Skip to content

Errata and changes

This edition is built from the ncurses documentation by Thomas E. Dickey, Zeyd M. Ben-Halim, Eric S. Raymond, Juergen Pfeifer, Pradeep Padala and the other contributors. This appendix lists every change it makes to that text: errors corrected in the original, errors this edition introduced and then fixed, formatting, structure and new material. It also lists errors found but not yet fixed, so they can be reported upstream. Errata meta-verification is the independent best-of-three audit of these error claims (with the current pass’s handoff report), Lessons learned records what went wrong in the migration itself and what now catches it, and References lists where to read more.

The list lives in errata.csv, one row per change, and the site shows it as the tables below. The list grows with the migration: every pull request that changes a guide or a manual page adds its rows, and the site’s workflow fails a pull request that changes the docs without touching the file.

Column Holds
category original-error, edition-error, formatting, structure, addition, open (found, not yet fixed) or rejected (checked, not an error)
where The file, as a path from the repository root
anchor The section in that file, when the change is in one section
was The text or behavior before the change; for an error, the claim as printed
now The text or behavior after it, or the fix to propose for an open row
source Where the error came from: the original guide, this edition, or the check that reported it
pr The pull request that made the change
commit The full SHA of the commit that made the change
checked What was actually opened to check the claim: files and lines, or URLs, not a citation
quote What those say, quoted
verdict matches, disagrees, missing, impossible, superseded or can't tell: does the evidence bear out the claim?
confidence high, medium or low
earlier What would have been concluded one step earlier, before the last thing checked

Every row that claims an error (original-error, edition-error, open, rejected) carries the five-step check in the last five columns, and the site shows it under each table. An error row’s verdict is matches; a claim the evidence doesn’t bear out moves to rejected with its verdict, so a finding is never dropped silently.

This edition’s own fixes can be wrong. When a later check shows that the original was right after all, a type I error (an error reported that wasn’t there), the row moves to rejected with the new evidence, so the record of the mistake stays. If the wrong fix also changed the text, the original wording is restored in its own commit, and a new edition-error row names that commit. The adjudication of the first outside pass lists the ten rows this happened to, and Lessons learned says why the first check got them wrong.

A row names the commit that made its change, so add the rows in a later commit than the change itself, in the same pull request. Pull requests here are merged with merge commits, which keep those SHAs.

Where Was Now Source Verdict Change
docs/client-library/forms/overview.md Post the form with form_post() Post the form using post_form() NCURSES Programming HOWTO, Forms chapter matches, high #1, f4e21cd
docs/client-library/forms/overview.md Unpost the menu with form_unpost() Unpost the form using unpost_form() NCURSES Programming HOWTO, Forms chapter matches, high #1, f4e21cd
docs/client-library/forms/overview.md Free the memory allocated to menu by free_form() Free the form, using free_form() NCURSES Programming HOWTO, Forms chapter matches, high #1, f4e21cd
docs/client-library/forms/custom-field-types.md Successor and predecessor hooks set with set_fieldtype_arg() set_fieldtype_choice(), as declared in form.h ncurses-intro.html, The Forms Library matches, high #6, 3a0e739
docs/client-library/forms/custom-field-types.md new_fieldtype() prototype without its closing semicolon Semicolon added ncurses-intro.html, The Forms Library matches, high #6, 3a0e739
docs/client-library/forms/custom-field-types.md The library routines will that call them will return error indication The library routines that call them will return an error indication ncurses-intro.html, The Forms Library matches, high #6, 2b1d4b7
docs/client-library/forms/custom-field-types.md should never see a NULL file pointer should never see a NULL pile pointer (the argument is the pile built by make_str) ncurses-intro.html, The Forms Library matches, medium #6, 2b1d4b7
docs/client-library/forms/field-options.md Example comment: This filed is like a password field This field is like a password field NCURSES Programming HOWTO, Field Options example matches, high #6, 2b1d4b7
man/curs_getch.3x see infocmp(1) (line 579) infocmp(1M); the pages declare section 1M, and the other 21 references to these programs use (1M) man/curs_getch.3x matches, high #9, 4df7807
man/curs_trace.3x tic(1) (line 207) tic(1M); the pages declare section 1M, and the other 21 references to these programs use (1M) man/curs_trace.3x matches, high #9, 4df7807
man/curs_extend.3x @TIC@(1) (line 77) @TIC@(1M); the pages declare section 1M, and the other 21 references to these programs use (1M) man/curs_extend.3x matches, high #9, 4df7807
man/curs_inopts.3x @TIC@(1) (line 669) @TIC@(1M); the pages declare section 1M, and the other 21 references to these programs use (1M) man/curs_inopts.3x matches, high #9, 4df7807
docs/man/term/02-description.5 @TIC@(1) (line 63) @TIC@(1M); the pages declare section 1M, and the other 21 references to these programs use (1M) man/term.5 matches, high #9, 4df7807
man/term_variables.3x @TIC@(1) and @INFOCMP@(1) (line 142) @TIC@(1M) and @INFOCMP@(1M); the pages declare section 1M, and the other 21 references to these programs use (1M) man/term_variables.3x matches, high #9, 4df7807
man/curs_bkgd.3x The background character’s set of attributes becomes a property of the character cell and move with it (line 101) … and moves with it ncurses manual page curs_bkgd(3X) matches, high #6, 63838d4
man/curs_bkgrnd.3x The background character’s set of attributes becomes a property of the character cell and move with it (line 92) … and moves with it ncurses manual page curs_bkgrnd(3X) matches, high #6, 63838d4
docs/client-library/menus/overview.md Post the menu with menu_post(); Unpost the menu with menu_unpost() Post the menu using post_menu(); Unpost the menu using unpost_menu() NCURSES Programming HOWTO, Menus chapter matches, high #9, d03f995
docs/client-library/menus/overview.md The menu can then by posted The menu can then be posted Writing Programs with NCURSES, The Menu Library matches, high #9, d03f995
docs/client-library/menus/menu-windows.md When you call unpost_menu(), you erase the subwindow, However, neither of these… …you erase the subwindow. However, neither of these… Writing Programs with NCURSES, The Menu Library matches, high #9, d03f995
docs/client-library/menus/menu-display.md see the menu_attribs(3x) manual page. (no closing parenthesis) see the menu_attributes(3X) manual page) Writing Programs with NCURSES, The Menu Library matches, high #9, d03f995
docs/client-library/menus/menu-driver.md menu_driver() accepts an ascii character (HOWTO); it accumulates printable ASCII characters (Writing Programs) A printable character NCURSES Programming HOWTO, Menus chapter; Writing Programs with NCURSES, The Menu Library matches, medium #9, d03f995
docs/client-library/menus/menu-display.md Example 21. Milt Columnar Menus Example Example: Multi Columnar Menus Example NCURSES Programming HOWTO, Menus chapter matches, high #9, d03f995
docs/client-library/menus/compiling.md The Menu Basics listing frees only items 0 and 1 of five, and calls free_menu() on a menu that is still posted The listing unposts the menu, frees it, then frees every item NCURSES Programming HOWTO, Menus chapter matches, high #9, d03f995
docs/client-library/menus/selecting-items.md The Multi Valued Menus listing frees only items 0 and 1 of eight, and calls free_menu() on a menu that is still posted The listing unposts the menu, frees it, then frees every item NCURSES Programming HOWTO, Menus chapter matches, high #9, d03f995
docs/client-library/menus/attributes.md The Menu Options listing frees the items before the menu The listing frees the menu, then the items NCURSES Programming HOWTO, Menus chapter matches, high #9, d03f995
docs/client-library/menus/user-pointer-and-hooks.md The Menu User Pointer listing frees the items before the menu The listing frees the menu, then the items NCURSES Programming HOWTO, Menus chapter matches, high #9, d03f995
docs/client-library/menus/user-pointer-and-hooks.md The Menu User Pointer listing stores the address of MY_DATA data, a variable local to the loop body, as each item’s user pointer and calls through it after the loop data is static, so the pointer stays valid NCURSES Programming HOWTO, Menus chapter matches, high #9, d03f995
docs/client-library/menus/menu-display.md The list of menu options ends at O_NONCYCLIC The list includes O_MOUSE_MENU NCURSES Programming HOWTO, Menus chapter matches, high #9, d03f995
docs/client-library/menus/menu-driver.md The mouse requests are quoted from an older menu_driver(3X), with its line breaks and soft hyphens (decora­tion), and without the double-click on an item The current behavior, from menu_driver(3X), as a list, including the double-click and E_REQUEST_DENIED NCURSES Programming HOWTO, Menus chapter matches, high #9, d03f995
docs/client-library/menus/attributes.md A link to text mode dos games at http://www.jersey.net/~debinjoe/games/ Link removed NCURSES Programming HOWTO, Menus chapter matches, high #9, d03f995
[man/menu_driver.3x](/ncurses/man/3x/menu_driver/#APPLICATION-DEFINED COMMANDS) the drive assumes it is an application-specific command the driver assumes it is an application-specific command ncurses manual pages matches, high #9, 3b3c5fd
[man/menu_driver.3x](/ncurses/man/3x/menu_driver/#RETURN VALUE) menu_driver return one of the following error codes menu_driver returns one of the following error codes ncurses manual pages matches, high #9, 3b3c5fd
docs/client-library/panels/overview.md replace a panel’s window with a different window by calling replace_window calling replace_panel() ncurses-intro.html, The Panels Library matches, high #10, e57bdd2
docs/client-library/panels/input-and-stdscr.md The stsdcr window is a special case The stdscr window ncurses-intro.html, The Panels Library matches, high #10, e57bdd2
docs/client-library/panels/hiding.md The panel_update code ignores hidden panels The update_panels() function ignores hidden panels ncurses-intro.html, The Panels Library matches, high #10, e57bdd2
docs/client-library/panels/hiding.md You cannot do top_panel() or bottom_panel on a hidden panel() In ncurses both work on a hidden panel and make it visible; use show_panel() for portability to System V ncurses-intro.html, The Panels Library matches, high #10, e57bdd2
docs/client-library/panels/input-and-stdscr.md Note that wgetch automatically calls wrefresh wgetch() calls wrefresh() on its window when the window has changed since it was last refreshed ncurses-intro.html, The Panels Library matches, medium #10, e57bdd2
docs/client-library/panels/hiding.md For some reason the function panel_hidden() which tells whether a panel is hidden or not is not working, with a link to a 1999 bug report panel_hidden() returns TRUE for a hidden panel, and the example uses it NCURSES Programming HOWTO, Panel Library chapter matches, high #10, e57bdd2
docs/client-library/panels/browsing.md /* Update the stacking order. 2nd panel will be on top */ (also in the moving and hiding listings) The last panel created is on top NCURSES Programming HOWTO, Panel Library chapter matches, high #10, e57bdd2
docs/client-library/panels/moving-and-resizing.md The program reads a panel’s position and size from its PANEL_DATA when it comes to the top, but never writes them back, so a moved panel jumps back to where it started the next time it is moved The loop writes the position and size back after each key NCURSES Programming HOWTO, Panel Library chapter matches, high #10, e57bdd2
docs/client-library/panels/hiding.md “Show or Hide a window with ‘a’(first window) ‘b’(Second Window) ‘c’(Third Window)”, 83 characters on an 80-column screen A 72-character line NCURSES Programming HOWTO, Panel Library chapter matches, high #10, e57bdd2
docs/client-library/panels/overview.md Mainpulate; is used to created panels; removes it form the stack; Entered Resizing :Use Manipulate; create; from; Entered Resizing: Use NCURSES Programming HOWTO, Panel Library chapter matches, high #10, e57bdd2
docs/client-library/panels/hiding.md bottom_panel on a hidden panel() bottom_panel() on a hidden panel ncurses-intro.html, The Panels Library matches, high #10, e57bdd2
docs/client-library/getting-started/initialization.md In the cbreak() mode these control characters are interpreted as any other character by the terminal driver In cbreak() mode the terminal driver still interprets them, and generates the signal as usual NCURSES Programming HOWTO, chapters 1–4 matches, high #11, eb995ee
docs/client-library/getting-started/hello-world.md endwin() frees the memory taken by curses sub-system and its data structures and puts the terminal in normal mode endwin() restores the terminal modes and cursor; it does not free memory, and a later refresh() resumes curses mode NCURSES Programming HOWTO, chapters 1–4 matches, high #11, eb995ee
docs/client-library/getting-started/hello-world.md <curses.h> also includes <termios.h>, <termio.h>, or <sgtty.h> depending on your system Removed; curses.h includes stdio.h, and no terminal-driver header ncurses-intro.html, Introduction and An Overview of Curses matches, high #11, eb995ee
docs/client-library/getting-started/hello-world.md In linking with curses you need to have -lncurses … There is no need for any other libraries A build configured –with-termlib also needs tinfo when linking statically; pkg-config –libs ncurses prints what an installation needs ncurses-intro.html, Introduction and An Overview of Curses matches, medium #11, eb995ee
docs/client-library/getting-started/hello-world.md To use ncurses library functions, you have to include ncurses.h in your programs #include <curses.h>; ncurses also installs it as ncurses.h NCURSES Programming HOWTO, chapters 1–4 matches, high #11, eb995ee
docs/client-library/getting-started/naming-and-variables.md bool: boolean type, actually a “char” C’s bool from stdbool.h where available, otherwise a type configure chooses ncurses-intro.html, Introduction and An Overview of Curses matches, high #11, eb995ee
docs/client-library/getting-started/naming-and-variables.md OK: error flag returned by routines when things go right flag returned by routines when things go right (0) ncurses-intro.html, Introduction and An Overview of Curses matches, high #11, eb995ee
docs/client-library/getting-started/naming-and-variables.md Changes should be made to through the API; To write to a different window. use the routine waddch() … is provided Changes should be made through the API; To write to a different window, use the routine waddch() ncurses-intro.html, Introduction and An Overview of Curses matches, high #11, eb995ee
docs/client-library/getting-started/initialization.md rearing to get started; halfdelay()is called; if he doesn’t respond with in certain time halfdelay() is called; if they don’t respond within a certain time NCURSES Programming HOWTO, chapters 1–4 matches, high #11, eb995ee
docs/client-library/getting-started/history.md Mail links to zmbenhal@netcom.com and esr@snark.thyrsus.com, and a link to adahome.com for the Ada95 binding Names without the old addresses and link ncurses-intro.html, Introduction and An Overview of Curses matches, medium #11, eb995ee
docs/client-library/windows/border-functions.md create_box() puts the corners at x + w and y + h, so a box of width 10 and height 3 is drawn 11 columns wide and 4 rows tall, and the erase loop clears one row and column more Corners at x + w - 1 and y + h - 1, lines of w - 2 and h - 2, and the erase loop stops before x + w and y + h NCURSES Programming HOWTO, chapters 5 and 9 matches, high #12, 2b29d57
docs/client-library/windows/windows-and-refresh.md printw(string); mvprintw(y, x, string); … printw(fmt, …); and so on NCURSES Programming HOWTO, chapters 5 and 9 matches, medium #12, 2b29d57
docs/client-library/windows/creating-windows.md let me clear few things; It doesn’t create any thing; as user press a key; Let’s dissect it by line by line a few things; anything; as the user presses a key; line by line NCURSES Programming HOWTO, chapters 5 and 9 matches, high #12, 2b29d57
docs/client-library/output/forms-characters.md If the terminal does not have such characters, curses.h will map them to a recognizable (though ugly) set of ASCII defaults curses maps them to ASCII defaults when it starts; the header only defines the ACS_ names Writing Programs with NCURSES, Output and Using Forms Characters matches, medium #13, eea2d03
docs/client-library/output/ let’s interact with world; message to be appeared on the screen; Try resizing your window(if possible); ch(of type char); the character printed … until it is turned off let’s interact with the world; message to appear on the screen; window (if possible); characters printed … until they are turned off NCURSES Programming HOWTO, chapter 6 matches, high #13, eea2d03
docs/client-library/windows/border-functions.md With _DEBUG defined, print_win_params() prints at row 25, below the bottom of a 24-row screen, so nothing is shown Prints on the bottom line (LINES - 1), then moves the cursor back to the top left so “Press F1 to exit” stays at the top NCURSES Programming HOWTO, chapter 9 matches, high #14, 16fc371
docs/client-library/input/characters-and-strings.md char str[80]; … getstr(str); getnstr(str, (int) sizeof(str) - 1); NCURSES Programming HOWTO, chapter 7 matches, high #15, d341cd3
docs/client-library/input/characters-and-strings.md Example 4. A Simple scanw example A Simple getstr example NCURSES Programming HOWTO, chapter 7 matches, high #15, d341cd3
docs/client-library/input/characters-and-strings.md These pseudo-line-oriented functions turn on echoing while they execute If echo is set, the characters are echoed as they are typed Writing Programs with NCURSES, Input matches, high #15, d341cd3
docs/client-library/input/ begin read them; must used; one or the other has to changed; The resulting string of characters are pointed to begin reading them; must be used; has to be changed; is pointed to NCURSES Programming HOWTO, chapter 7; Writing Programs with NCURSES, Input matches, high #15, d341cd3
docs/client-library/attributes/colors.md Initially RED color is defined with content 1000(r), 0(g), 0(b); init_color(COLOR_RED, 700, 0, 0) is described as lightening red by a minuscule Initially ncurses defines red as 680, 0, 0, so the call makes it slightly brighter NCURSES Programming HOWTO, chapters 8 and 10 matches, high #16, 282ca13
docs/client-library/attributes/attribute-functions.md attr_get() is useful in scanning areas of screen when we are not sure what attribute each character is associated with attr_get() reports the window’s current attributes and color pair, for the next characters written; inch() reads characters already on the screen NCURSES Programming HOWTO, chapters 8 and 10 matches, medium #16, 282ca13
docs/client-library/attributes/chgat.md Other functions wchgat(), mvchgat(), wchgat() behave similarly wchgat(), mvchgat() and mvwchgat() NCURSES Programming HOWTO, chapters 8 and 10 matches, high #16, 282ca13
docs/client-library/attributes/attributes.md /* pager functionality by Joseph Spainhour” spainhou@bellsouth.net */ The stray double quote after the name is removed NCURSES Programming HOWTO, chapters 8 and 10 matches, high #16, 282ca13
docs/client-library/attributes/ screeen; are we are at; cut bold on; Forth parameter; Viola !!!; loose track; the the source; Don’t worry about all those initialization and other crap screen; are we at; turn bold on; Fourth parameter; Voila !!!; lose track; the source; Don’t worry about all that initialization NCURSES Programming HOWTO, chapters 8 and 10 matches, high #16, 282ca13
docs/client-library/keyboard/key-usage-example.md mvprintw(24, 0, “Character pressed is …”); mvprintw(23, 0, “You chose choice …”) mvprintw(LINES - 1, 0, …); mvprintw(LINES - 2, 0, …) NCURSES Programming HOWTO, chapter 11 matches, high #17, 8aff0ee
docs/client-library/keyboard/key-usage-example.md startx = (80 - WIDTH) / 2; starty = (24 - HEIGHT) / 2; startx = (COLS - WIDTH) / 2; starty = (LINES - HEIGHT) / 2; NCURSES Programming HOWTO, chapter 11 matches, high #17, 8aff0ee
docs/client-library/keyboard/ the Functions keys; See the initialization section; if(ch == KEY_LEFT); User did a choice come out of the infinite loop the function keys; a link to Initialization; if (ch == KEY_LEFT); User made a choice; leave the loop NCURSES Programming HOWTO, chapter 11 matches, high #17, 8aff0ee
docs/client-library/mouse/mouse-events.md The default is one fifth of a second The default is one sixth of a second NCURSES Programming HOWTO, chapter 12 matches, high #19, 02ebd3a
docs/client-library/mouse/mouse-menu-example.md if (event.bstate & BUTTON1_PRESSED) with mousemask(ALL_MOUSE_EVENTS, NULL) if (event.bstate & (BUTTON1_PRESSED | BUTTON1_CLICKED)) NCURSES Programming HOWTO, chapter 12 matches, high #19, 02ebd3a
docs/client-library/mouse/mouse-menu-example.md report_choice(event.x + 1, event.y + 1, …); j = starty + 3; mouse_x <= i + strlen(…) report_choice(event.x, event.y, …); j = starty + 2; mouse_x < i + len NCURSES Programming HOWTO, chapter 12 matches, high #19, 02ebd3a
docs/client-library/mouse/mouse-menu-example.md startx = (80 - WIDTH) / 2; starty = (24 - HEIGHT) / 2; mvprintw(23, 1, …); mvprintw(22, 1, …) COLS and LINES; LINES - 1 and LINES - 2 NCURSES Programming HOWTO, chapter 12 matches, high #19, 02ebd3a
docs/client-library/mouse/mouse-events.md The event mask table stops at BUTTON4_TRIPLE_CLICKED The table includes buttons 5 and notes NCURSES_MOUSE_VERSION NCURSES Programming HOWTO, chapter 12 matches, high #19, 02ebd3a
docs/client-library/mouse/ lets do the same thing from mouse; any thing; a class of mouse events have been enabled; Do some thing; thousands of a second; convert to mouse co-ordinates to screen relative let’s do the same thing with the mouse; anything; has been enabled; Do something; thousandths of a second; convert between window-relative and screen-relative coordinates NCURSES Programming HOWTO, chapter 12 matches, high #19, 02ebd3a
docs/client-library/screen/coordinates.md getbegyx() and getmaxyx() store the window’s beginning and maximum co-ordinates getmaxyx() stores the size: the number of lines and columns, one more than the largest coordinate NCURSES Programming HOWTO, chapter 13 matches, high #24, dbe947c
docs/client-library/init-and-debugging/terminal-descriptions.md The values returned can be 1 (all is well), 0 (no such terminal), or -1 (some problem locating the terminfo database) With OK the status is 1; with ERR, 1 means a hard-copy terminal, 0 a type not found or generic, and -1 no terminfo database Writing Programs with NCURSES, Function Descriptions matches, high #26, 808b628
docs/client-library/init-and-debugging/debugging.md trace() sets the trace level; _tracef() is available only if you link with -lncurses_g curses_trace() sets the level and replaces the deprecated trace(); the trace functions are in the debugging library or any build with TRACE defined; tracing is skipped if the trace file already exists Writing Programs with NCURSES, Function Descriptions matches, high #26, 808b628
docs/client-library/init-and-debugging/initialization-and-wrapup.md endwin() should called once Call endwin() once Writing Programs with NCURSES, Function Descriptions matches, high #26, 808b628
docs/client-library/init-and-debugging/debugging.md It is only available only if (sentence rewritten) Writing Programs with NCURSES, Function Descriptions matches, high #26, 808b628
docs/client-library/init-and-debugging/terminal-descriptions.md Setupterm() also stores setupterm() also copies Writing Programs with NCURSES, Function Descriptions matches, high #26, 808b628
docs/client-library/hints/resizing.md When ncurses returns that code, it calls resizeterm ncurses calls resizeterm() with the new size at the next wgetch() or refresh, and wgetch() then returns KEY_RESIZE Writing Programs with NCURSES, Hints, Tips, and Tricks matches, high #32, c812489
docs/client-library/hints/resizing.md The refresh will pick up the new screen size from the xterm’s environment The refresh() picks up the new screen size from the terminal Writing Programs with NCURSES, Hints, Tips, and Tricks matches, medium #32, c812489
docs/client-library/hints/terminals.md Note that you will also have to call def_shell_mode and def_prog_mode on each tty yourself newterm() saves each terminal’s shell and program modes, as initscr() does, and each terminal has its own saved modes Writing Programs with NCURSES, Hints, Tips, and Tricks matches, high #32, c812489
docs/client-library/hints/extensions.md Ncurses supports up 16 colors, unlike SVr4 curses which defines only 8. While most terminals which provide color allow only 8 colors, about a quarter (including XFree86 xterm) support 16 colors. ncurses supports as many colors as the terminal description’s colors capability gives, such as 256 for xterm-256color, and init_extended_color() and init_extended_pair() take int arguments past the limit of short Writing Programs with NCURSES, Special Features of NCURSES matches, high #32, c812489
docs/client-library/hints/compatibility.md the attribute of erased blanks is normal unless and until it is modified by the functions bkgdset() or wbkgdset() Erased blanks have the window’s background rendition, which is normal until you change it with bkgd(), bkgdset(), wbkgd() or wbkgdset() Writing Programs with NCURSES, Compatibility with Older Versions matches, high #32, c812489
docs/client-library/hints/compatibility.md every macro entry point have a corresponding function which may be linked A function the standard defines, and ncurses also implements as a macro, is available as a function too; getyx(), getmaxyx() and the others the standard defines only as macros have no function Writing Programs with NCURSES, XSI Curses Conformance matches, high #32, c812489
docs/client-library/more/hanoi.md mvprintw(24, 0, “Press F1 to Exit”) mvprintw(LINES - 1, 0, “Press F1 to Exit”) NCURSES Programming HOWTO, Just For Fun (example archive) matches, high #33, bec4c6d
docs/client-library/more/shuffle.md mvprintw(24, 0, …) for “Press F1 to Exit” and “You Win !!!” mvprintw(LINES - 1, 0, …) NCURSES Programming HOWTO, Just For Fun (example archive) matches, high #33, bec4c6d
docs/client-library/more/typing-tutor.md #define STATUSY 25 #define STATUSY (LINES - 1) NCURSES Programming HOWTO, Just For Fun (example archive) matches, high #33, bec4c6d
docs/client-library/more/widget-libraries.md http://www2.linuxjournal.com/lj-issues/issue5/2807.html https://www.linuxjournal.com/article/2807 NCURSES Programming HOWTO, Tools and Widget Libraries matches, high #33, bec4c6d
docs/contributor/hackguide/objective.md The third objective (standards conformance) The second objective A Hacker’s Guide to NCURSES matches, high #45, 6e167a6
docs/contributor/hackguide/reporting-bugs.md This is a majordomo list This is a GNU Mailman list, with a public archive A Hacker’s Guide to NCURSES matches, high #45, 6e167a6
docs/contributor/hackguide/reporting-bugs.md trace() curses_trace() A Hacker’s Guide to NCURSES matches, high #45, 6e167a6
docs/contributor/hackguide/reporting-bugs.md Test frames for the library algorithms are in the progs directory They are in ncurses/, built with make test_progs; the mvcur tester is lib_mvcur A Hacker’s Guide to NCURSES matches, high #45, 6e167a6
docs/contributor/hackguide/engine-room.md If the entire sequence takes less than 1 second Each character of the sequence must arrive within ESCDELAY, 1000 ms by default A Hacker’s Guide to NCURSES matches, medium #45, 6e167a6
docs/contributor/hackguide/engine-room.md timed_wait() _nc_timed_wait() A Hacker’s Guide to NCURSES matches, high #45, 6e167a6
docs/contributor/hackguide/engine-room.md lib_doupdate.c ncurses/tty/tty_update.c A Hacker’s Guide to NCURSES matches, high #45, 6e167a6
docs/contributor/hackguide/engine-room.md _nc_mvcur_scrolln() in lib_mvcur.c _nc_scrolln() in tty_update.c A Hacker’s Guide to NCURSES matches, high #45, 6e167a6
docs/contributor/hackguide/engine-room.md See include/curses.h.in for mask values See include/curses.tail A Hacker’s Guide to NCURSES matches, high #45, 6e167a6
docs/contributor/hackguide/engine-room.md The lib_mouse() code The code in lib_mouse.c A Hacker’s Guide to NCURSES matches, medium #45, 6e167a6
docs/contributor/hackguide/forms-and-menus.md will settle for BSD regexps will settle for the older compile() and step() from regexp.h or regexpr.h A Hacker’s Guide to NCURSES matches, medium #45, 6e167a6
docs/contributor/hackguide/terminfo-compiler.md ncurses/dump_entry.c progs/dump_entry.c A Hacker’s Guide to NCURSES matches, high #45, 6e167a6
docs/contributor/hackguide/style-tips.md ncurses/MKlib_gen.sh ncurses/base/MKlib_gen.sh A Hacker’s Guide to NCURSES matches, high #45, 6e167a6
docs/contributor/hackguide/style-tips.md The fifth field (flags) of include/Caps The seventh column A Hacker’s Guide to NCURSES matches, high #45, 6e167a6
docs/contributor/hackguide/porting-hints.md lib_getstr.c is pure curses, but calls outstr() Pure curses; it echoes through waddch() and wechochar() A Hacker’s Guide to NCURSES matches, high #45, 6e167a6
docs/contributor/hackguide/porting-hints.md lib_insstr.c lib_insnstr.c, here and in the library tour A Hacker’s Guide to NCURSES matches, high #45, 6e167a6
docs/contributor/hackguide/porting-hints.md lib_unctrl.c unctrl.c, generated at build, here and in the library tour A Hacker’s Guide to NCURSES matches, high #45, 6e167a6
docs/contributor/hackguide/porting-hints.md comp_main.c progs/tic.c A Hacker’s Guide to NCURSES matches, high #45, 6e167a6
docs/contributor/install/installation-procedure.md The configure script attempts to locate previous installation of ncurses, and will set the default prefix according to where it finds the ncurses headers The configure script chooses the default prefix from the system type alone; it does not look for an earlier installation INSTALL matches, high #49, 2ff9553
docs/contributor/install/installation-procedure.md Typing ‘configure’ with no arguments is equivalent to: ./configure –with-normal –with-debug –enable-overwrite ./configure –with-normal –with-debug, with overwrite on by default only when the prefix is /usr INSTALL matches, high #49, 2ff9553
docs/contributor/install/installation-procedure.md read the PORTABILITY section in the file ncurses/README read the porting hints in the hacker’s guide INSTALL matches, high #49, 2ff9553
docs/contributor/install/installation-procedure.md tic(1), captoinfo(1), infocmp(1), toe(1), clear(1) tset(1), reset(1), and tput(1) tic(1), captoinfo(1), infotocap(1), infocmp(1), toe(1), clear(1), tabs(1), tset(1), reset(1) and tput(1); step 1’s list also gains infotocap INSTALL matches, high #49, 2ff9553
docs/contributor/install/configure-options/with-m-z.md If the option is not given, this defaults to “long”. intptr_t for ABI 6, the default ABI, and long for ABI 5 INSTALL matches, high #49, 2ff9553
docs/contributor/install/configure-options/with-m-z.md –with-weak-symbols –enable-weak-symbols INSTALL matches, high #49, 2ff9553
docs/contributor/install/system-integrators.md FreeBSD, NetBSD, OpenBSD, Cygwin, MinGW Linux, GNU/kFreeBSD and other kbsd-gnu systems, FreeBSD, OpenBSD, MirBSD, Cygwin, MSYS and MinGW; NetBSD gets /usr/local INSTALL matches, high #49, 2ff9553
docs/contributor/install/system-integrators.md listed in the misc/terminfo file listed in the misc/terminfo.src file INSTALL matches, high #49, 2ff9553
docs/contributor/install/bsd-conversion.md $HOME/terminfo (twice) $HOME/.terminfo INSTALL matches, high #49, 2ff9553
docs/contributor/install/bug-reports.md To subscribe send mail to bug-ncurses-request@gnu.org with body that reads: subscribe ncurses It is a GNU Mailman list; join it from its subscription page INSTALL matches, medium #49, 2ff9553
docs/contributor/install/mingw.md assuming environment variable MINGW_ROOT holds the root directory name of your MinGW build with the MinGW root installed at /mingw64 README.MinGW matches, medium #49, 2ff9553
docs/contributor/ada/limitations.md I provide no functions to switch on/off curses tracing options. Tracing is supported through Terminal_Interface.Curses.Trace (Trace_On and Trace_Put) Ada95 Binding for ncurses matches, high #50, e3fa730
docs/contributor/ada/index.md Just make the boolean member bold of the record true. Set the Bold_Character member of a Character_Attribute_Set record to true Ada95 Binding for ncurses matches, high #50, e3fa730
docs/contributor/ada/index.md Support for wide characters is currently not in the binding, as it is not really in ncurses at this point in time. The binding does not support wide characters Ada95 Binding for ncurses matches, high #50, e3fa730
docs/contributor/ada/implementation-details.md The (n)curses documentation says, that the String arrays to be passed to an TYPE_ENUM fieldtype must not be automatic variables. ncurses also copies the list for TYPE_ENUM, so a C list may live in automatic variables too Ada95 Binding for ncurses matches, high #50, e3fa730
docs/contributor/announce/release-notes.md The most important bug-fixes/improvements (a link to “Interesting bug-fixes”, a section with no content) The most important bug-fixes and improvements dealt with robustness issues (no link); the empty section is dropped announce.html.in matches, high #52, 54088dc
docs/contributor/announce/program-improvements.md https://invisible-island.net/ncurses/man/tput.1m.html https://invisible-island.net/ncurses/man/tput.1.html announce.html.in matches, high #52, 54088dc
docs/contributor/announce/development.md Patches to the current release are made available at https://invisible-island.net/archives/ncurses/6.5/ …/archives/ncurses/6.6/ announce.html.in matches, high #52, 54088dc
docs/contributor/announce/development.md To join the ncurses mailing list, please write email to bug-ncurses-request@gnu.org containing the line: subscribe @<host.domain> Join it from its Mailman subscription page announce.html.in matches, medium #52, 54088dc
docs/contributor/announce/portability.md modify MKlib_gen.c to allow for Solaris’s definition of NULL as 0L modify MKlib_gen.sh … announce.html.in matches, high #52, 54088dc
docs/client-library/getting-started/introduction.md to revert to the original mode, type this: printf ‘\033[0;37;40m\n’ printf ‘\033[0m\n’ NCURSES Programming HOWTO, Introduction matches, high #54, 74a7f5a
docs/client-library/getting-started/initialization.md control characters like suspend (CTRL-Z), interrupt and quit (CTRL-C) suspend (CTRL-Z), interrupt (CTRL-C) and quit (CTRL-) NCURSES Programming HOWTO, Initialization matches, high #54, 74a7f5a
docs/client-library/attributes/chgat.md Fifth is the color index. It is the index given during init_pair() Fifth is the color pair number, the one given to init_pair() NCURSES Programming HOWTO, chgat() matches, high #55, 2b69b21
docs/client-library/mouse/mouse-events.md Do this before the next wgetch(); otherwise another mouse event might come in and make the first one inaccessible ncurses keeps mouse events in a separate queue of eight; read each one with getmouse() when KEY_MOUSE arrives Writing Programs with NCURSES, Mouse Interfacing matches, high #55, 2b69b21
docs/client-library/output/line-drawing-characters.md An example showing all the characters, with 31 SHOW() lines and no ACS_URCORNER 32 SHOW() lines, adding the upper right corner, and a note on the eleven ACS_ aliases NCURSES Programming HOWTO, Line Graphics matches, high #55, 2b69b21
docs/client-library/forms/fields.md Note that new_field() does not copy the pointer array into private storage Note that new_form() does not copy the pointer array Writing Programs with NCURSES, The Forms Library matches, high #56, e117d45
docs/client-library/forms/field-options.md By default, all options are on. The ten standard options are on by default; the four ncurses extension options are off Writing Programs with NCURSES, The Forms Library matches, high #56, e117d45
docs/client-library/forms/field-options.md Each field structure contains one character pointer slot … The (char *) type is retained for System V compatibility One pointer slot; the char * remark is removed Writing Programs with NCURSES, The Forms Library matches, high #56, e117d45
docs/client-library/forms/field-options.md The growth limit can be changed whether or not the O_STATIC bit is on, but has no effect until it is. … but has no effect until that bit is turned off. Writing Programs with NCURSES, The Forms Library matches, high #56, e117d45
docs/client-library/forms/form-driver.md A REQ_MOVE_RIGHT from A will go to B A REQ_RIGHT_FIELD from A will go to B Writing Programs with NCURSES, The Forms Library matches, high #56, e117d45
docs/client-library/forms/form-driver.md the position of a form for purposes of these requests is its upper-left corner the position of a field for purposes of these requests Writing Programs with NCURSES, The Forms Library matches, high #56, e117d45
docs/client-library/forms/form-driver.md REQ_DEL_PREV: Delete previous word at cursor. REQ_DEL_PREV: Delete character before the cursor. Writing Programs with NCURSES, The Forms Library matches, high #56, e117d45
docs/client-library/forms/hooks-and-form-options.md When the current page is changed by set_current_field() call; When the current field is changed by a set_form_page() call set_current_field() changes the current field; set_form_page() changes the current page Writing Programs with NCURSES, The Forms Library matches, high #56, e117d45
docs/client-library/hints/compatibility.md ncurses is intended to be base-level conformant with the X/Open Curses standard … almost all the features not directly concerned with wide characters and internationalization ncurses implements X/Open Curses Issue 7, including the wide-character functions in its wide-character library Writing Programs with NCURSES, X/Open Curses Conformance matches, high #56, e117d45
docs/client-library/more/shuffle.md Slide the tiles with the arrow keys until they’re in order A note that the program deals a random arrangement without checking it can be solved, and half of all arrangements can’t be NCURSES Programming HOWTO, Just For Fun (JustForFun/shuffle.c) matches, high #56, e117d45
docs/contributor/hackguide/terminfo-compiler.md parse an entire terminfo file into a doubly-linked circular list of entry structures doubly-linked list A Hacker’s Guide to NCURSES matches, high #57, 769b14b
docs/contributor/hackguide/other-utilities.md The tput and clear utilities just do an entry load followed by a tputs() of a selected capability. clear also sends E3 when present, and tput also handles init and reset A Hacker’s Guide to NCURSES matches, high #57, 769b14b
docs/contributor/install/installation-procedure.md and also interprets the contents of the $TERM environment variable the $TERMCAP environment variable INSTALL matches, high #57, 769b14b
docs/contributor/install/installation-procedure.md $(prefix)/man: the manual pages $(prefix)/share/man INSTALL matches, high #57, 769b14b
docs/contributor/install/installation-procedure.md the library searches /etc/termcap before the terminfo database after the terminfo database, when no terminfo entry is found INSTALL matches, high #59, 8e5de3e
docs/client-library/getting-started/get-ncurses.md su root (become root), then make install sudo make install (install it as root) NCURSES Programming HOWTO, Where to get it matches, high #65, 90f5129
docs/man/mitem_visible/03-description.3x A menu item is visible when it is in the portion of a posted menu that is mapped onto the screen (if the menu is scrollable, in particular, this portion will be smaller than the whole menu). A menu item is visible when it is in the part of a posted menu that the menu’s subwindow shows. In a scrollable menu, that part is smaller than the whole menu; top_row (see mitem_current(3X)) gives the menu row it starts at. man/mitem_visible.3x matches, high #78, b912aee
docs/man/mitem_visible/05-portability.3x These routines emulate the System V menu library. They were not supported on Version 7 or BSD versions. This routine emulates the System V menu library. It was not supported on Version 7 or BSD versions. man/mitem_visible.3x matches, high #78, b912aee
docs/man/captoinfo/03-description.1m If no files are specified, captoinfo interprets the content of the environment variable TERMCAP as a file name, and extracts only the entry for the terminal type named in the environment variable TERM from it. If no files are specified, captoinfo translates only the entry for the terminal type named in TERM, reading it from TERMCAP. If the value of TERMCAP names an existing file, captoinfo reads that file; otherwise it takes the value as the termcap entry itself. If either variable is not set, captoinfo reads /etc/termcap. man/captoinfo.1m matches, high #81, 50b61fe
docs/man/menu_requestname/04-return-value.3x menu_request_by_name returns E_NO_MATCH on error. It does not set errno. menu_request_by_name returns E_NO_MATCH when no request has the given name, and sets errno to E_NO_MATCH. man/menu_requestname.3x matches, high #81, 50b61fe
docs/man/form_requestname/04-return-value.3x form_request_by_name returns E_NO_MATCH on error. It does not set errno. form_request_by_name returns E_NO_MATCH when no request has the given name, and sets errno to E_NO_MATCH. man/form_requestname.3x matches, high #81, 50b61fe
docs/man/term/04-portability.5 the name field cannot exceed 128 bytes. the name field cannot exceed 512 bytes, including its terminating null character. ncurses writes and reads back names of up to 511 characters; at 512 or more tic writes an entry ncurses cannot read, a bug tracked in this repository’s issue “Bugs in ncurses found while migrating the docs” man/term.5 matches, high #82, 2c2df01
docs/man/tput/05-exit-status.1 0: Boolean or string capability present; 1: Boolean or numeric capability absent 0: Boolean or string capability present, or any numeric capability; 1: Boolean or string capability absent; and a note that an absent numeric capability gives 0 and is written as -1 man/tput.1 matches, high #84, a422f86
docs/man/tput/05-exit-status.1 With -S: 4: some operands not interpreted With -S: 4: unrecognized capability code, later operands not interpreted; >4: 4 + the number of operands whose status was not 0, or system error man/tput.1 matches, high #84, a422f86
docs/man/tput/04-options.1 Only one cap-code is allowed per line. A line may hold more than one cap-code, each followed by its parameters. man/tput.1 matches, high #84, a422f86
docs/man/define_key/04-return-value.3x define_key returns ERR if definition is NULL and key-code is nonpositive, or insufficient memory is available to bind definition to key-code. Otherwise, define_key returns OK. define_key also returns ERR when no screen has been initialized, when definition is NULL and key-code has no binding, when key-code is nonpositive and definition has no binding, when definition is empty, and when definition and a bound string are prefixes of one another man/define_key.3x matches, high #85, 72b9115
How each was checked
  • Post the form with form_post() (docs/client-library/forms/overview.md)
    1. Claim, as printed: Post the form with form_post()
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:5528; form/form.h:429
    3. It says: “HOWTO: Post the form with form_post() and refresh the screen. form.h: extern FORM_EXPORT(int) post_form (FORM *);”
    4. Verdict: matches, high confidence
    5. One step earlier: Without form.h I would have had only the HOWTO’s word against the forms man pages; form.h shows no form_post exists
  • Unpost the menu with form_unpost() (docs/client-library/forms/overview.md)
    1. Claim, as printed: Unpost the menu with form_unpost()
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:5538; form/form.h:430
    3. It says: “HOWTO: Unpost the menu with form_unpost(). form.h: extern FORM_EXPORT(int) unpost_form (FORM *);”
    4. Verdict: matches, high confidence
    5. One step earlier: Reading only the HOWTO, the step looks like a slip of the pen about menus; form.h shows the name is wrong as well
  • Free the memory allocated to menu by free_form() (docs/client-library/forms/overview.md)
    1. Claim, as printed: Free the memory allocated to menu by free_form()
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:5542; doc/html/NCURSES-Programming-HOWTO.html:4325
    3. It says: “Line 5542: Free the memory allocated to menu by free_form(). Line 4325, in the Menus chapter: Free the memory allocated to menu by free_menu()”
    4. Verdict: matches, high confidence
    5. One step earlier: Without line 4325 I would not have seen that the forms steps were copied from the menus steps, which explains all three slips
  • Successor and predecessor hooks set with set_fieldtype_arg() (docs/client-library/forms/custom-field-types.md)
    1. Claim, as printed: Successor and predecessor hooks set with set_fieldtype_arg()
    2. Opened: doc/html/ncurses-intro.html:3358; form/form.h:346-352; man/form_fieldtype.3x:127
    3. It says: “ncurses-intro: int set_fieldtype_arg(FIELDTYPE *type, INTHOOK succ, INTHOOK pred); form.h: set_fieldtype_arg takes make_arg, copy_arg, free_arg; set_fieldtype_choice takes next_choice, prev_choice. form_fieldtype(3X): The set_fieldtype_choice function allows forms programmers to define successor and predecessor functions”
    4. Verdict: matches, high confidence
    5. One step earlier: With only ncurses-intro, set_fieldtype_arg appears twice and looks deliberate; form.h shows its second use has the other function’s arguments
  • new_fieldtype() prototype without its closing semicolon (docs/client-library/forms/custom-field-types.md)
    1. Claim, as printed: new_fieldtype() prototype without its closing semicolon
    2. Opened: doc/html/ncurses-intro.html:3260-3261
    3. It says: “FIELDTYPE new_fieldtype(HOOK f_validate, / field validator / / HOOK c_validate) / character validator */”
    4. Verdict: matches, high confidence
    5. One step earlier: Reading the rendered HTML, the trailing comment hides the missing semicolon; it shows only in the source line
  • The library routines will that call them will return error indication (docs/client-library/forms/custom-field-types.md)
    1. Claim, as printed: The library routines will that call them will return error indication
    2. Opened: doc/html/ncurses-intro.html:3342
    3. It says: “routines will that call them will return error indication when this happens”
    4. Verdict: matches, high confidence
    5. One step earlier: None; the sentence is ungrammatical as printed
  • should never see a NULL file pointer (docs/client-library/forms/custom-field-types.md)
    1. Claim, as printed: should never see a NULL file pointer
    2. Opened: doc/html/ncurses-intro.html:3344; doc/html/ncurses-intro.html Validation Function Arguments; man/form_fieldtype.3x:115-120
    3. It says: “ncurses-intro: your validation functions should never see a NULL file pointer. Earlier in the same section: If no such arguments are defined for the field type, this pile pointer argument will be NULL”
    4. Verdict: matches, medium confidence
    5. One step earlier: Read alone, file pointer could be taken literally; the section never mentions files, and the argument it describes is the pile pointer, so this is most likely a typo for pile
  • Example comment: This filed is like a password field (docs/client-library/forms/field-options.md)
    1. Claim, as printed: Example comment: This filed is like a password field
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:6074
    3. It says: “field_opts_off(field[1], O_PUBLIC); /* This filed is like a password field */”
    4. Verdict: matches, high confidence
    5. One step earlier: None; filed is a typo for field
  • see infocmp(1) (line 579) (man/curs_getch.3x)
    1. Claim, as printed: see infocmp(1) (line 579)
    2. Opened: man/curs_getch.3x at the line given; man/infocmp.1m .TH; every reference to tic and infocmp in man/; man/man_db.renames.in; aclocal.m4 CF_MANPAGE_RENAMES
    3. It says: “see infocmp(1) as written; the page itself: .TH @TIC@ 1M or .TH @INFOCMP@ 1M; 21 other references use (1M)”
    4. Verdict: matches, high confidence
    5. One step earlier: At the .TH line alone I would have said it is wrong everywhere; on Debian, man_db.renames installs the page in section 1, where (1) happens to resolve
  • tic(1) (line 207) (man/curs_trace.3x)
    1. Claim, as printed: tic(1) (line 207)
    2. Opened: man/curs_trace.3x at the line given; man/tic.1m .TH; every reference to tic and infocmp in man/; man/man_db.renames.in; aclocal.m4 CF_MANPAGE_RENAMES
    3. It says: “tic(1) as written; the page itself: .TH @TIC@ 1M or .TH @INFOCMP@ 1M; 21 other references use (1M)”
    4. Verdict: matches, high confidence
    5. One step earlier: At the .TH line alone I would have said it is wrong everywhere; on Debian, man_db.renames installs the page in section 1, where (1) happens to resolve
  • @TIC@(1) (line 77) (man/curs_extend.3x)
    1. Claim, as printed: @TIC@(1) (line 77)
    2. Opened: man/curs_extend.3x at the line given; man/tic.1m .TH; every reference to tic and infocmp in man/; man/man_db.renames.in; aclocal.m4 CF_MANPAGE_RENAMES
    3. It says: “@TIC@(1) as written; the page itself: .TH @TIC@ 1M or .TH @INFOCMP@ 1M; 21 other references use (1M)”
    4. Verdict: matches, high confidence
    5. One step earlier: At the .TH line alone I would have said it is wrong everywhere; on Debian, man_db.renames installs the page in section 1, where (1) happens to resolve
  • @TIC@(1) (line 669) (man/curs_inopts.3x)
    1. Claim, as printed: @TIC@(1) (line 669)
    2. Opened: man/curs_inopts.3x at the line given; man/tic.1m .TH; every reference to tic and infocmp in man/; man/man_db.renames.in; aclocal.m4 CF_MANPAGE_RENAMES
    3. It says: “@TIC@(1) as written; the page itself: .TH @TIC@ 1M or .TH @INFOCMP@ 1M; 21 other references use (1M)”
    4. Verdict: matches, high confidence
    5. One step earlier: At the .TH line alone I would have said it is wrong everywhere; on Debian, man_db.renames installs the page in section 1, where (1) happens to resolve
  • @TIC@(1) (line 63) (docs/man/term/02-description.5)
    1. Claim, as printed: @TIC@(1) (line 63)
    2. Opened: man/term.5 at the line given; man/tic.1m .TH; every reference to tic and infocmp in man/; man/man_db.renames.in; aclocal.m4 CF_MANPAGE_RENAMES
    3. It says: “@TIC@(1) as written; the page itself: .TH @TIC@ 1M or .TH @INFOCMP@ 1M; 21 other references use (1M)”
    4. Verdict: matches, high confidence
    5. One step earlier: At the .TH line alone I would have said it is wrong everywhere; on Debian, man_db.renames installs the page in section 1, where (1) happens to resolve
  • @TIC@(1) and @INFOCMP@(1) (line 142) (man/term_variables.3x)
    1. Claim, as printed: @TIC@(1) and @INFOCMP@(1) (line 142)
    2. Opened: man/term_variables.3x at the line given; man/infocmp.1m .TH; every reference to tic and infocmp in man/; man/man_db.renames.in; aclocal.m4 CF_MANPAGE_RENAMES
    3. It says: “@TIC@(1) and @INFOCMP@(1) as written; the page itself: .TH @TIC@ 1M or .TH @INFOCMP@ 1M; 21 other references use (1M)”
    4. Verdict: matches, high confidence
    5. One step earlier: At the .TH line alone I would have said it is wrong everywhere; on Debian, man_db.renames installs the page in section 1, where (1) happens to resolve
  • The background character’s set of attributes becomes a property of the character cell and move with it (line 101) (man/curs_bkgd.3x)
    1. Claim, as printed: The background character’s set of attributes becomes a property of the character cell and move with it (line 101)
    2. Opened: man/curs_bkgd.3x:101-104
    3. It says: “The background character’s set of attributes becomes a property of the character cell and move with it through any scrolling and insert/delete line/character operations.”
    4. Verdict: matches, high confidence
    5. One step earlier: Read as the attributes … move, it could pass; the verb before it, becomes, is singular, so its pair must be moves
  • The background character’s set of attributes becomes a property of the character cell and move with it (line 92) (man/curs_bkgrnd.3x)
    1. Claim, as printed: The background character’s set of attributes becomes a property of the character cell and move with it (line 92)
    2. Opened: man/curs_bkgrnd.3x:92-95
    3. It says: “The background character’s set of attributes becomes a property of the character cell and move with it through any scrolling and insert/delete line/character operations.”
    4. Verdict: matches, high confidence
    5. One step earlier: Read as the attributes … move, it could pass; the verb before it, becomes, is singular, so its pair must be moves
  • Post the menu with menu_post(); Unpost the menu with menu_unpost() (docs/client-library/menus/overview.md)
    1. Claim, as printed: Post the menu with menu_post(); Unpost the menu with menu_unpost()
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:4311 and 4321; menu/menu.h:239 and 263
    3. It says: “HOWTO: Post the menu with menu_post() and refresh the screen. / Unpost the menu with menu_unpost(). menu.h: extern MENU_EXPORT(int) post_menu(MENU *); extern MENU_EXPORT(int) unpost_menu(MENU *);”
    4. Verdict: matches, high confidence
    5. One step earlier: Without menu.h it reads like the same slip as the forms chapter; menu.h shows no menu_post or menu_unpost exists
  • The menu can then by posted (docs/client-library/menus/overview.md)
    1. Claim, as printed: The menu can then by posted
    2. Opened: doc/html/ncurses-intro.html:1579
    3. It says: “The menu can then by posted, that is written to an associated window.”
    4. Verdict: matches, high confidence
    5. One step earlier: The sentence alone shows the typo; nothing further to check
  • When you call unpost_menu(), you erase the subwindow, However, neither of these… (docs/client-library/menus/menu-windows.md)
    1. Claim, as printed: When you call unpost_menu(), you erase the subwindow, However, neither of these…
    2. Opened: doc/html/ncurses-intro.html:1711
    3. It says: “erase the subwindow, However, neither of these actually modifies”
    4. Verdict: matches, high confidence
    5. One step earlier: The sentence alone shows the comma splice; nothing further to check
  • see the menu_attribs(3x) manual page. (no closing parenthesis) (docs/client-library/menus/menu-display.md)
    1. Claim, as printed: see the menu_attribs(3x) manual page. (no closing parenthesis)
    2. Opened: doc/html/ncurses-intro.html:1691; the file list of man/
    3. It says: “menu_attribs(3x) manual page.

      — man/ has menu_attributes.3x and no menu_attribs.3x”
    4. Verdict: matches, high confidence
    5. One step earlier: From the HTML alone the page name looks plausible; only the man/ directory shows there is no such page
  • menu_driver() accepts an ascii character (HOWTO); it accumulates printable ASCII characters (Writing Programs) (docs/client-library/menus/menu-driver.md)
    1. Claim, as printed: menu_driver() accepts an ascii character (HOWTO); it accumulates printable ASCII characters (Writing Programs)
    2. Opened: menu/m_driver.c:446; menu/mf_common.h:71; man/menu_driver.3x:55 and 114
    3. It says: “m_driver.c: if (!(c & ~((int)MAX_REGULAR_CHARACTER)) && isprint(UChar(c))) — MAX_REGULAR_CHARACTER is 0xff. menu_driver.3x: The input is a printable character.”
    4. Verdict: matches, medium confidence
    5. One step earlier: The man page says printable character, which alone would not show that ASCII is too narrow; the driver code accepts any byte isprint() accepts in the locale
  • Example 21. Milt Columnar Menus Example (docs/client-library/menus/menu-display.md)
    1. Claim, as printed: Example 21. Milt Columnar Menus Example
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:4939
    3. It says: “Example 21. Milt Columnar Menus Example”
    4. Verdict: matches, high confidence
    5. One step earlier: The title alone shows the typo; nothing further to check
  • The Menu Basics listing frees only items 0 and 1 of five, and calls free_menu() on a menu that is still posted (docs/client-library/menus/compiling.md)
    1. Claim, as printed: The Menu Basics listing frees only items 0 and 1 of five, and calls free_menu() on a menu that is still posted
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:4420-4422; menu/m_new.c:131-132; menu/m_item_new.c:173-174
    3. It says: “HOWTO: free_item(my_items[0]); free_item(my_items[1]); free_menu(my_menu); m_new.c: if (menu->status & _POSTED) RETURN(E_POSTED); m_item_new.c: if (item->imenu) RETURN(E_CONNECTED);”
    4. Verdict: matches, high confidence
    5. One step earlier: The listing alone shows the cleanup is incomplete; m_new.c shows that even free_menu() fails, so nothing is freed at all
  • The Multi Valued Menus listing frees only items 0 and 1 of eight, and calls free_menu() on a menu that is still posted (docs/client-library/menus/selecting-items.md)
    1. Claim, as printed: The Multi Valued Menus listing frees only items 0 and 1 of eight, and calls free_menu() on a menu that is still posted
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:5192-5194; menu/m_new.c:131-132; menu/m_item_new.c:173-174
    3. It says: “HOWTO: free_item(my_items[0]); free_item(my_items[1]); free_menu(my_menu); m_new.c: if (menu->status & _POSTED) RETURN(E_POSTED); m_item_new.c: if (item->imenu) RETURN(E_CONNECTED);”
    4. Verdict: matches, high confidence
    5. One step earlier: The listing alone shows the cleanup is incomplete; m_new.c shows that even free_menu() fails
  • The Menu Options listing frees the items before the menu (docs/client-library/menus/attributes.md)
    1. Claim, as printed: The Menu Options listing frees the items before the menu
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:5330-5333; menu/m_item_new.c:173-174; menu/m_new.c:134-135
    3. It says: “HOWTO: unpost_menu(my_menu); for (…) free_item(my_items[i]); free_menu(my_menu); m_item_new.c: if (item->imenu) RETURN(E_CONNECTED); m_new.c: if (menu->items) _nc_Disconnect_Items(menu);”
    4. Verdict: matches, high confidence
    5. One step earlier: The order looks harmless until m_item_new.c shows free_item() refuses an item still connected to a menu; only free_menu() disconnects them
  • The Menu User Pointer listing frees the items before the menu (docs/client-library/menus/user-pointer-and-hooks.md)
    1. Claim, as printed: The Menu User Pointer listing frees the items before the menu
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:5456-5459; menu/m_item_new.c:173-174; menu/m_new.c:134-135
    3. It says: “HOWTO: unpost_menu(my_menu); for (…) free_item(my_items[i]); free_menu(my_menu); m_item_new.c: if (item->imenu) RETURN(E_CONNECTED);”
    4. Verdict: matches, high confidence
    5. One step earlier: As for the Menu Options listing: the order looks harmless until free_item() is read
  • The Menu User Pointer listing stores the address of MY_DATA data, a variable local to the loop body, as each item’s user pointer and calls through it after the loop (docs/client-library/menus/user-pointer-and-hooks.md)
    1. Claim, as printed: The Menu User Pointer listing stores the address of MY_DATA data, a variable local to the loop body, as each item’s user pointer and calls through it after the loop
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:5415-5420; ISO C (C11 6.2.4p6): an automatic object’s lifetime ends when its block ends
    3. It says: “HOWTO: for (…) { MY_DATA data = {func}; … set_item_userptr(my_items[i], (void *) &data); } … data = item_userptr(cur); data->my_func(item_name(cur));”
    4. Verdict: matches, high confidence
    5. One step earlier: Running the program shows nothing wrong, because the stack slot is not reused; only the scope rule shows the pointer dangles
  • The list of menu options ends at O_NONCYCLIC (docs/client-library/menus/menu-display.md)
    1. Claim, as printed: The list of menu options ends at O_NONCYCLIC
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:5075; menu/menu.h:72; menu/menu.priv.h:76-83; man/menu_opts.3x:90-95
    3. It says: “menu.h: #define O_MOUSE_MENU (0x40). menu.priv.h: ALL_MENU_OPTS includes O_MOUSE_MENU. menu_opts.3x: O_MOUSE_MENU If user clicks with the mouse and it does not fall on the currently active menu, push KEY_MOUSE …”
    4. Verdict: matches, high confidence
    5. One step earlier: The HOWTO quotes the man page of its day, so it looks complete; the current man page and menu.h add the option
  • The mouse requests are quoted from an older menu_driver(3X), with its line breaks and soft hyphens (decora­tion), and without the double-click on an item (docs/client-library/menus/menu-driver.md)
    1. Claim, as printed: The mouse requests are quoted from an older menu_driver(3X), with its line breaks and soft hyphens (decora­tion), and without the double-click on an item
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:4576-4588; man/menu_driver.3x:123-161
    3. It says: “menu_driver.3x: If you double-click an item a REQ_TOGGLE_ITEM is generated and E_UNKNOWN_COMMAND is returned. … If you clicked outside the user window … an E_REQUEST_DENIED is returned.”
    4. Verdict: matches, high confidence
    5. One step earlier: The quoted text is accurate as far as it goes; only the current man page shows what it leaves out
  • A link to text mode dos games at http://www.jersey.net/~debinjoe/games/ (docs/client-library/menus/attributes.md)
    1. Claim, as printed: A link to text mode dos games at http://www.jersey.net/~debinjoe/games/
    2. Opened: http://www.jersey.net/~debinjoe/games/ (fetched 2026-11-14)
    3. It says: “The page is a parked-domain stub: ”
    4. Verdict: matches, high confidence
    5. One step earlier: The server answers 200, so a status check alone would have passed it; the body is a parking redirect
  • the drive assumes it is an application-specific command ([man/menu_driver.3x](/ncurses/man/3x/menu_driver/#APPLICATION-DEFINED COMMANDS))
    1. Claim, as printed: the drive assumes it is an application-specific command
    2. Opened: man/menu_driver.3x:165
    3. It says: “the drive assumes it is an application-specific”
    4. Verdict: matches, high confidence
    5. One step earlier: Found while checking the menus guide against this page; the sentence alone shows the typo
  • menu_driver return one of the following error codes ([man/menu_driver.3x](/ncurses/man/3x/menu_driver/#RETURN VALUE))
    1. Claim, as printed: menu_driver return one of the following error codes
    2. Opened: man/menu_driver.3x:170
    3. It says: “\fBmenu_driver\fP return one of the following error codes:”
    4. Verdict: matches, high confidence
    5. One step earlier: The sentence alone shows the agreement error
  • replace a panel’s window with a different window by calling replace_window (docs/client-library/panels/overview.md)
    1. Claim, as printed: replace a panel’s window with a different window by calling replace_window
    2. Opened: doc/html/ncurses-intro.html:1470; man/panel.3x:63
    3. It says: “intro: You can replace a panel’s window with a different window by calling replace_window. panel.3x: int replace_panel(PANEL *pan, WINDOW *window);”
    4. Verdict: matches, high confidence
    5. One step earlier: The intro alone reads as if replace_window were a function; the synopsis has no such name, only replace_panel
  • The stsdcr window is a special case (docs/client-library/panels/input-and-stdscr.md)
    1. Claim, as printed: The stsdcr window is a special case
    2. Opened: doc/html/ncurses-intro.html:1506
    3. It says: “The stsdcr window is a special case. It is considered below all panels.”
    4. Verdict: matches, high confidence
    5. One step earlier: A transposition; the next sentence and the overview both name stdscr
  • The panel_update code ignores hidden panels (docs/client-library/panels/hiding.md)
    1. Claim, as printed: The panel_update code ignores hidden panels
    2. Opened: doc/html/ncurses-intro.html:1528; man/panel.3x update_panels
    3. It says: “intro: The panel_update code ignores hidden panels. panel.3x: update_panels() refreshes the virtual screen to reflect the relations between the panels in the stack”
    4. Verdict: matches, high confidence
    5. One step earlier: No function is named panel_update; the intro itself calls update_panels() earlier in the chapter
  • You cannot do top_panel() or bottom_panel on a hidden panel() (docs/client-library/panels/hiding.md)
    1. Claim, as printed: You cannot do top_panel() or bottom_panel on a hidden panel()
    2. Opened: doc/html/ncurses-intro.html:1528-1530; panel/p_top.c; panel/p_bottom.c; man/panel.3x PORTABILITY
    3. It says: “p_top.c: top_panel(PANEL * pan) { returnCode(show_panel(pan)); } panel.3x: The functions show_panel and top_panel are identical in this implementation, and work equally well with displayed or hidden panels.”
    4. Verdict: matches, high confidence
    5. One step earlier: The man page warns only about System V; p_bottom.c also links a hidden panel in at the bottom
  • Note that wgetch automatically calls wrefresh (docs/client-library/panels/input-and-stdscr.md)
    1. Claim, as printed: Note that wgetch automatically calls wrefresh
    2. Opened: doc/html/ncurses-intro.html:1512; ncurses/base/lib_getch.c:382
    3. It says: “lib_getch.c: if ((is_wintouched(win) || (win->_flags & _HASMOVED)) && !IS_PAD(win)) { wrefresh(win); }”
    4. Verdict: matches, medium confidence
    5. One step earlier: The advice that follows still holds; only the always is too strong
  • For some reason the function panel_hidden() which tells whether a panel is hidden or not is not working, with a link to a 1999 bug report (docs/client-library/panels/hiding.md)
    1. Claim, as printed: For some reason the function panel_hidden() which tells whether a panel is hidden or not is not working, with a link to a 1999 bug report
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:4067-4071; panel/p_hidden.c:44-54
    3. It says: “p_hidden.c: rc = (IS_LINKED(pan) ? FALSE : TRUE);”
    4. Verdict: matches, high confidence
    5. One step earlier: The bug report is from 1999 and its host, geocrawler.com, is gone; the hiding example now runs on the site using panel_hidden()
  • / Update the stacking order. 2nd panel will be on top / (also in the moving and hiding listings) (docs/client-library/panels/browsing.md)
    1. Claim, as printed: /* Update the stacking order. 2nd panel will be on top */ (also in the moving and hiding listings)
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:3680; man/panel.3x new_panel
    3. It says: “panel.3x: new_panel(win) … places the panel on the top of the stack”
    4. Verdict: matches, high confidence
    5. One step earlier: The comment means my_panels[2], the third panel; the first listing says Last-created panel will be on top
  • The program reads a panel’s position and size from its PANEL_DATA when it comes to the top, but never writes them back, so a moved panel jumps back to where it started the next time it is moved (docs/client-library/panels/moving-and-resizing.md)
    1. Claim, as printed: The program reads a panel’s position and size from its PANEL_DATA when it comes to the top, but never writes them back, so a moved panel jumps back to where it started the next time it is moved
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:3851; 3921; 3935
    3. It says: “set_user_ptrs: ptrs[i].x = x; main: case 9: … newx = top->x; (no other write to top->x)”
    4. Verdict: matches, high confidence
    5. One step earlier: Run on the site: move the third window, Tab round to it again, and move it; it now moves on from where it was left
  • “Show or Hide a window with ‘a’(first window) ‘b’(Second Window) ‘c’(Third Window)”, 83 characters on an 80-column screen (docs/client-library/panels/hiding.md)
    1. Claim, as printed: “Show or Hide a window with ‘a’(first window) ‘b’(Second Window) ‘c’(Third Window)”, 83 characters on an 80-column screen
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:4197
    3. It says: “mvprintw(LINES - 3, 0, “Show or Hide a window with ‘a’(first window) ‘b’(Second Window) ‘c’(Third Window)”);”
    4. Verdict: matches, high confidence
    5. One step earlier: The line wraps onto the next row, where F1 to Exit overwrites it
  • Mainpulate; is used to created panels; removes it form the stack; Entered Resizing :Use (docs/client-library/panels/overview.md)
    1. Claim, as printed: Mainpulate; is used to created panels; removes it form the stack; Entered Resizing :Use
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:3450; 3460; 3945; 4057
    3. It says: “Mainpulate the panels with show_panel() … is used to created panels … merely removes it form the stack”
    4. Verdict: matches, high confidence
    5. One step earlier: Typos
  • bottom_panel on a hidden panel() (docs/client-library/panels/hiding.md)
    1. Claim, as printed: bottom_panel on a hidden panel()
    2. Opened: doc/html/ncurses-intro.html:1529-1530
    3. It says: “cannot do top_panel() or bottom_panel on a hidden panel().”
    4. Verdict: matches, high confidence
    5. One step earlier: The parentheses belong to the function name
  • In the cbreak() mode these control characters are interpreted as any other character by the terminal driver (docs/client-library/getting-started/initialization.md)
    1. Claim, as printed: In the cbreak() mode these control characters are interpreted as any other character by the terminal driver
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:1342; man/curs_inopts.3x:148-158
    3. It says: “curs_inopts: cbreak configures the terminal in cbreak mode, which disables line buffering and erase and kill character processing — the interrupt, quit, suspend, and flow control characters are unaffected”
    4. Verdict: matches, high confidence
    5. One step earlier: The HOWTO says the opposite of what it means: the contrast with raw() is that cbreak() keeps the signals
  • endwin() frees the memory taken by curses sub-system and its data structures and puts the terminal in normal mode (docs/client-library/getting-started/hello-world.md)
    1. Claim, as printed: endwin() frees the memory taken by curses sub-system and its data structures and puts the terminal in normal mode
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:1282; man/curs_initscr.3x:185-232, 266-268
    3. It says: “curs_initscr: endwin … restores terminal modes … Calling refresh(3X) or doupdate(3X) after a temporary suspension causes curses to resume managing the terminal. delscreen frees the storage backing the supplied SCREEN”
    4. Verdict: matches, high confidence
    5. One step earlier: If endwin() freed the data structures, curses could not resume after it; delscreen() is the function that frees them
  • <curses.h> also includes <termios.h>, <termio.h>, or <sgtty.h> depending on your system (docs/client-library/getting-started/hello-world.md)
    1. Claim, as printed: <curses.h> also includes <termios.h>, <termio.h>, or <sgtty.h> depending on your system
    2. Opened: doc/html/ncurses-intro.html:487-489; include/curses.h.in:250
    3. It says: “curses.h.in: #include <stdio.h> (no termios.h, termio.h or sgtty.h anywhere in the file)”
    4. Verdict: matches, high confidence
    5. One step earlier: The sentence describes older curses headers
  • In linking with curses you need to have -lncurses … There is no need for any other libraries (docs/client-library/getting-started/hello-world.md)
    1. Claim, as printed: In linking with curses you need to have -lncurses … There is no need for any other libraries
    2. Opened: doc/html/ncurses-intro.html:490-493; man/ncurses.3x:1880-1901; INSTALL:713-714
    3. It says: “ncurses.3x: An application desirous of wide-character support then uses -lncursesw (or -ltinfow) instead of -lncurses (or -ltinfo). INSTALL: generate .pc files for each of the libraries”
    4. Verdict: matches, medium confidence
    5. One step earlier: True for the default shared build; not for every configuration the project supports
  • To use ncurses library functions, you have to include ncurses.h in your programs (docs/client-library/getting-started/hello-world.md)
    1. Claim, as printed: To use ncurses library functions, you have to include ncurses.h in your programs
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:1174; man/ncurses.3x:1844; include/Makefile (generated) ln -s curses.h ncurses.h
    3. It says: “ncurses.3x: #include <curses.h>. The HOWTO’s own Hello World includes <curses.h>”
    4. Verdict: matches, high confidence
    5. One step earlier: ncurses.h is a link to curses.h, so it works, but it is not required and is not portable
  • bool: boolean type, actually a “char” (docs/client-library/getting-started/naming-and-variables.md)
    1. Claim, as printed: bool: boolean type, actually a “char”
    2. Opened: doc/html/ncurses-intro.html:599; include/curses.h.in:290-302
    3. It says: “curses.h.in: typedef @cf_cv_type_of_bool@ NCURSES_BOOL; … #include <stdbool.h> /* use whatever the C compiler decides bool really is */”
    4. Verdict: matches, high confidence
    5. One step earlier: configure measures the type; on a C99 compiler it is _Bool
  • OK: error flag returned by routines when things go right (docs/client-library/getting-started/naming-and-variables.md)
    1. Claim, as printed: OK: error flag returned by routines when things go right
    2. Opened: doc/html/ncurses-intro.html:620; include/curses.h.in:429
    3. It says: “#define OK (0)”
    4. Verdict: matches, high confidence
    5. One step earlier: OK is not an error flag; the value is added to match ERR’s entry
  • Changes should be made to through the API; To write to a different window. use the routine waddch() … is provided (docs/client-library/getting-started/naming-and-variables.md)
    1. Claim, as printed: Changes should be made to through the API; To write to a different window. use the routine waddch() … is provided
    2. Opened: doc/html/ncurses-intro.html:534; 540
    3. It says: “Changes should be made to through the API … To write to a different window. use the routine waddch()”
    4. Verdict: matches, high confidence
    5. One step earlier: Typos
  • rearing to get started; halfdelay()is called; if he doesn’t respond with in certain time (docs/client-library/getting-started/initialization.md)
    1. Claim, as printed: rearing to get started; halfdelay()is called; if he doesn’t respond with in certain time
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:809; 1389; 1396
    3. It says: “you must be rearing to get started … halfdelay()is called … if he doesn’t respond with in certain time”
    4. Verdict: matches, high confidence
    5. One step earlier: Typos; the sentence with rearing was dropped with the HOWTO’s chapter 1.3 lead-in
  • Mail links to zmbenhal@netcom.com and esr@snark.thyrsus.com, and a link to adahome.com for the Ada95 binding (docs/client-library/getting-started/history.md)
    1. Claim, as printed: Mail links to zmbenhal@netcom.com and esr@snark.thyrsus.com, and a link to adahome.com for the Ada95 binding
    2. Opened: doc/html/ncurses-intro.html:417-421
    3. It says: “The original maintainer of this package is Zeyd Ben-Halim zmbenhal@netcom.com”
    4. Verdict: matches, medium confidence
    5. One step earlier: Not checked that the addresses bounce; they predate the current maintainer, and the maintainers’ list address stays as the contact. adahome.com now redirects elsewhere
  • create_box() puts the corners at x + w and y + h, so a box of width 10 and height 3 is drawn 11 columns wide and 4 rows tall, and the erase loop clears one row and column more (docs/client-library/windows/border-functions.md)
    1. Claim, as printed: create_box() puts the corners at x + w and y + h, so a box of width 10 and height 3 is drawn 11 columns wide and 4 rows tall, and the erase loop clears one row and column more
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:2501-2507
    3. It says: “mvaddch(y, x + w, p_win->border.tr); … mvvline(y + 1, x + w, p_win->border.rs, h - 1);”
    4. Verdict: matches, high confidence
    5. One step earlier: Run on the site, the original box was 11 by 4 beside the first example’s 10 by 3 window from the same height and width; it now matches
  • printw(string); mvprintw(y, x, string); … (docs/client-library/windows/windows-and-refresh.md)
    1. Claim, as printed: printw(string); mvprintw(y, x, string); …
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:1520; man/curs_printw.3x
    3. It says: “curs_printw: int printw(const char *fmt, …);”
    4. Verdict: matches, medium confidence
    5. One step earlier: The first argument is a format; passing an arbitrary string there misprints any % it contains
  • let me clear few things; It doesn’t create any thing; as user press a key; Let’s dissect it by line by line (docs/client-library/windows/creating-windows.md)
    1. Claim, as printed: let me clear few things; It doesn’t create any thing; as user press a key; Let’s dissect it by line by line
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:1487; 2366
    3. It says: “let me clear few things about windows … destroys windows as user press a key”
    4. Verdict: matches, high confidence
    5. One step earlier: Typos
  • If the terminal does not have such characters, curses.h will map them to a recognizable (though ugly) set of ASCII defaults (docs/client-library/output/forms-characters.md)
    1. Claim, as printed: If the terminal does not have such characters, curses.h will map them to a recognizable (though ugly) set of ASCII defaults
    2. Opened: doc/html/ncurses-intro.html:804; man/curs_addch.3x:293-299
    3. It says: “curses uses the ACS default listed below if the terminal type lacks the acs_chars (acsc) capability”
    4. Verdict: matches, medium confidence
    5. One step earlier: The fallback comes from the library at run time, from the terminal description, not from the header at compile time
  • let’s interact with world; message to be appeared on the screen; Try resizing your window(if possible); ch(of type char); the character printed … until it is turned off (docs/client-library/output/)
    1. Claim, as printed: let’s interact with world; message to be appeared on the screen; Try resizing your window(if possible); ch(of type char); the character printed … until it is turned off
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:1539
    3. It says: “Now that curses is initialized, let’s interact with world.”
    4. Verdict: matches, high confidence
    5. One step earlier: Typos and grammar
  • With _DEBUG defined, print_win_params() prints at row 25, below the bottom of a 24-row screen, so nothing is shown (docs/client-library/windows/border-functions.md)
    1. Claim, as printed: With _DEBUG defined, print_win_params() prints at row 25, below the bottom of a 24-row screen, so nothing is shown
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:2480; man/curs_move.3x:83
    3. It says: “mvprintw(25, 0, “%d %d %d %d”, p_win->startx, p_win->starty,”
    4. Verdict: matches, high confidence
    5. One step earlier: Run on the site with _DEBUG defined, the original line was never shown on an 80x24 screen; it now shows 35 10 10 3
  • char str[80]; … getstr(str); (docs/client-library/input/characters-and-strings.md)
    1. Claim, as printed: char str[80]; … getstr(str);
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:1927; include/curses.h.in:1222; man/curs_getstr.3x:155-160
    3. It says: “#define wgetstr(w, s) wgetnstr(w, s, -1); a negative n reads up to LINE_MAX characters”
    4. Verdict: matches, high confidence
    5. One step earlier: getstr() can write up to LINE_MAX (usually 2048) bytes into an 80-byte buffer; typing a long line overruns it
  • Example 4. A Simple scanw example (docs/client-library/input/characters-and-strings.md)
    1. Claim, as printed: Example 4. A Simple scanw example
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:1908
    3. It says: “The program calls getstr() and never scanw()”
    4. Verdict: matches, high confidence
    5. One step earlier: The title named a function the program does not use
  • These pseudo-line-oriented functions turn on echoing while they execute (docs/client-library/input/characters-and-strings.md)
    1. Claim, as printed: These pseudo-line-oriented functions turn on echoing while they execute
    2. Opened: doc/html/ncurses-intro.html:778; man/curs_getstr.3x:118-123; ncurses/base/lib_getstr.c:99,123
    3. It says: “If the screen’s echo option is enabled, wgetstr updates win with waddch”
    4. Verdict: matches, high confidence
    5. One step earlier: ncurses turns echo off while reading and echoes each character itself only if echo was on
  • begin read them; must used; one or the other has to changed; The resulting string of characters are pointed to (docs/client-library/input/)
    1. Claim, as printed: begin read them; must used; one or the other has to changed; The resulting string of characters are pointed to
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:1826-1828; doc/html/ncurses-intro.html:770
    3. It says: “will begin read them only after a new line”
    4. Verdict: matches, high confidence
    5. One step earlier: Typos and grammar
  • Initially RED color is defined with content 1000(r), 0(g), 0(b); init_color(COLOR_RED, 700, 0, 0) is described as lightening red by a minuscule (docs/client-library/attributes/colors.md)
    1. Claim, as printed: Initially RED color is defined with content 1000(r), 0(g), 0(b); init_color(COLOR_RED, 700, 0, 0) is described as lightening red by a minuscule
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:2691,2703; ncurses/base/lib_color.c:106,109-113
    3. It says: “#define RGB_ON 680 … DATA(RGB_ON, RGB_OFF, RGB_OFF), /* COLOR_RED */”
    4. Verdict: matches, high confidence
    5. One step earlier: ncurses starts its eight basic colors at 680, not 1000
  • attr_get() is useful in scanning areas of screen when we are not sure what attribute each character is associated with (docs/client-library/attributes/attribute-functions.md)
    1. Claim, as printed: attr_get() is useful in scanning areas of screen when we are not sure what attribute each character is associated with
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:2109; man/curs_attr.3x:80,133-140
    3. It says: “wattr_set sets the current attributes of the given window … wattr_get”
    4. Verdict: matches, medium confidence
    5. One step earlier: attr_get() returns the window’s rendition, not the attributes of any cell on the screen
  • Other functions wchgat(), mvchgat(), wchgat() behave similarly (docs/client-library/attributes/chgat.md)
    1. Claim, as printed: Other functions wchgat(), mvchgat(), wchgat() behave similarly
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:2163; man/curs_attr.3x:64-67
    3. It says: “chgat, wchgat, mvchgat, mvwchgat”
    4. Verdict: matches, high confidence
    5. One step earlier: wchgat() was listed twice and mvwchgat() was missing
  • / pager functionality by Joseph Spainhour” spainhou@bellsouth.net / (docs/client-library/attributes/attributes.md)
    1. Claim, as printed: /* pager functionality by Joseph Spainhour” spainhou@bellsouth.net */
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:1957
    3. It says: “Joseph Spainhour” spainhou@bellsouth.net”
    4. Verdict: matches, high confidence
    5. One step earlier: An unmatched quote in the credit comment
  • screeen; are we are at; cut bold on; Forth parameter; Viola !!!; loose track; the the source; Don’t worry about all those initialization and other crap (docs/client-library/attributes/)
    1. Claim, as printed: screeen; are we are at; cut bold on; Forth parameter; Viola !!!; loose track; the the source; Don’t worry about all those initialization and other crap
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:1979,1984,1994,2013,2091,2192,2631,2660
    3. It says: “are we are at the end of the screen”
    4. Verdict: matches, high confidence
    5. One step earlier: Typos and grammar
  • mvprintw(24, 0, “Character pressed is …”); mvprintw(23, 0, “You chose choice …”) (docs/client-library/keyboard/key-usage-example.md)
    1. Claim, as printed: mvprintw(24, 0, “Character pressed is …”); mvprintw(23, 0, “You chose choice …”)
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:2867,2877
    3. It says: “ mvprintw(24, 0,”
    4. Verdict: matches, high confidence
    5. One step earlier: Row 24 is off a 24-row screen (rows 0 to 23), so the key code was never shown
  • startx = (80 - WIDTH) / 2; starty = (24 - HEIGHT) / 2; (docs/client-library/keyboard/key-usage-example.md)
    1. Claim, as printed: startx = (80 - WIDTH) / 2; starty = (24 - HEIGHT) / 2;
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:2839-2840
    3. It says: “startx = (80 - WIDTH) / 2;”
    4. Verdict: matches, high confidence
    5. One step earlier: The menu was centered for an 80x24 screen whatever the terminal size
  • the Functions keys; See the initialization section; if(ch == KEY_LEFT); User did a choice come out of the infinite loop (docs/client-library/keyboard/)
    1. Claim, as printed: the Functions keys; See the initialization section; if(ch == KEY_LEFT); User did a choice come out of the infinite loop
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:2742-2743,2768,2874
    3. It says: “be enabled to get the Functions keys”
    4. Verdict: matches, high confidence
    5. One step earlier: Typos and grammar
  • The default is one fifth of a second (docs/client-library/mouse/mouse-events.md)
    1. Claim, as printed: The default is one fifth of a second
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:3138; ncurses/curses.priv.h:184; man/curs_mouse.3x:436-438
    3. It says: “#define DEFAULT_MAXCLICK 166”
    4. Verdict: matches, high confidence
    5. One step earlier: ncurses starts the mouse interval at 166 ms
  • if (event.bstate & BUTTON1_PRESSED) with mousemask(ALL_MOUSE_EVENTS, NULL) (docs/client-library/mouse/mouse-menu-example.md)
    1. Claim, as printed: if (event.bstate & BUTTON1_PRESSED) with mousemask(ALL_MOUSE_EVENTS, NULL)
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:3096,3103; man/curs_mouse.3x:402-428
    3. It says: “If you do not make clicks visible, they will be reported as press-release pairs”
    4. Verdict: matches, high confidence
    5. One step earlier: With clicks in the mask, a quick click arrives only as BUTTON1_CLICKED, so the menu ignored ordinary clicks; checked on a native pty and in the browser
  • report_choice(event.x + 1, event.y + 1, …); j = starty + 3; mouse_x <= i + strlen(…) (docs/client-library/mouse/mouse-menu-example.md)
    1. Claim, as printed: report_choice(event.x + 1, event.y + 1, …); j = starty + 3; mouse_x <= i + strlen(…)
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:3053,3058,3104
    3. It says: “report_choice(event.x + 1, event.y + 1, &choice);”
    4. Verdict: matches, high confidence
    5. One step earlier: The +1 offsets accepted a click in the blank column left of each choice
  • startx = (80 - WIDTH) / 2; starty = (24 - HEIGHT) / 2; mvprintw(23, 1, …); mvprintw(22, 1, …) (docs/client-library/mouse/mouse-menu-example.md)
    1. Claim, as printed: startx = (80 - WIDTH) / 2; starty = (24 - HEIGHT) / 2; mvprintw(23, 1, …); mvprintw(22, 1, …)
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:3086,3107
    3. It says: “mvprintw(23, 1,”
    4. Verdict: matches, high confidence
    5. One step earlier: The menu assumed an 80x24 screen
  • The event mask table stops at BUTTON4_TRIPLE_CLICKED (docs/client-library/mouse/mouse-events.md)
    1. Claim, as printed: The event mask table stops at BUTTON4_TRIPLE_CLICKED
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:2940; man/curs_mouse.3x:216-220; include/curses.tail:52-55
    3. It says: “BUTTON5_PRESSED mouse button 5 down”
    4. Verdict: matches, high confidence
    5. One step earlier: Button 5 events were missing
  • lets do the same thing from mouse; any thing; a class of mouse events have been enabled; Do some thing; thousands of a second; convert to mouse co-ordinates to screen relative (docs/client-library/mouse/)
    1. Claim, as printed: lets do the same thing from mouse; any thing; a class of mouse events have been enabled; Do some thing; thousands of a second; convert to mouse co-ordinates to screen relative
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:2894,2904,2954,2966,3131,3135
    3. It says: “in thousands of a second”
    4. Verdict: matches, high confidence
    5. One step earlier: Typos and grammar
  • getbegyx() and getmaxyx() store the window’s beginning and maximum co-ordinates (docs/client-library/screen/coordinates.md)
    1. Claim, as printed: getbegyx() and getmaxyx() store the window’s beginning and maximum co-ordinates
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:3179; man/curs_getyx.3x:74-79
    3. It says: “curs_getyx: maximum valid line and column dimensions … Those are one more than the corresponding maximum positions.”
    4. Verdict: matches, high confidence
    5. One step earlier: A program that loops up to the value getmaxyx() gives, reading it as the last coordinate, writes one column past the edge
  • The values returned can be 1 (all is well), 0 (no such terminal), or -1 (some problem locating the terminfo database) (docs/client-library/init-and-debugging/terminal-descriptions.md)
    1. Claim, as printed: The values returned can be 1 (all is well), 0 (no such terminal), or -1 (some problem locating the terminfo database)
    2. Opened: doc/html/ncurses-intro.html:1066-1068; man/curs_terminfo.3x:244-290
    3. It says: “curs_terminfo: If ERR is returned, examine errret. 1 means that the terminal is a hard-copy type … 0 means that the terminal could not be found, or that it is a generic type”
    4. Verdict: matches, high confidence
    5. One step earlier: Read alone, a status of 1 looks like success even when setupterm() returned ERR for a hard-copy terminal
  • trace() sets the trace level; _tracef() is available only if you link with -lncurses_g (docs/client-library/init-and-debugging/debugging.md)
    1. Claim, as printed: trace() sets the trace level; _tracef() is available only if you link with -lncurses_g
    2. Opened: doc/html/ncurses-intro.html:1101-1120; man/curs_trace.3x:138,300
    3. It says: “curs_trace: The original trace routine was deprecated because it often conflicted with application names. … If the file already exists, no tracing is done.”
    4. Verdict: matches, high confidence
    5. One step earlier: A reader following the HOWTO would call a deprecated function and wonder why no trace file appears when an old one is still in the directory
  • endwin() should called once (docs/client-library/init-and-debugging/initialization-and-wrapup.md)
    1. Claim, as printed: endwin() should called once
    2. Opened: doc/html/ncurses-intro.html:992
    3. It says: “endwin() should called once at wrapup time for each terminal opened using this function.”
    4. Verdict: matches, high confidence
    5. One step earlier: The sentence alone shows the missing be; nothing further to check
  • It is only available only if (docs/client-library/init-and-debugging/debugging.md)
    1. Claim, as printed: It is only available only if
    2. Opened: doc/html/ncurses-intro.html:1119
    3. It says: “It is only available only if you link with -lncurses_g.”
    4. Verdict: matches, high confidence
    5. One step earlier: The doubled only is plain from the sentence; the rest of the claim is corrected in the curses_trace row
  • Setupterm() also stores (docs/client-library/init-and-debugging/terminal-descriptions.md)
    1. Claim, as printed: Setupterm() also stores
    2. Opened: doc/html/ncurses-intro.html:1085
    3. It says: “Setupterm() also stores the names section of the terminal description”
    4. Verdict: matches, high confidence
    5. One step earlier: The function name is lowercase everywhere else, and C names are case-sensitive
  • When ncurses returns that code, it calls resizeterm (docs/client-library/hints/resizing.md)
    1. Claim, as printed: When ncurses returns that code, it calls resizeterm
    2. Opened: doc/html/ncurses-intro.html:1234; man/resizeterm.3x:155
    3. It says: “The ncurses library provides a SIGWINCH signal handler, which pushes a KEY_RESIZE via the wgetch() calls. When ncurses returns that code, it calls resizeterm”
    4. Verdict: matches, high confidence
    5. One step earlier: resizeterm(3X) gives the order: the handler sets a flag, wgetch() or doupdate() calls resizeterm(), which ungetch()es KEY_RESIZE
  • The refresh will pick up the new screen size from the xterm’s environment (docs/client-library/hints/resizing.md)
    1. Claim, as printed: The refresh will pick up the new screen size from the xterm’s environment
    2. Opened: doc/html/ncurses-intro.html:1221
    3. It says: “The refresh will pick up the new screen size from the xterm’s environment.”
    4. Verdict: matches, medium confidence
    5. One step earlier: ncurses asks the terminal driver for the size; the LINES and COLUMNS environment variables override it rather than supply it
  • Note that you will also have to call def_shell_mode and def_prog_mode on each tty yourself (docs/client-library/hints/terminals.md)
    1. Claim, as printed: Note that you will also have to call def_shell_mode and def_prog_mode on each tty yourself
    2. Opened: doc/html/ncurses-intro.html:1256; man/curs_kernel.3x:92
    3. It says: “Note that you will also have to call def_shell_mode and def_prog_mode on each tty yourself. / curs_kernel: This is done automatically by initscr. There is one such save area for each screen context allocated by newterm.”
    4. Verdict: matches, high confidence
    5. One step earlier: initscr() does its work through newterm(), so the saved modes come with each new screen
  • Ncurses supports up 16 colors, unlike SVr4 curses which defines only 8. While most terminals which provide color allow only 8 colors, about a quarter (including XFree86 xterm) support 16 colors. (docs/client-library/hints/extensions.md)
    1. Claim, as printed: Ncurses supports up 16 colors, unlike SVr4 curses which defines only 8. While most terminals which provide color allow only 8 colors, about a quarter (including XFree86 xterm) support 16 colors.
    2. Opened: doc/html/ncurses-intro.html:1305; man/curs_color.3x:198
    3. It says: “Ncurses supports up 16 colors, unlike SVr4 curses which defines only 8.”
    4. Verdict: matches, high confidence
    5. One step earlier: The 16-color limit is long gone; COLORS comes from the terminal description, and the sentence was also missing a to
  • the attribute of erased blanks is normal unless and until it is modified by the functions bkgdset() or wbkgdset() (docs/client-library/hints/compatibility.md)
    1. Claim, as printed: the attribute of erased blanks is normal unless and until it is modified by the functions bkgdset() or wbkgdset()
    2. Opened: doc/html/ncurses-intro.html:1388; man/curs_bkgd.3x:86
    3. It says: “Instead, the attribute of erased blanks is normal unless and until it is modified by the functions bkgdset() or wbkgdset().”
    4. Verdict: matches, high confidence
    5. One step earlier: bkgd() and wbkgd() set the background too, and erasing fills with the whole background character, not only its attribute
  • every macro entry point have a corresponding function which may be linked (docs/client-library/hints/compatibility.md)
    1. Claim, as printed: every macro entry point have a corresponding function which may be linked
    2. Opened: doc/html/ncurses-intro.html:1407; man/curs_getyx.3x:107
    3. It says: “ncurses meets the XSI requirement that every macro entry point have a corresponding function which may be linked / curs_getyx: All of these interfaces are implemented as macros.”
    4. Verdict: matches, high confidence
    5. One step earlier: The getyx() family store into their arguments, so they can only be macros
  • mvprintw(24, 0, “Press F1 to Exit”) (docs/client-library/more/hanoi.md)
    1. Claim, as printed: mvprintw(24, 0, “Press F1 to Exit”)
    2. Opened: docs/client-library/more/hanoi.md:120 (the program as this edition runs it on the site); the original: ncurses_programs.tar.gz JustForFun/hanoi.c:107
    3. It says: “mvprintw(24, 0, “Press F1 to Exit”);”
    4. Verdict: matches, high confidence
    5. One step earlier: Rows count from 0, so on a 24-line terminal row 24 is off the screen and the hint never showed; the smoke test runs on 80x24
  • mvprintw(24, 0, …) for “Press F1 to Exit” and “You Win !!!” (docs/client-library/more/shuffle.md)
    1. Claim, as printed: mvprintw(24, 0, …) for “Press F1 to Exit” and “You Win !!!”
    2. Opened: docs/client-library/more/shuffle.md:174 and 232 (the program as this edition runs it on the site); the original: ncurses_programs.tar.gz JustForFun/shuffle.c:165; ncurses_programs.tar.gz JustForFun/shuffle.c:223
    3. It says: “mvprintw(24, 0, “You Win !!!\n”);”
    4. Verdict: matches, high confidence
    5. One step earlier: Row 24 is past the last row of a 24-line terminal, so neither message showed
  • #define STATUSY 25 (docs/client-library/more/typing-tutor.md)
    1. Claim, as printed: #define STATUSY 25
    2. Opened: docs/client-library/more/typing-tutor.md:25 (the program as this edition runs it on the site); the original: ncurses_programs.tar.gz JustForFun/tt.c:16
    3. It says: “#define STATUSY 25”
    4. Verdict: matches, high confidence
    5. One step earlier: The status line was two rows below a 24-line screen, so its prompts never showed
  • http://www2.linuxjournal.com/lj-issues/issue5/2807.html (docs/client-library/more/widget-libraries.md)
    1. Claim, as printed: http://www2.linuxjournal.com/lj-issues/issue5/2807.html
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:7339
    3. It says: “article on dialog in Linux Journal”
    4. Verdict: matches, high confidence
    5. One step earlier: The old host no longer answers; the new address serves the same article, Dialog: An Introductory Tutorial
  • The third objective (standards conformance) (docs/contributor/hackguide/objective.md)
    1. Claim, as printed: The third objective (standards conformance)
    2. Opened: docs/contributor/hackguide/objective.md:8 (the second item of the list); doc/html/hackguide.html:162
    3. It says: “- Conformant with the XSI Curses standard issued as part of XPG4 by X/Open.”
    4. Verdict: matches, high confidence
    5. One step earlier: Without counting the list, the cross-reference reads as plausible
  • This is a majordomo list (docs/contributor/hackguide/reporting-bugs.md)
    1. Claim, as printed: This is a majordomo list
    2. Opened: https://lists.gnu.org/mailman/listinfo/bug-ncurses; INSTALL:2584; ANNOUNCE:660-687
    3. It says: “Bug-ncurses Info Page (the Mailman listinfo page)”
    4. Verdict: matches, high confidence
    5. One step earlier: INSTALL gives only the address, so the list software would have been unchecked
  • trace() (docs/contributor/hackguide/reporting-bugs.md)
    1. Claim, as printed: trace()
    2. Opened: man/curs_trace.3x:300
    3. It says: “The original trace routine was deprecated because”
    4. Verdict: matches, high confidence
    5. One step earlier: trace() still links as a compatibility name, so a build test alone would have passed
  • Test frames for the library algorithms are in the progs directory (docs/contributor/hackguide/reporting-bugs.md)
    1. Claim, as printed: Test frames for the library algorithms are in the progs directory
    2. Opened: ncurses/Makefile.in:205-216, 352, 378; ncurses/tty/lib_mvcur.c:1209
    3. It says: “(void) puts(“The mvcur tester. Type ? for help”);”
    4. Verdict: matches, high confidence
    5. One step earlier: progs/ exists and holds the utilities, so the directory name alone looked right
  • If the entire sequence takes less than 1 second (docs/contributor/hackguide/engine-room.md)
    1. Claim, as printed: If the entire sequence takes less than 1 second
    2. Opened: ncurses/base/lib_getch.c:66; ncurses/base/lib_getch.c:682-690
    3. It says: “NCURSES_EXPORT_VAR(int) ESCDELAY = 1000;”
    4. Verdict: matches, medium confidence
    5. One step earlier: The default is 1 second, so the figure matches; only the kgetch comment shows the timer is per character
  • timed_wait() (docs/contributor/hackguide/engine-room.md)
    1. Claim, as printed: timed_wait()
    2. Opened: ncurses/tty/lib_twait.c:39; ncurses/tty/lib_twait.c:165
    3. It says: “_nc_timed_wait(const SCREEN *sp MAYBE_UNUSED,”
    4. Verdict: matches, high confidence
    5. One step earlier: The file name lib_twait.c matches, which would have hidden the renamed function
  • lib_doupdate.c (docs/contributor/hackguide/engine-room.md)
    1. Claim, as printed: lib_doupdate.c
    2. Opened: ncurses/tty/tty_update.c
    3. It says: “doupdate(NCURSES_SP_DCL0) is defined in tty_update.c; no lib_doupdate.c exists in the tree”
    4. Verdict: matches, high confidence
    5. One step earlier: The function name doupdate matches, so only a file search shows the rename
  • _nc_mvcur_scrolln() in lib_mvcur.c (docs/contributor/hackguide/engine-room.md)
    1. Claim, as printed: _nc_mvcur_scrolln() in lib_mvcur.c
    2. Opened: ncurses/tty/tty_update.c:2023; ncurses/tty/tty_update.c:2160
    3. It says: “NCURSES_SP_NAME(_nc_scrolln)(NCURSES_SP_DCLx”
    4. Verdict: matches, high confidence
    5. One step earlier: lib_mvcur.c still exists, so the file looked current
  • See include/curses.h.in for mask values (docs/contributor/hackguide/engine-room.md)
    1. Claim, as printed: See include/curses.h.in for mask values
    2. Opened: include/curses.tail:231-234
    3. It says: “#define OPTIMIZE_MVCUR 0x01 /* cursor movement optimization */”
    4. Verdict: matches, high confidence
    5. One step earlier: curses.h.in exists, so the citation looked live
  • The lib_mouse() code (docs/contributor/hackguide/engine-room.md)
    1. Claim, as printed: The lib_mouse() code
    2. Opened: ncurses/base/lib_mouse.c
    3. It says: “No function lib_mouse() is defined; lib_mouse.c holds _nc_mouse_event, _nc_mouse_inline and _nc_mouse_parse”
    4. Verdict: matches, medium confidence
    5. One step earlier: Read loosely as the file, the phrase is only imprecise
  • will settle for BSD regexps (docs/contributor/hackguide/forms-and-menus.md)
    1. Claim, as printed: will settle for BSD regexps
    2. Opened: aclocal.m4:7305; aclocal.m4:7320
    3. It says: “AC_CHECK_FUNC(compile,[cf_regex_func=compile],[ … for cf_regex_hdr in regexp.h regexpr.h”
    4. Verdict: matches, medium confidence
    5. One step earlier: BSD has its own regex functions, so the phrase sounded right without the macro
  • ncurses/dump_entry.c (docs/contributor/hackguide/terminfo-compiler.md)
    1. Claim, as printed: ncurses/dump_entry.c
    2. Opened: progs/dump_entry.c
    3. It says: “The file exists in progs/; ncurses/ has none”
    4. Verdict: matches, high confidence
    5. One step earlier: The file name matches; only the directory moved
  • ncurses/MKlib_gen.sh (docs/contributor/hackguide/style-tips.md)
    1. Claim, as printed: ncurses/MKlib_gen.sh
    2. Opened: ncurses/base/MKlib_gen.sh
    3. It says: “The script exists under ncurses/base/”
    4. Verdict: matches, high confidence
    5. One step earlier: The file name matches; only the directory moved
  • The fifth field (flags) of include/Caps (docs/contributor/hackguide/style-tips.md)
    1. Claim, as printed: The fifth field (flags) of include/Caps
    2. Opened: include/Caps:49-57
    3. It says: “# Column 7: Lead with `Y’ if capability should be emitted in termcap”
    4. Verdict: matches, high confidence
    5. One step earlier: Without the header comment in Caps, the field count could not have been checked
  • lib_getstr.c is pure curses, but calls outstr() (docs/contributor/hackguide/porting-hints.md)
    1. Claim, as printed: lib_getstr.c is pure curses, but calls outstr()
    2. Opened: doc/html/hackguide.html:835; ncurses/base/lib_getstr.c; a search for outstr in the whole tree
    3. It says: “No outstr() is defined or called anywhere in the tree”
    4. Verdict: matches, high confidence
    5. One step earlier: The pure-curses half matches; only a tree search shows outstr() is gone
  • lib_insstr.c (docs/contributor/hackguide/porting-hints.md)
    1. Claim, as printed: lib_insstr.c
    2. Opened: doc/html/hackguide.html:409; doc/html/hackguide.html:828; ncurses/base/lib_insnstr.c:47
    3. It says: “winsnstr(WINDOW *win, const char *s, int n)”
    4. Verdict: matches, high confidence
    5. One step earlier: The function names are the same, which would have hidden the renamed file
  • lib_unctrl.c (docs/contributor/hackguide/porting-hints.md)
    1. Claim, as printed: lib_unctrl.c
    2. Opened: ncurses/Makefile.in; ncurses/base/MKunctrl.awk
    3. It says: “unctrl.c is written by MKunctrl.awk”
    4. Verdict: matches, high confidence
    5. One step earlier: unctrl() still exists, so the name looked current
  • comp_main.c (docs/contributor/hackguide/porting-hints.md)
    1. Claim, as printed: comp_main.c
    2. Opened: progs/tic.c
    3. It says: “tic’s main() is in progs/tic.c; no comp_main.c exists”
    4. Verdict: matches, high confidence
    5. One step earlier: The other comp_*.c files are still there, which made comp_main.c plausible
  • The configure script attempts to locate previous installation of ncurses, and will set the default prefix according to where it finds the ncurses headers (docs/contributor/install/installation-procedure.md)
    1. Claim, as printed: The configure script attempts to locate previous installation of ncurses, and will set the default prefix according to where it finds the ncurses headers
    2. Opened: aclocal.m4:972-985 (CF_CFG_DEFAULTS); a configure run on Linux
    3. It says: “(openbsd*|freebsd*|mirbsd*|linux*|cygwin*|msys*|kbsd-gnu|mingw*) prefix=/usr ;; (*) prefix=$ac_default_prefix”
    4. Verdict: matches, high confidence
    5. One step earlier: The configure run printed “checking for prefix… /usr” and no search for headers; the macro only uses existing headers to pick includedir, and only when the prefix is still unset
  • Typing ‘configure’ with no arguments is equivalent to: ./configure –with-normal –with-debug –enable-overwrite (docs/contributor/install/installation-procedure.md)
    1. Claim, as printed: Typing ‘configure’ with no arguments is equivalent to: ./configure –with-normal –with-debug –enable-overwrite
    2. Opened: configure.in:563-568
    3. It says: “AC_ARG_ENABLE(overwrite, … [if test “$prefix” = “/usr” ; then with_overwrite=yes; else with_overwrite=no; fi])”
    4. Verdict: matches, high confidence
    5. One step earlier: With the /usr/local default most systems get, the equivalence was wrong; on Linux, where the prefix is /usr, it happened to hold
  • read the PORTABILITY section in the file ncurses/README (docs/contributor/install/installation-procedure.md)
    1. Claim, as printed: read the PORTABILITY section in the file ncurses/README
    2. Opened: ncurses/README:31
    3. It says: “For discussion of the package internals, see hackguide.html in the doc/html”
    4. Verdict: matches, high confidence
    5. One step earlier: ncurses/README has no PORTABILITY section; it points to the hacker’s guide, whose last chapter is the porting hints
  • tic(1), captoinfo(1), infocmp(1), toe(1), clear(1) tset(1), reset(1), and tput(1) (docs/contributor/install/installation-procedure.md)
    1. Claim, as printed: tic(1), captoinfo(1), infocmp(1), toe(1), clear(1) tset(1), reset(1), and tput(1)
    2. Opened: progs/Makefile.in:202-217; INSTALL step 1
    3. It says: “actual_captoinfo … actual_infotocap”
    4. Verdict: matches, high confidence
    5. One step earlier: Step 1 of the same file lists tabs among the installed programs, which is what showed step 3’s list was short
  • If the option is not given, this defaults to “long”. (docs/contributor/install/configure-options/with-m-z.md)
    1. Claim, as printed: If the option is not given, this defaults to “long”.
    2. Opened: aclocal.m4:92,109 (CF_ABI_DEFAULTS); configure.in:1086-1089; a configure run
    3. It says: “cf_dft_tparm_arg=long … (x[[6789]]) cf_dft_tparm_arg=intptr_t”
    4. Verdict: matches, high confidence
    5. One step earlier: The configure run reported intptr_t; reading only configure.in, the default is a variable and looks like it could still be long
  • –with-weak-symbols (docs/contributor/install/configure-options/with-m-z.md)
    1. Claim, as printed: –with-weak-symbols
    2. Opened: configure.in:1394-1398
    3. It says: “AC_ARG_ENABLE(weak-symbols, [ –enable-weak-symbols enable weak-symbols for pthreads]”
    4. Verdict: matches, high confidence
    5. One step earlier: INSTALL’s own option list spells it –enable-weak-symbols, so only this cross-reference was wrong
  • FreeBSD, NetBSD, OpenBSD, Cygwin, MinGW (docs/contributor/install/system-integrators.md)
    1. Claim, as printed: FreeBSD, NetBSD, OpenBSD, Cygwin, MinGW
    2. Opened: aclocal.m4:979
    3. It says: “(openbsd*|freebsd*|mirbsd*|linux*|cygwin*|msys*|kbsd-gnu|mingw*)”
    4. Verdict: matches, high confidence
    5. One step earlier: NetBSD is in INSTALL’s list but not in the case pattern, and MirBSD, MSYS and kbsd-gnu are in the pattern but not the list
  • listed in the misc/terminfo file (docs/contributor/install/system-integrators.md)
    1. Claim, as printed: listed in the misc/terminfo file
    2. Opened: misc/terminfo.src:1-7
    3. It says: “This version of terminfo.src is distributed with ncurses and is maintained by Thomas E. Dickey”
    4. Verdict: matches, high confidence
    5. One step earlier: misc/ has no file named terminfo, only terminfo.src, so the path INSTALL gave led nowhere
  • $HOME/terminfo (twice) (docs/contributor/install/bsd-conversion.md)
    1. Claim, as printed: $HOME/terminfo (twice)
    2. Opened: ncurses/tinfo/home_terminfo.c:35; INSTALL:2484
    3. It says: “home_terminfo.c – return the $HOME/.terminfo string, expanded”
    4. Verdict: matches, high confidence
    5. One step earlier: The same section says $HOME/.terminfo two paragraphs earlier, so this was a slip rather than a second directory
  • To subscribe send mail to bug-ncurses-request@gnu.org with body that reads: subscribe ncurses (docs/contributor/install/bug-reports.md)
    1. Claim, as printed: To subscribe send mail to bug-ncurses-request@gnu.org with body that reads: subscribe ncurses
    2. Opened: https://lists.gnu.org/mailman/listinfo/bug-ncurses
    3. It says: “Bug-ncurses Info Page (the Mailman listinfo page)”
    4. Verdict: matches, medium confidence
    5. One step earlier: Mailman reads the word after subscribe as a password, so the old body would not subscribe the address it names
  • assuming environment variable MINGW_ROOT holds the root directory name of your MinGW build (docs/contributor/install/mingw.md)
    1. Claim, as printed: assuming environment variable MINGW_ROOT holds the root directory name of your MinGW build
    2. Opened: README.MinGW:101-136
    3. It says: “./configure –prefix=/mingw64 …”
    4. Verdict: matches, medium confidence
    5. One step earlier: The command that follows never uses MINGW_ROOT; it names /mingw64 directly
  • I provide no functions to switch on/off curses tracing options. (docs/contributor/ada/limitations.md)
    1. Claim, as printed: I provide no functions to switch on/off curses tracing options.
    2. Opened: Ada95/gen/terminal_interface-curses-trace.ads.m4:45,115-119
    3. It says: “procedure Trace_On (x : Trace_Attribute_Set); … procedure Trace_Put (str : String);”
    4. Verdict: matches, high confidence
    5. One step earlier: The package list in the same document leaves Trace out, so nothing in the original hinted it had been added
  • Just make the boolean member bold of the record true. (docs/contributor/ada/index.md)
    1. Claim, as printed: Just make the boolean member bold of the record true.
    2. Opened: Ada95/gen/terminal_interface-curses.ads.m4:420-427
    3. It says: “type Character_Attribute_Set is record … Bold_Character : Boolean;”
    4. Verdict: matches, high confidence
    5. One step earlier: The original wrote the name in bold type, so it read like an identifier; the record has no member called Bold
  • Support for wide characters is currently not in the binding, as it is not really in ncurses at this point in time. (docs/contributor/ada/index.md)
    1. Claim, as printed: Support for wide characters is currently not in the binding, as it is not really in ncurses at this point in time.
    2. Opened: Ada95/gen/terminal_interface-curses.ads.m4; INSTALL –enable-widec
    3. It says: “INSTALL: Currently (2023/10/21) it does not use the extended features from the wide-character code”
    4. Verdict: matches, high confidence
    5. One step earlier: The binding part is still true, but ncurses has had wide-character support since 5.x and builds it by default since 6.5, so the reason given no longer holds
  • The (n)curses documentation says, that the String arrays to be passed to an TYPE_ENUM fieldtype must not be automatic variables. (docs/contributor/ada/implementation-details.md)
    1. Claim, as printed: The (n)curses documentation says, that the String arrays to be passed to an TYPE_ENUM fieldtype must not be automatic variables.
    2. Opened: man/form_field_validation.3x:119-120
    3. It says: “The library copies the string list, so you may use a list that lives in automatic variables on the stack.”
    4. Verdict: matches, high confidence
    5. One step earlier: Taken alone, the binding text reads as a workaround for ncurses; the man page shows ncurses now does the same copy
  • The most important bug-fixes/improvements (a link to “Interesting bug-fixes”, a section with no content) (docs/contributor/announce/release-notes.md)
    1. Claim, as printed: The most important bug-fixes/improvements (a link to “Interesting bug-fixes”, a section with no content)
    2. Opened: announce.html.in:171,926-930
    3. It says: “

      …Interesting bug-fixes

      followed directly by

      …Configuration changes

      ”
    4. Verdict: matches, high confidence
    5. One step earlier: The table of contents lists the section too, so the gap only shows on reaching it
  • https://invisible-island.net/ncurses/man/tput.1m.html (docs/contributor/announce/program-improvements.md)
    1. Claim, as printed: https://invisible-island.net/ncurses/man/tput.1m.html
    2. Opened: announce.html.in:544; man/tput.1; a request to tput.1m.html
    3. It says: “The Overview links tput.1.html; the tree has man/tput.1 and no tput.1m; the 1m URL returned 404”
    4. Verdict: matches, high confidence
    5. One step earlier: The other utilities in that list are in section 1m, which makes 1m look right for tput too
  • Patches to the current release are made available at https://invisible-island.net/archives/ncurses/6.5/ (docs/contributor/announce/development.md)
    1. Claim, as printed: Patches to the current release are made available at https://invisible-island.net/archives/ncurses/6.5/
    2. Opened: announce.html.in:1748-1751; dist.mk:39-44; https://invisible-island.net/archives/ncurses/6.6/
    3. It says: “NCURSES_MAJOR = 6, NCURSES_MINOR = 6; the 6.6 directory returned 200”
    4. Verdict: matches, high confidence
    5. One step earlier: The same page announces 6.6 as this release, so 6.5 is the previous one
  • To join the ncurses mailing list, please write email to bug-ncurses-request@gnu.org containing the line: subscribe @<host.domain> (docs/contributor/announce/development.md)
    1. Claim, as printed: To join the ncurses mailing list, please write email to bug-ncurses-request@gnu.org containing the line: subscribe @<host.domain>
    2. Opened: https://lists.gnu.org/mailman/listinfo/bug-ncurses
    3. It says: “Bug-ncurses Info Page (the Mailman listinfo page)”
    4. Verdict: matches, medium confidence
    5. One step earlier: Mailman reads the word after subscribe as a password, so that line would not subscribe the address it names
  • modify MKlib_gen.c to allow for Solaris’s definition of NULL as 0L (docs/contributor/announce/portability.md)
    1. Claim, as printed: modify MKlib_gen.c to allow for Solaris’s definition of NULL as 0L
    2. Opened: ncurses/base/MKlib_gen.sh:191-194; NEWS:804
    3. It says: “s/((()0[L]([ ]!=[ ]*(const void)/\1NULL\2/g”
    4. Verdict: matches, high confidence
    5. One step earlier: NEWS has the same .c name, so the slip came from there; no MKlib_gen.c exists, and the 0[L]* patterns are in the .sh script
  • to revert to the original mode, type this: printf ‘\033[0;37;40m\n’ (docs/client-library/getting-started/introduction.md)
    1. Claim, as printed: to revert to the original mode, type this: printf ‘\033[0;37;40m\n’
    2. Opened: misc/terminfo.src:882 and 1291 (sgr0 for VT-style entries)
    3. It says: “sgr0=\E[m … sgr0=\E[m\E(B”
    4. Verdict: matches, high confidence
    5. One step earlier: Reading only the guide, 37;40 looks like the terminal’s normal colors; the terminfo sgr0 strings show the reset is SGR 0, and 37;40 forces white on black
  • control characters like suspend (CTRL-Z), interrupt and quit (CTRL-C) (docs/client-library/getting-started/initialization.md)
    1. Claim, as printed: control characters like suspend (CTRL-Z), interrupt and quit (CTRL-C)
    2. Opened: /usr/include/x86_64-linux-gnu/sys/ttydefaults.h:71
    3. It says: “#define CQUIT 034 /* FS, ^\ */”
    4. Verdict: matches, high confidence
    5. One step earlier: Reading only the guide, CTRL-C could be meant for interrupt alone; the system default shows quit is a different key
  • Fifth is the color index. It is the index given during init_pair() (docs/client-library/attributes/chgat.md)
    1. Claim, as printed: Fifth is the color index. It is the index given during init_pair()
    2. Opened: docs/man/curs_attr/03-description.3x:128; man/curs_color.3x (DESCRIPTION)
    3. It says: “the color pair argument is a color pair index; init_pair initializes a color pair identifier … from a pair of color indices”
    4. Verdict: matches, high confidence
    5. One step earlier: Reading only the HOWTO, index and pair look interchangeable; curs_color uses color index for a single color and pair for the combination
  • Do this before the next wgetch(); otherwise another mouse event might come in and make the first one inaccessible (docs/client-library/mouse/mouse-events.md)
    1. Claim, as printed: Do this before the next wgetch(); otherwise another mouse event might come in and make the first one inaccessible
    2. Opened: ncurses/curses.priv.h:185; docs/man/curs_mouse (DESCRIPTION)
    3. It says: “#define EV_MAX 8 /* size of mouse circular event queue */; A queue distinct from that used for keyboard events accumulates the details of mouse events.”
    4. Verdict: matches, high confidence
    5. One step earlier: Without the source I would have kept the warning as a cautious rule; curses.priv.h shows a queue that holds more than one event
  • An example showing all the characters, with 31 SHOW() lines and no ACS_URCORNER (docs/client-library/output/line-drawing-characters.md)
    1. Claim, as printed: An example showing all the characters, with 31 SHOW() lines and no ACS_URCORNER
    2. Opened: include/curses.h.in (ACS_ defines); doc/html/NCURSES-Programming-HOWTO.html:3332-3363
    3. It says: “#define ACS_URCORNER NCURSES_ACS(‘k’) /* upper right corner */; the HOWTO has no ACS_URCORNER”
    4. Verdict: matches, high confidence
    5. One step earlier: Counting the list alone, 31 looked complete; curses.h names 32 distinct characters
  • Note that new_field() does not copy the pointer array into private storage (docs/client-library/forms/fields.md)
    1. Claim, as printed: Note that new_field() does not copy the pointer array into private storage
    2. Opened: doc/html/ncurses-intro.html:1965; form/frm_def.c:157-167
    3. It says: “new_field() does not copy the …; Connect_Fields(FORM *form, FIELD **fields) … form->field = fields;”
    4. Verdict: matches, high confidence
    5. One step earlier: The pointer array is the argument of new_form(), not new_field(); frm_def.c shows new_form() keeps the caller’s array
  • By default, all options are on. (docs/client-library/forms/field-options.md)
    1. Claim, as printed: By default, all options are on.
    2. Opened: form/form.priv.h:151-168; form/fld_def.c:57
    3. It says: “STD_FIELD_OPTS (O_VISIBLE … O_STATIC); ALL_FIELD_OPTS adds O_DYNAMIC_JUSTIFY, O_NO_LEFT_STRIP, O_EDGE_INSERT_STAY, O_INPUT_LIMIT; the default field uses STD_FIELD_OPTS”
    4. Verdict: matches, high confidence
    5. One step earlier: True when written; the extension options came later and the default field still uses only the standard set
  • *Each field structure contains one character pointer slot … The (char ) type is retained for System V compatibility (docs/client-library/forms/field-options.md)
    1. Claim, as printed: Each field structure contains one character pointer slot … The (char *) type is retained for System V compatibility
    2. Opened: form/form.h:376 and 389; doc/html/ncurses-intro.html:2222
    3. It says: “extern FORM_EXPORT(int) set_field_userptr (FIELD *, void *); extern FORM_EXPORT(void *) field_userptr (const FIELD *);”
    4. Verdict: matches, high confidence
    5. One step earlier: The guide predates the change; form.h now declares void *
  • The growth limit can be changed whether or not the O_STATIC bit is on, but has no effect until it is. (docs/client-library/forms/field-options.md)
    1. Claim, as printed: The growth limit can be changed whether or not the O_STATIC bit is on, but has no effect until it is.
    2. Opened: form/fld_max.c:72; doc/html/ncurses-intro.html:2258
    3. It says: “if (!((unsigned)field->opts & O_STATIC)) { … may grow”
    4. Verdict: matches, high confidence
    5. One step earlier: Read literally, “until it is” means until O_STATIC is on, the opposite of what fld_max.c does
  • A REQ_MOVE_RIGHT from A will go to B (docs/client-library/forms/form-driver.md)
    1. Claim, as printed: A REQ_MOVE_RIGHT from A will go to B
    2. Opened: form/form.h:259; doc/html/ncurses-intro.html:2745
    3. It says: “#define REQ_RIGHT_FIELD (KEY_MAX + 14); form.h has no REQ_MOVE_RIGHT”
    4. Verdict: matches, high confidence
    5. One step earlier: A reader copying the name would get a compile error
  • the position of a form for purposes of these requests is its upper-left corner (docs/client-library/forms/form-driver.md)
    1. Claim, as printed: the position of a form for purposes of these requests is its upper-left corner
    2. Opened: doc/html/ncurses-intro.html:2740
    3. It says: “Note, however, that the position of a form for purposes of these requests is its upper-left corner.”
    4. Verdict: matches, high confidence
    5. One step earlier: The example that follows is about fields A, B and C, so form is a slip for field
  • REQ_DEL_PREV: Delete previous word at cursor. (docs/client-library/forms/form-driver.md)
    1. Claim, as printed: REQ_DEL_PREV: Delete previous word at cursor.
    2. Opened: docs/man/form_driver (requests table); doc/html/ncurses-intro.html:2951
    3. It says: “REQ_DEL_PREV Delete character before the cursor.”
    4. Verdict: matches, high confidence
    5. One step earlier: The guide itself later says REQ_DEL_PREV deletes the previous character; the man page settles it
  • When the current page is changed by set_current_field() call; When the current field is changed by a set_form_page() call (docs/client-library/forms/hooks-and-form-options.md)
    1. Claim, as printed: When the current page is changed by set_current_field() call; When the current field is changed by a set_form_page() call
    2. Opened: doc/html/ncurses-intro.html:3120-3121; docs/man/form_page
    3. It says: “When the current page is changed by set_current_field() call”
    4. Verdict: matches, high confidence
    5. One step earlier: The function names alone show the two were swapped
  • ncurses is intended to be base-level conformant with the X/Open Curses standard … almost all the features not directly concerned with wide characters and internationalization (docs/client-library/hints/compatibility.md)
    1. Claim, as printed: ncurses is intended to be base-level conformant with the X/Open Curses standard … almost all the features not directly concerned with wide characters and internationalization
    2. Opened: man/ncurses.3x (DESCRIPTION)
    3. It says: “ncurses implements the interface described by X/Open Curses Issue 7.”
    4. Verdict: matches, high confidence
    5. One step earlier: True of the 1990s library; ncursesw has since added the wide-character functions
  • Slide the tiles with the arrow keys until they’re in order (docs/client-library/more/shuffle.md)
    1. Claim, as printed: Slide the tiles with the arrow keys until they’re in order
    2. Opened: git show bec4c6d:docs/client-library/more/shuffle.md (line 5, the sentence as first written); docs/client-library/more/shuffle.md:95-110 init_board(); Johnson and Story, American Journal of Mathematics 2 (1879)
    3. It says: “init_board() fills temp_board with rand() % (n * n) values and no parity check; exactly half of the permutations of a sliding puzzle are solvable”
    4. Verdict: matches, high confidence
    5. One step earlier: Without reading init_board() I would have assumed the deal was made by legal moves, which is always solvable
  • parse an entire terminfo file into a doubly-linked circular list of entry structures (docs/contributor/hackguide/terminfo-compiler.md)
    1. Claim, as printed: parse an entire terminfo file into a doubly-linked circular list of entry structures
    2. Opened: ncurses/tinfo/comp_parse.c:70-75
    3. It says: “newp->last = _nc_tail; newp->next = NULL;”
    4. Verdict: matches, high confidence
    5. One step earlier: The list ends in NULL at both ends; it is not circular
  • The tput and clear utilities just do an entry load followed by a tputs() of a selected capability. (docs/contributor/hackguide/other-utilities.md)
    1. Claim, as printed: The tput and clear utilities just do an entry load followed by a tputs() of a selected capability.
    2. Opened: progs/clear_cmd.c:56-61; progs/tput.c:118-119,440-441
    3. It says: “tputs(clear_screen, …); char *E3 = tigetstr(UserCap(E3)); my_init[] = “init”; my_reset[] = “reset””
    4. Verdict: matches, high confidence
    5. One step earlier: Both programs grew beyond a single tputs() after the guide was written
  • and also interprets the contents of the $TERM environment variable (docs/contributor/install/installation-procedure.md)
    1. Claim, as printed: and also interprets the contents of the $TERM environment variable
    2. Opened: INSTALL:177-181; ncurses/tinfo/read_termcap.c:804,994,1053
    3. It says: “will also interpret the contents of the $TERM environment variable; getenv(“TERMCAP”)”
    4. Verdict: matches, high confidence
    5. One step earlier: The termcap reader reads TERMCAP (and TERMPATH), not TERM, for termcap data
  • $(prefix)/man: the manual pages (docs/contributor/install/installation-procedure.md)
    1. Claim, as printed: $(prefix)/man: the manual pages
    2. Opened: INSTALL:93; configure:176,186,691
    3. It says: “Under $(prefix)/man: the manual pages; datarootdir=‘${prefix}/share’; mandir=‘${datarootdir}/man’”
    4. Verdict: matches, high confidence
    5. One step earlier: Autoconf has defaulted mandir to DATAROOTDIR/man since 2.60, and this configure does too
  • the library searches /etc/termcap before the terminfo database (docs/contributor/install/installation-procedure.md)
    1. Claim, as printed: the library searches /etc/termcap before the terminfo database
    2. Opened: INSTALL:177-180; include/term_entry.h:95-108; ncurses/tinfo/read_entry.c:946-958
    3. It says: “will search /etc/termcap before the terminfo database; dbdEnvOnce … dbdCfgOnce (terminfo) come before dbdEnvOnce2, dbdEnvList2, dbdCfgList2 (termcap)”
    4. Verdict: matches, high confidence
    5. One step earlier: PR 57 fixed $TERM in the same sentence but kept ‘before’; the database iterator tries every terminfo location first
  • su root (become root), then make install (docs/client-library/getting-started/get-ncurses.md)
    1. Claim, as printed: su root (become root), then make install
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:851-852; INSTALL:212-214
    3. It says: “HOWTO: su root # become root / make install # install it. INSTALL: Once you’ve tested, you can type ‘make install’ to install libraries, the programs, the terminfo database and the manual pages.”
    4. Verdict: matches, high confidence
    5. One step earlier: Typed one at a time on a system with a root password, the two steps work, so the HOWTO alone looks right; run as a sequence, su opens a root shell and waits, and make install then runs as the reader, and su root fails at once on Ubuntu (no root password) and macOS (root disabled)
  • A menu item is visible when it is in the portion of a posted menu that is mapped onto the screen (if the menu is scrollable, in particular, this portion will be smaller than the whole menu). (docs/man/mitem_visible/03-description.3x)
    1. Claim, as printed: A menu item is visible when it is in the portion of a posted menu that is mapped onto the screen (if the menu is scrollable, in particular, this portion will be smaller than the whole menu).
    2. Opened: menu/m_item_vis.c:46-47 and 59-62
    3. It says: “Description: A item is visible if it currently appears in the subwindow of a posted menu. / (menu->toprow + menu->arows) > (item->y) && (item->y >= menu->toprow)”
    4. Verdict: matches, high confidence
    5. One step earlier: Read alone, “mapped onto the screen” sounds right; the code compares only the item’s row with the menu’s top row and displayed rows, so an item counts as visible before the screen is refreshed
  • These routines emulate the System V menu library. They were not supported on Version 7 or BSD versions. (docs/man/mitem_visible/05-portability.3x)
    1. Claim, as printed: These routines emulate the System V menu library. They were not supported on Version 7 or BSD versions.
    2. Opened: man/mitem_visible.3x SYNOPSIS; menu/menu.h:269
    3. It says: “bool item_visible(const ITEM *item); (the page documents one function)”
    4. Verdict: matches, high confidence
    5. One step earlier: The plural is boilerplate shared with the other menu pages; this page has a single routine
  • If no files are specified, captoinfo interprets the content of the environment variable TERMCAP as a file name, and extracts only the entry for the terminal type named in the environment variable TERM from it. (docs/man/captoinfo/03-description.1m)
    1. Claim, as printed: If no files are specified, captoinfo interprets the content of the environment variable TERMCAP as a file name, and extracts only the entry for the terminal type named in the environment variable TERM from it.
    2. Opened: progs/tic.c:923-942
    3. It says: “if (access(termcap, F_OK) == 0) { /* file exists */ source_file = termcap; } else { … fprintf(tmp_fp, “%s\n”, termcap);”
    4. Verdict: matches, high confidence
    5. One step earlier: TERMCAP usually names a file, so the page reads as complete until the other branch is read
  • menu_request_by_name returns E_NO_MATCH on error. It does not set errno. (docs/man/menu_requestname/04-return-value.3x)
    1. Claim, as printed: menu_request_by_name returns E_NO_MATCH on error. It does not set errno.
    2. Opened: menu/m_req_name.c:128; menu/mf_common.h:73-78
    3. It says: “RETURN(E_NO_MATCH); … #define RETURN(code) return( SET_ERROR(code) ) … #define SET_ERROR(code) (errno=(code))”
    4. Verdict: matches, high confidence
    5. One step earlier: The RETURN macro hides the errno assignment; the return statement alone reads as if errno is untouched
  • form_request_by_name returns E_NO_MATCH on error. It does not set errno. (docs/man/form_requestname/04-return-value.3x)
    1. Claim, as printed: form_request_by_name returns E_NO_MATCH on error. It does not set errno.
    2. Opened: form/frm_req_name.c:176; menu/mf_common.h:73-78
    3. It says: “RETURN(E_NO_MATCH); … #define RETURN(code) return( SET_ERROR(code) )”
    4. Verdict: matches, high confidence
    5. One step earlier: The RETURN macro hides the errno assignment; the return statement alone reads as if errno is untouched
  • the name field cannot exceed 128 bytes. (docs/man/term/04-portability.5)
    1. Claim, as printed: the name field cannot exceed 128 bytes.
    2. Opened: ncurses/tinfo/write_entry.c:771,812-822; ncurses/tinfo/read_entry.c:362,381-391; include/tic.h:104; site/formal/term-name-field.als (Alloy 6.2.0: NameNeverExceeds128 has a counterexample, LimitIs511 holds); site/formal/term-name-field-roundtrip.sh with tic and infocmp 6.6.20261107 (2026-11-18)
    3. It says: “#define MAX_NAME_SIZE 512 /* maximum legal name field size (XSI:127) */ … want = Min(MAX_NAME_SIZE, (unsigned) name_size);”
    4. Verdict: matches, high confidence
    5. One step earlier: The page lists it beside two entry-size limits that are right, and 128 is close to the XSI minimum of 127 that the code’s comment cites
  • 0: Boolean or string capability present; 1: Boolean or numeric capability absent (docs/man/tput/05-exit-status.1)
    1. Claim, as printed: 0: Boolean or string capability present; 1: Boolean or numeric capability absent
    2. Opened: progs/tput.c:132-148,213-215,387-389; site/formal/utilities-check.sh with tput 6.6.20261107 (2026-11-18), output in site/formal/evidence/tput-clear-tabs/utilities-check.txt: tput -T dumb lines exits 0 and writes -1, tput -T dumb bold exits 1
    3. It says: “case NUMBER: result = 0; /* always zero / … case STRING: result = value; / 0=normal, 1=missing */”
    4. Verdict: matches, high confidence
    5. One step earlier: DESCRIPTION already said an absent numeric capability is written as -1; the table looks like it swapped “string” and “numeric” in status 1
  • With -S: 4: some operands not interpreted (docs/man/tput/05-exit-status.1)
    1. Claim, as printed: With -S: 4: some operands not interpreted
    2. Opened: progs/tput.c:240,505-512; progs/progs.priv.h:96-99; site/formal/utilities-check.sh with tput 6.6.20261107 (2026-11-18), output in site/formal/evidence/tput-clear-tabs/utilities-check.txt: -S with hz gives 5, hz hz gives 6, hz hz hz gives 7, nosuchcap then bold gives 4 and writes nothing for bold
    3. It says: “if (code != 0) { if (result == 0) result = ErrSystem(0); /* will return value >4 */ ++result; }”
    4. Verdict: matches, high confidence
    5. One step earlier: The code comment itself says the value is above 4; the page kept a single status 4 for any failure
  • Only one cap-code is allowed per line. (docs/man/tput/04-options.1)
    1. Claim, as printed: Only one cap-code is allowed per line.
    2. Opened: progs/tput.c:481-516; site/formal/utilities-check.sh with tput 6.6.20261107 (2026-11-18), output in site/formal/evidence/tput-clear-tabs/utilities-check.txt: -S with “hz hz hz” on one line gives 7, and “cup 1 2 bold” writes both sequences
    3. It says: “/* split the buffer into tokens */ … while (argnum > 0) { … code = tput_cmd(fd, &tty_settings, argnum, argnow, &used); … argnum -= used; argnow += used; }”
    4. Verdict: matches, high confidence
    5. One step earlier: NEWS for patch 20210925 says modify tput to allow multiple commands per line; the sentence in OPTIONS predates that change
  • define_key returns ERR if definition is NULL and key-code is nonpositive, or insufficient memory is available to bind definition to key-code. Otherwise, define_key returns OK. (docs/man/define_key/04-return-value.3x)
    1. Claim, as printed: define_key returns ERR if definition is NULL and key-code is nonpositive, or insufficient memory is available to bind definition to key-code. Otherwise, define_key returns OK.
    2. Opened: ncurses/base/define_key.c:42-77; ncurses/base/key_defined.c:39-61; ncurses/tinfo/add_tries.c:57-58; site/formal/define-key-check.sh with ncurses 6.6.20261107 (2026-11-18), output in site/formal/evidence/define-key/define-key-check.txt: cases 5, 6, 9, 10 and 11 return ERR, and define_key before newterm returns ERR
    3. It says: “int code = ERR; … } else if (CallHasKey(keycode)) { while (_nc_remove_key(&(SP_PARM->_keytry), ukey)) code = OK; } … if (NCURSES_SP_NAME(key_defined)(NCURSES_SP_ARGx str) == 0) { … } else { code = ERR; }”
    4. Verdict: matches, high confidence
    5. One step earlier: The page lists only the argument check and memory; the code starts from ERR and sets OK only when a binding is removed or added, so every case that changes nothing fails

Errors this edition introduced, then fixed

Section titled “Errors this edition introduced, then fixed”
Where Was Now Source Verdict Change
docs/client-library/forms/fields.md Merging the guides split the field_pad() comment out of its code block; the stray fence turned the next paragraph and the Form Attributes heading into code Comment back on the field_pad() line, fence removed This edition matches, high #3, f52aae7
docs/client-library/forms/fields.md Backspace at the start of a field moved to the previous field but left the cursor at its start REQ_END_LINE after REQ_DEL_PREV when the field changes This edition matches, high #6, 3a0e739
[man/resizeterm.3x](/ncurses/man/3x/resizeterm/#SEE ALSO) The site’s man pages showed literal ** around page references and after italic words, e.g. **curs_getch(3X),** **curs_variables(3X),** **wresize(3X)** Page references render as bold links; check-rendered.mjs finds no literal markup on any page This edition’s site build (sync.mjs through pandoc 3.1.3) matches, high #7, c9e0bd1
man/infocmp.1m The site showed the page as troff source because pandoc stopped at the table format Lf(BI) The page renders; the table font is rewritten to fBI before pandoc reads it This edition’s site build (sync.mjs through pandoc 3.1.3) matches, high #7, 0809cab
man/curs_terminfo.3x The site showed the page as troff source because pandoc stopped at a table cell continued with a trailing backslash, and the backticks in troff’s quote strings closed the fallback’s code fence early The page renders; continued lines are joined before pandoc reads them, and the fallback fence is longer than any run of backticks in the page This edition’s site build (sync.mjs through pandoc 3.1.3) matches, high #7, 0809cab
site/src/components/run-example.ts In the runnable examples the cursor blinked white and black inside a blue form field It blinks white and the cell’s own blue; the terminal uses xterm.js’s WebGL renderer Reader report on the fields page matches, high #7, 1019410
site/src/content/docs/index.mdx Thomas E. Dickey has maintained ncurses since 1.9.9e Has recorded every change since 1.9.9e, and maintained ncurses from 1.9.9g through 4.1 and again from 5.0 This edition’s front page matches, high #44, 3c61bb7
site/src/content/docs/credits.md Thomas E. Dickey, maintainer since ncurses 1.9.9e Maintainer from ncurses 1.9.9g through 4.1 and again from 5.0; NEWS records every change since 1.9.9e in his hand This edition’s Credits page matches, high #44, 3c61bb7
docs/client-library/output/forms-characters.md Other border functions draws lines with them (the forms-drawing characters) Other border functions draws with the ASCII characters |, - and +; the window border program gets forms-drawing characters by passing 0 to box() This edition (commit eea2d03, Output guide) matches, high #54, 74a7f5a
docs/client-library/input/characters-and-strings.md While getnstr() is reading the line, F1 is ignored, like the other function keys Without keypad() F1 arrives as its escape sequence and is stored; in keypad mode getnstr() beeps at function keys, except that the down arrow ends the line and the left arrow erases This edition (commit d341cd3, Input guide) matches, high #54, 74a7f5a
docs/client-library/hints/resizing.md pads stdscr with blanks: the glossary link was on the verb The link moves to the noun later in the paragraph (can’t resize pads) This edition (glossary links, PR 53, commit df81beb) matches, high #56, e117d45
docs/contributor/ada/implementation-details.md ncurses also uses the parameter’s absolute value as the number of lines to take, and the binding’s Rip_Off_Lines routine relies on this. The public ripoffline() passes on only the sign; the internal _nc_ripoffline() uses the absolute value, and Rip_Off_Lines calls it directly This edition (Ada95 guide, commit e3fa730) matches, high #57, 769b14b
docs/glossary/attribute.md The color pair is stored with the attributes, and the wide-character functions use the attr_t type and names such as attr_on(). The X/Open functions, such as attr_on() and attr_set(), and the wide-character functions use attr_t; attr_set() takes the pair separately This edition (glossary, PR 53, commit df81beb) matches, high #59, 8e5de3e
docs/client-library/input/characters-and-strings.md ncurses handles that case itself: getch() reads a whole line with wgetnstr(), echoing it …, so the program gets nothing until the user presses Enter The original warning is back: one or the other has to be changed before calling getch(); otherwise, the program’s output will be unpredictable This edition (Input guide review, PR 54, commit 74a7f5a) matches, medium #63, 3cc6b89
docs/contributor/hackguide/terminfo-compiler.md This, for example, is where the XENIX forms-drawing capabilities … get translated to an acsc string. The reverse translation, from acsc to the AIX box1 capability, happens on output, in dump_entry.c. The original example is back: parse_entry.c is where the AIX box1 capability gets translated to an acsc string This edition (Hacker’s guide review, PR 57, commit 769b14b) matches, high #63, 3cc6b89
docs/glossary/wide-character-library.md most add w or _w to the 8-bit name, so waddstr() becomes waddwstr() and waddch() becomes wadd_wch() Most are named by adding _w to the 8-bit name, so waddch() becomes wadd_wch(); the addwstr() and inwstr() families add only w, as in waddwstr() This edition (glossary review, PR 59, commit 8e5de3e) matches, high #63, 3cc6b89
docs/client-library/more/widget-libraries.md Curses::Forms, linked to the CursesForms distribution, and Curses::Forms(3) Curses::Form, linked to its page in the Curses distribution, and Curses::Form(3), as the original had them This edition matches, medium #63, bc49d13
docs/client-library/menus/attributes.md The DOS games collection shown as a plain-text URL marked as a dead link Linked to the Internet Archive’s 2001 copy of the collection This edition matches, medium #63, 0b3b1fc
site/src/content/docs/index.mdx He has revised every manual page and wrote several of them. He has revised every manual page, which Eric S. Raymond first wrote. This edition matches, high #63, 567f078
site/src/components/Footer.astro When a runnable example started, a vertical scroll bar flashed beside the terminal No scroll bar on the example terminal Reader report on the runnable examples matches, high #69, a14e7f4
site/code-links.mjs In code listings, curses types and variables such as WINDOW, SCREEN, bool and stdscr were plain text with no hint; only function calls linked They link to their man page with the same hint as functions; FIELD, FORM, MENU, ITEM, PANEL and FIELDTYPE link to the page of the function that creates them Reader report on curs_initscr(3X) matches, high #69, 4d540ca
site/scripts/sync.mjs A function’s link opened the top of its man page, however far down the function is described It opens the heading that documents the name, or the section that first describes it or its w-family member (getch goes to curs_getch(3X)’s Reading Characters) Reader report on the code hints matches, high #69, 4d540ca
site/src/components/boot-screen.js The December solstice from Meeus’s mean formula, up to 14 minutes from the published instant The December solstice for 2025–2036 from the US Naval Observatory’s published instants; Lunar New Year cites the Hong Kong Observatory tables, and a test checks Hanukkah against Hebcal and Lunar New Year against the Observatory Check of the festival dates against published calendars matches, high #71, eb5852c
site/src/components/run-example.ts The exit note after a program’s last screen began with a new line, so a program that ended with the cursor on the bottom row lost its top two lines The note goes on the bottom row when that row is empty This edition matches, high #72, 0769fa6
site/examples/pty.js A program that read keys with nodelay() or timeout(0) froze until a key came: Emscripten answers a zero-timeout poll() with __syscall_poll_nonblocking, which called stdin always ready, so getch() blocked in read() pty.js answers __syscall_poll_nonblocking with the same readiness check as __syscall_poll This edition matches, high #73, ce5a458
site/README.md The pretend shell keeps them in /usr/games: ls leaves them out /usr/games is a folder like any other: ls /usr/games or cd /usr/games finds them, and they run by name from any folder because it’s on the PATH This edition matches, high #73, 3ad103c
site/examples/pty.js Ctrl-C reached every program as a key, even in cbreak() mode, where the initialization guide says the terminal driver generates the signal While a program’s terminal modes keep ISIG on (cbreak(), or no mode call), Ctrl-C sends SIGINT and interrupts the read; under raw() it stays a key. pty.js keeps the modes set with tcsetattr(), which Emscripten’s terminal ignored This edition matches, high #73, 89203d4
site/src/components/MigrationProgress.astro One agent session did the work, with the maintainer reviewing and merging each pull request; each pull request reviewed and merged by the maintainer Betsalel Williamson, who runs this edition, reviewed and merged each pull request; the ncurses maintainer, Thomas E. Dickey, hasn’t reviewed this work. Maintainer time is now Reviewer time This edition matches, high #76, 846bc7c
site/scripts/man-render.mjs The site’s manual pages dropped every quotation mark written with the pages’ quote strings (88 pages: see subsection Keypad Mode below), and term(5)’s carets (bel=G, cr=M) Quotation marks and carets render as the troff source writes them: see subsection “Keypad Mode” below; bel=^G, cr=^M. pandoc now reads each .ie/.if as groff takes it This edition matches, high #78, d91dc70
site/scripts/man-render.mjs The site printed TABLE in place of every table whose header spans columns or rows (captoinfo(1M), curs_attr(3X), term(5), and nine capability tables in terminfo(5)), dropped strings inside tables (curs_addch(3X) lost the cells for ACS_DEGREE, ACS_BULLET and ACS_PI), and infocmp(1M) showed %p1% %+%c The adapter merges a spanned header into one row, shows a one-column table of preformatted lines as code, expands the strings each page defines, and drops point-size changes; no page shows TABLE, and infocmp(1M) shows %p1%’ ’%+%c This edition matches, high #81, 5f5ca9f
How each was checked
  • Merging the guides split the field_pad() comment out of its code block; the stray fence turned the next paragraph and the Form Attributes heading into code (docs/client-library/forms/fields.md)
    1. Claim, as printed: Merging the guides split the field_pad() comment out of its code block; the stray fence turned the next paragraph and the Form Attributes heading into code
    2. Opened: git show f52aae7^:docs/client-library/forms/fields.md (lines 126-130); the live fields page
    3. It says: “An unmatched ``` after the blank line, holding only /* returns present pad character */, then the prose rendered inside the code block”
    4. Verdict: matches, high confidence
    5. One step earlier: From the screenshot alone I would have blamed the site renderer; the shard itself had the stray fence
  • Backspace at the start of a field moved to the previous field but left the cursor at its start (docs/client-library/forms/fields.md)
    1. Claim, as printed: Backspace at the start of a field moved to the previous field but left the cursor at its start
    2. Opened: form/frm_driver.c:2586-2591 and 1563; the second runnable example in headless Chromium
    3. It says: “frm_driver.c: if ((fct == FE_Delete_Previous) && ((unsigned)form->opts & O_BS_OVERLOAD) && First_Position_In_Current_Field(form)) res = Inter_Field_Navigation(FN_Previous_Field, form); and in _nc_Set_Current_Field: form->currow = form->curcol = form->toprow = form->begincol = 0;”
    4. Verdict: matches, high confidence
    5. One step earlier: From the report alone it could have been the browser terminal; frm_driver.c shows the library puts the cursor at the start of the field it moves to
  • The site’s man pages showed literal ** around page references and after italic words, e.g. **curs_getch(3X),** **curs_variables(3X),** **wresize(3X)** ([man/resizeterm.3x](/ncurses/man/3x/resizeterm/#SEE ALSO))
    1. Claim, as printed: The site’s man pages showed literal ** around page references and after italic words, e.g. **curs_getch(3X),** **curs_variables(3X),** **wresize(3X)**
    2. Opened: man/resizeterm.3x:200-203; the output of pandoc 3.1.3 -f man -t gfm for that page; https://betsalel-williamson.github.io/ncurses/man/3x/resizeterm/
    3. It says: “troff: \fB\%curs_getch\fP(3X), \fB\%curs_variables\fP(3X), \fB\%wresize\fP(3X). pandoc: **curs_getch(3X),** **curs_variables(3X),** **wresize(3X)**”
    4. Verdict: matches, high confidence
    5. One step earlier: Reading only the sync script, I blamed its link rewriting; pandoc’s own output shows the ** before any link is added
  • The site showed the page as troff source because pandoc stopped at the table format Lf(BI) (man/infocmp.1m)
    1. Claim, as printed: The site showed the page as troff source because pandoc stopped at the table format Lf(BI)
    2. Opened: man/infocmp.1m:205-208; pandoc 3.1.3 -f man on that page
    3. It says: “Format line: Lf(BI) Lf(BI) L. pandoc: Error at (line 207, column 3): unexpected ‘(’”
    4. Verdict: matches, high confidence
    5. One step earlier: The page’s caution said only that pandoc couldn’t read it, which suggests a fault in the page; the format line is valid tbl syntax
  • The site showed the page as troff source because pandoc stopped at a table cell continued with a trailing backslash, and the backticks in troff’s quote strings closed the fallback’s code fence early (man/curs_terminfo.3x)
    1. Claim, as printed: The site showed the page as troff source because pandoc stopped at a table cell continued with a trailing backslash, and the backticks in troff’s quote strings closed the fallback’s code fence early
    2. Opened: man/curs_terminfo.3x:1083-1084; pandoc 3.1.3 -f man on that page
    3. It says: “resetterm set terminal modes to \*(``out of \fIcurses\fP\*('' \ (continued on the next line). pandoc: Error at (line 1083, column 65): unexpected end of input”
    4. Verdict: matches, high confidence
    5. One step earlier: Without the line number from pandoc’s error I would have looked for a macro pandoc doesn’t know, as in other pages
  • In the runnable examples the cursor blinked white and black inside a blue form field (site/src/components/run-example.ts)
    1. Claim, as printed: In the runnable examples the cursor blinked white and black inside a blue form field
    2. Opened: site/node_modules/@xterm/xterm/lib/xterm.js 5.5.0 (DOM renderer CSS); pixel samples of the cursor cell in headless Chromium; the cell’s attributes from term.buffer
    3. It says: “CSS: @keyframes blink_block { 0% { background-color: #ffffff } 50% { background-color: inherit } }. Cell under the cursor: bg=4 (blue). Samples before: rgb(255,255,255) rgb(30,30,30); after: rgb(255,255,255) rgb(52,101,164)”
    4. Verdict: matches, high confidence
    5. One step earlier: Setting the theme’s cursor colors alone changed nothing in the samples; only the cell’s attributes showed the cell itself was blue
  • Thomas E. Dickey has maintained ncurses since 1.9.9e (site/src/content/docs/index.mdx)
    1. Claim, as printed: Thomas E. Dickey has maintained ncurses since 1.9.9e
    2. Opened: README:159; README:168; NEWS:36; https://invisible-island.net/ncurses/#involved
    3. It says: “Thomas Dickey (maintainer for 1.9.9g through 4.1, resuming with FSF’s 5.0)”
    4. Verdict: matches, high confidence
    5. One step earlier: NEWS and AUTHORS both start at 1.9.9e, which reads as a match; only README names his years as maintainer, and Florian La Roche’s for 4.2
  • Thomas E. Dickey, maintainer since ncurses 1.9.9e (site/src/content/docs/credits.md)
    1. Claim, as printed: Thomas E. Dickey, maintainer since ncurses 1.9.9e
    2. Opened: README:159; NEWS:36; https://invisible-island.net/ncurses/#involved
    3. It says: “Thomas Dickey (maintainer for 1.9.9g through 4.1, resuming with FSF’s 5.0)”
    4. Verdict: matches, high confidence
    5. One step earlier: The same date as the front page; NEWS alone supports 1.9.9e as the start of his record but not of his maintainership
  • Other border functions draws lines with them (the forms-drawing characters) (docs/client-library/output/forms-characters.md)
    1. Claim, as printed: Other border functions draws lines with them (the forms-drawing characters)
    2. Opened: docs/client-library/windows/border-functions.md (init_win_params); docs/man/curs_border/03-description.3x:36
    3. It says: “p_win->border.ls = ‘|’; … p_win->border.tl = ‘+’; If any chtype argument is 0, then curses uses forms-drawing characters”
    4. Verdict: matches, high confidence
    5. One step earlier: Without reading init_win_params I would have assumed the second program used ACS_ names like the first
  • While getnstr() is reading the line, F1 is ignored, like the other function keys (docs/client-library/input/characters-and-strings.md)
    1. Claim, as printed: While getnstr() is reading the line, F1 is ignored, like the other function keys
    2. Opened: ncurses/base/lib_getstr.c:118-148
    3. It says: “ch == KEY_DOWN || ch == KEY_ENTER … break; ch == erasec || ch == KEY_LEFT || ch == KEY_BACKSPACE … else if (ch >= KEY_MIN …) beep”
    4. Verdict: matches, high confidence
    5. One step earlier: Before reading lib_getstr.c I would have said every function key beeps; the example also never calls keypad(), so F1 is never a key code there
  • pads stdscr with blanks: the glossary link was on the verb (docs/client-library/hints/resizing.md)
    1. Claim, as printed: pads stdscr with blanks: the glossary link was on the verb
    2. Opened: docs/client-library/hints/resizing.md:11; docs/man/resizeterm/03-description.3x:65
    3. It says: “and pads stdscr with blanks if the screen grew; It is not possible to automatically resize pads.”
    4. Verdict: matches, high confidence
    5. One step earlier: The glossary script matched the word, not its sense; I checked every pad link after finding this one
  • ncurses also uses the parameter’s absolute value as the number of lines to take, and the binding’s Rip_Off_Lines routine relies on this. (docs/contributor/ada/implementation-details.md)
    1. Claim, as printed: ncurses also uses the parameter’s absolute value as the number of lines to take, and the binding’s Rip_Off_Lines routine relies on this.
    2. Opened: ncurses/base/lib_set_term.c:650,774-782; Ada95/gen/terminal_interface-curses.adb.m4:2322-2332; doc/html/Ada95.html:297-301
    3. It says: “count = (rop->line < 0) ? -rop->line : rop->line; _nc_ripoffline (line < 0) ? -1 : 1; pragma Import (C, Ripoffline, “_nc_ripoffline”); ncurses has internally implemented it”
    4. Verdict: matches, high confidence
    5. One step earlier: The original said ncurses did this internally; the edition dropped ‘internally’, which made it read as true of ripoffline() itself
  • The color pair is stored with the attributes, and the wide-character functions use the attr_t type and names such as attr_on(). (docs/glossary/attribute.md)
    1. Claim, as printed: The color pair is stored with the attributes, and the wide-character functions use the attr_t type and names such as attr_on().
    2. Opened: include/curses.h.in:637-638; man/curs_attr.3x:56-57,87
    3. It says: “attr_on (attr_t, void ); / generated */ (no WIDEC mark); attr_set (attr_t, NCURSES_PAIRS_T, void *)”
    4. Verdict: matches, high confidence
    5. One step earlier: attr_on() is in the 8-bit library too; it is an X/Open function, not a wide-character one
  • ncurses handles that case itself: getch() reads a whole line with wgetnstr(), echoing it …, so the program gets nothing until the user presses Enter (docs/client-library/input/characters-and-strings.md)
    1. Claim, as printed: ncurses handles that case itself: getch() reads a whole line with wgetnstr(), echoing it …, so the program gets nothing until the user presses Enter
    2. Opened: man/curs_getch.3x:631-633; ncurses/base/lib_getch.c:453-477
    3. It says: “curs_getch(3X): Use of wgetch with echo(3X) and neither cbreak(3X) nor raw(3X) is not well-defined. lib_getch.c: Handle cooked mode. Grab a string from the screen”
    4. Verdict: matches, medium confidence
    5. One step earlier: The source does handle cooked mode, so the edition’s sentence describes the implementation; the manual keeps the case undefined, which is what a program can rely on
  • This, for example, is where the XENIX forms-drawing capabilities … get translated to an acsc string. The reverse translation, from acsc to the AIX box1 capability, happens on output, in dump_entry.c. (docs/contributor/hackguide/terminfo-compiler.md)
    1. Claim, as printed: This, for example, is where the XENIX forms-drawing capabilities … get translated to an acsc string. The reverse translation, from acsc to the AIX box1 capability, happens on output, in dump_entry.c.
    2. Opened: ncurses/tinfo/parse_entry.c:1077-1115; include/Caps:1031
    3. It says: “Translate AIX forms characters. if (PRESENT(box_chars_1)) … _nc_warning(“acsc string synthesized from AIX capabilities”); Caps: box_chars_1 box1 str bx”
    4. Verdict: matches, high confidence
    5. One step earlier: parse_entry.c has no literal box1; only the Caps table ties box1 to box_chars_1
  • most add w or _w to the 8-bit name, so waddstr() becomes waddwstr() and waddch() becomes wadd_wch() (docs/glossary/wide-character-library.md)
    1. Claim, as printed: most add w or _w to the 8-bit name, so waddstr() becomes waddwstr() and waddch() becomes wadd_wch()
    2. Opened: man/ncurses.3x:518-530; include/curses.wide
    3. It says: “where a non-wide function name contains ch or str, prefix it with _w … (Exceptions that add only w comprise addwstr, inwstr, and their variants.)”
    4. Verdict: matches, high confidence
    5. One step earlier: Counting names in curses.wide alone gives 49 with _w and 16 with w; the man page names the exceptions
  • Curses::Forms, linked to the CursesForms distribution, and Curses::Forms(3) (docs/client-library/more/widget-libraries.md)
    1. Claim, as printed: Curses::Forms, linked to the CursesForms distribution, and Curses::Forms(3)
    2. Opened: https://fastapi.metacpan.org/v1/module/Curses::Form (fetched 2026-11-15)
    3. It says: ““distribution” : “Curses”, “path” : “Curses.pm””
    4. Verdict: matches, medium confidence
    5. One step earlier: Curses::Forms is the separate CursesForms distribution, which made the original look wrong
  • The DOS games collection shown as a plain-text URL marked as a dead link (docs/client-library/menus/attributes.md)
    1. Claim, as printed: The DOS games collection shown as a plain-text URL marked as a dead link
    2. Opened: https://web.archive.org/web/20010607231808/http://www.jersey.net/~debinjoe/games/index.htm (opened by the owner 2026-11-15; this environment can’t reach the Wayback Machine)
    3. It says: “The snapshot shows the collection of text-mode DOS games”
    4. Verdict: matches, medium confidence
    5. One step earlier: Without a person to open it, the snapshot’s content couldn’t be checked, so the link was marked dead
  • He has revised every manual page and wrote several of them. (site/src/content/docs/index.mdx)
    1. Claim, as printed: He has revised every manual page and wrote several of them.
    2. Opened: README:171-176; PR 63 review on site/src/content/docs/index.mdx
    3. It says: “README: Eric S. Raymond: the man pages”
    4. Verdict: matches, high confidence
    5. One step earlier: The meta review replaced the original sentence on AUTHORS evidence alone; review rejected that change, so the original wording is restored
  • When a runnable example started, a vertical scroll bar flashed beside the terminal (site/src/components/Footer.astro)
    1. Claim, as printed: When a runnable example started, a vertical scroll bar flashed beside the terminal
    2. Opened: site/node_modules/@xterm/xterm/css/xterm.css 5.5.0; the viewport’s scroll bar width on every frame of a Hello World run in Chromium with scroll bars shown
    3. It says: “CSS: .xterm .xterm-viewport { overflow-y: scroll; }. Before: 15 px in all 140 frames; after: 0 px in all 143”
    4. Verdict: matches, high confidence
    5. One step earlier: Headless Chromium hides scroll bars by default, so the first screenshots showed no bar at all
  • In code listings, curses types and variables such as WINDOW, SCREEN, bool and stdscr were plain text with no hint; only function calls linked (site/code-links.mjs)
    1. Claim, as printed: In code listings, curses types and variables such as WINDOW, SCREEN, bool and stdscr were plain text with no hint; only function calls linked
    2. Opened: site/code-links.mjs (the CALL test); site/.generated/symbols.json
    3. It says: “Before: a name linked only if followed by (; WINDOW was in the index as typedef WINDOW; but never linked. After: WINDOW links to curs_variables(3X)#window”
    4. Verdict: matches, high confidence
    5. One step earlier: The links were first built and checked for calls only, so the listings’ types were never looked at
  • A function’s link opened the top of its man page, however far down the function is described (site/scripts/sync.mjs)
    1. Claim, as printed: A function’s link opened the top of its man page, however far down the function is described
    2. Opened: Every anchor in site/.generated/symbols.json checked against the ids in the built pages
    3. It says: “1,013 of 1,267 links carry an anchor and all 1,013 ids exist; the rest go to a page with no better section”
    4. Verdict: matches, high confidence
    5. One step earlier: A first version stopped at the first mention and sent getch to NOTES, and COLOR_PAIR to DESCRIPTION instead of its own heading
  • The December solstice from Meeus’s mean formula, up to 14 minutes from the published instant (site/src/components/boot-screen.js)
    1. Claim, as printed: The December solstice from Meeus’s mean formula, up to 14 minutes from the published instant
    2. Opened: USNO’s December solstice instants (aa.usno.navy.mil/api/seasons) for 2025–2036, the Hong Kong Observatory’s Lunar New Year, and Hebcal’s first Hanukkah candle
    3. It says: “2028 solstice: formula 08:32 UT, USNO 08:19 UT; the fixed code gives USNO’s instants, the Observatory’s 12 dates and Hebcal’s 12 dates”
    4. Verdict: matches, high confidence
    5. One step earlier: The mean formula, first ported from the mockup in this pull request, drifted up to 14 minutes and could put a reader near midnight on the wrong day
  • The exit note after a program’s last screen began with a new line, so a program that ended with the cursor on the bottom row lost its top two lines (site/src/components/run-example.ts)
    1. Claim, as printed: The exit note after a program’s last screen began with a new line, so a program that ended with the cursor on the bottom row lost its top two lines
    2. Opened: The Leaving curses mode example, run in a local build, before and after
    3. It says: “Before: only “System returned 768” showed; after: “Hello World !!!”, “Another String” and “System returned 768””
    4. Verdict: matches, high confidence
    5. One step earlier: The note was written as \r\n\r\n since the runnable examples were added
  • A program that read keys with nodelay() or timeout(0) froze until a key came: Emscripten answers a zero-timeout poll() with __syscall_poll_nonblocking, which called stdin always ready, so getch() blocked in read() (site/examples/pty.js)
    1. Claim, as printed: A program that read keys with nodelay() or timeout(0) froze until a key came: Emscripten answers a zero-timeout poll() with __syscall_poll_nonblocking, which called stdin always ready, so getch() blocked in read()
    2. Opened: A test program calling getch() five times after timeout(0), run in Node before and after
    3. It says: “Before: tick 0, then nothing until a key; after: tick 0 to tick 4, then done”
    4. Verdict: matches, high confidence
    5. One step earlier: pty.js has answered poll() since the runnable examples were added; no guide listing uses a zero timeout, so the test programs were the first to hit it
  • The pretend shell keeps them in /usr/games: ls leaves them out (site/README.md)
    1. Claim, as printed: The pretend shell keeps them in /usr/games: ls leaves them out
    2. Opened: site/src/components/page-shell.js: what ls, ls / and cd /usr/games do
    3. It says: “node(‘/’) lists home/ and usr/, and /usr/games is a folder like ~/menus”
    4. Verdict: matches, high confidence
    5. One step earlier: Written in this same pull request when the games were first added, before the examples moved into folders
  • Ctrl-C reached every program as a key, even in cbreak() mode, where the initialization guide says the terminal driver generates the signal (site/examples/pty.js)
    1. Claim, as printed: Ctrl-C reached every program as a key, even in cbreak() mode, where the initialization guide says the terminal driver generates the signal
    2. Opened: docs/client-library/getting-started/initialization.md, raw() and cbreak(); then bs (cbreak) and the raw() listing run in Node with Ctrl-C pressed
    3. It says: “In cbreak() mode the terminal driver still interprets them, and generates the signal as usual”
    4. Verdict: matches, high confidence
    5. One step earlier: pty.js has passed every byte through since the runnable examples were added; bs, which only quits mid-game on SIGINT, was the first program to need the signal
  • One agent session did the work, with the maintainer reviewing and merging each pull request; each pull request reviewed and merged by the maintainer (site/src/components/MigrationProgress.astro)
    1. Claim, as printed: One agent session did the work, with the maintainer reviewing and merging each pull request; each pull request reviewed and merged by the maintainer
    2. Opened: The Credits page and the front page’s cards, which name Thomas E. Dickey as the maintainer, and the repository’s merge history
    3. It says: “The front page said the maintainer reviewed and merged each pull request, while the Credits page names Thomas E. Dickey as the maintainer; the pull requests were merged by betsalel-williamson”
    4. Verdict: matches, high confidence
    5. One step earlier: Without the Credits page the word maintainer would have read as whoever merges the pull requests
  • The site’s manual pages dropped every quotation mark written with the pages’ quote strings (88 pages: see subsection Keypad Mode below), and term(5)’s carets (bel=G, cr=M) (site/scripts/man-render.mjs)
    1. Claim, as printed: The site’s manual pages dropped every quotation mark written with the pages’ quote strings (88 pages: see subsection Keypad Mode below), and term(5)’s carets (bel=G, cr=M)
    2. Opened: man/curs_getch.3x line 101 and man/term.5 against the site’s pages for them, and pandoc –verbose on every page
    3. It says: “see subsection *(Keypad Mode\*('' below. / [INFO] Skipped 'unknown string ’ (640 times across the pages)”
    4. Verdict: matches, high confidence
    5. One step earlier: The rendered page alone reads naturally without the quotes; only the troff source or pandoc’s verbose log shows them missing
  • The site printed TABLE in place of every table whose header spans columns or rows (captoinfo(1M), curs_attr(3X), term(5), and nine capability tables in terminfo(5)), dropped strings inside tables (curs_addch(3X) lost the cells for ACS_DEGREE, ACS_BULLET and ACS_PI), and infocmp(1M) showed %p1% %+%c (site/scripts/man-render.mjs)
    1. Claim, as printed: The site printed TABLE in place of every table whose header spans columns or rows (captoinfo(1M), curs_attr(3X), term(5), and nine capability tables in terminfo(5)), dropped strings inside tables (curs_addch(3X) lost the cells for ACS_DEGREE, ACS_BULLET and ACS_PI), and infocmp(1M) showed %p1% %+%c
    2. Opened: Rendered output of all 134 pages before and after, with pandoc 3.1.3 (2026-11-18)
    3. It says: “TABLE”
    4. Verdict: matches, high confidence
    5. One step earlier: pandoc reports a dropped table only as a warning on standard error, so the page built without failing
Where Was Now Source Change
docs/client-library/forms/ 39 function names, identifiers and constants written as plain prose (set_field_fore(), O_AUTOSKIP, TYPE_ENUM and others) Code spans; a Vale rule keeps it that way NCURSES Programming HOWTO and ncurses-intro.html #4, af8b605
docs/client-library/forms/fields.md field_pad() comment two columns left of the comments around it Lined up NCURSES Programming HOWTO #6, 3a0e739
docs/client-library/forms/custom-field-types.md Typedef comments out of line with the declarations below them Lined up ncurses-intro.html #6, 3a0e739
docs/client-library/forms/ Listings in mixed styles Complete programs follow the ncurses source style (docs/.clang-format) and trailing comments line up; a check in the site workflow keeps it that way This edition #6, 3a0e739
docs/client-library/menus/user-pointer-and-hooks.md See menu_opts(3x) for details. (the code span holds the rest of the sentence) See menu_opts(3X) for details. Writing Programs with NCURSES, The Menu Library #9, d03f995
docs/client-library/menus/menu-display.md Listings formatted by hand: #define CTRLD 4, the choices array brace on its own line, a case block with a second break after it The listings pass clang-format with docs/.clang-format; the Enter case keeps one break, inside its block This edition #9, d03f995
docs/client-library/forms/compiling.md #include <form.h> indented 10 spaces No indent ncurses-intro.html #10, 062f34d
docs/client-library/menus/compiling.md #include <menu.h> indented 10 spaces No indent ncurses-intro.html #10, 062f34d
docs/client-library/forms/ Trailing comments in listings ran past the content column, so the listing scrolled sideways (compiling.md, fields.md, field-options.md, custom-field-types.md) Each comment is on its own line above its statement. The two color comments in fields.md are one comment, with its missing closing parenthesis added NCURSES Programming HOWTO and ncurses-intro.html #10, 062f34d
docs/client-library/menus/ Long calloc and mvprintw calls ran past the content column (compiling.md, attributes.md, menu-windows.md, selecting-items.md, user-pointer-and-hooks.md) Split over two lines This edition #10, 062f34d
docs/client-library/panels/ Trailing comments past the content column, the calloc test split before == NULL, a bool named move beside the move() macro, and a two-line resize test Comments above their statements, the calloc result tested on its own line, the flag named moving, and two plain assignments before the test NCURSES Programming HOWTO #10, e57bdd2
docs/client-library/panels/compiling.md #include <panel.h> indented 10 spaces No indent ncurses-intro.html #10, e57bdd2
docs/client-library/getting-started/introduction.md echo “^[[0;31;40mIn Color”, typed with CTRL+V and Esc printf ‘\033[0;31;40mIn Color\n’, which can be copied and pasted NCURSES Programming HOWTO #11, eb995ee
docs/client-library/windows/ Trailing comments past the content column in the printw table and the centering comment; WIN * p_win Comments above their statements; WIN *p_win, as clang-format writes it NCURSES Programming HOWTO #12, 2b29d57
docs/client-library/output/printw.md Trailing comments past column 77 in the printw example, and the comment after the move() call Comments above their statements; long calls split under 77 columns, as clang-format writes them NCURSES Programming HOWTO, chapter 6 #13, eea2d03
docs/client-library/input/characters-and-strings.md Trailing comments past column 77 in the getstr example Comments above their statements, as clang-format writes them NCURSES Programming HOWTO, chapter 7 #15, d341cd3
docs/client-library/attributes/ Trailing comments past column 77 and braces on their own lines in the pager; long print_in_middle() signature Comments above their statements and calls split under 77 columns, as clang-format writes them NCURSES Programming HOWTO, chapters 8 and 10 #16, 282ca13
docs/client-library/keyboard/key-usage-example.md Long string arguments past column 77; the choices initializer brace on its own line Strings split with adjacent literals under 77 columns; the brace on the declaration line, as clang-format writes it NCURSES Programming HOWTO, chapter 11 #17, 8aff0ee
docs/client-library/mouse/mouse-menu-example.md The choices initializer on one line; a long condition and a trailing comment past column 77 The initializer one item a line; a len variable; the comment above its statement, as clang-format writes them NCURSES Programming HOWTO, chapter 12 #19, 02ebd3a
docs/client-library/output/line-drawing-characters.md SHOW() printed the 31 characters one a line, past the bottom of a 24-line screen Two columns of 16 with mvprintw(); clang-format off and on markers in place of INDENT-OFF and INDENT-ON NCURSES Programming HOWTO, chapter 14 #24, dbe947c
docs/client-library/screen/leaving-curses-mode.md Trailing comments on each statement, some wrapped onto a second line The comments above their statements, as clang-format writes them NCURSES Programming HOWTO, chapter 14 #24, dbe947c
docs/contributor/hackguide/objective.md A missing closing parenthesis in the objectives list Closed A Hacker’s Guide to NCURSES #45, 6e167a6
docs/contributor/hackguide/porting-hints.md This modules assist This module assists A Hacker’s Guide to NCURSES #45, 6e167a6
docs/contributor/hackguide/documentation-conventions.md it make it makes A Hacker’s Guide to NCURSES #45, 6e167a6
docs/contributor/install/compatibility/ncurses-6.md opqaue opaque INSTALL #49, 2ff9553
docs/contributor/install/configure-options/enable.md binary- compatible (three options) binary-compatible INSTALL #49, 2ff9553
docs/contributor/install/configure-options/enable.md which stores characters as wide-characters, which stores characters as wide-characters. INSTALL #49, 2ff9553
docs/contributor/install/configure-options/enable.md either version of the the ncurses library either version of the ncurses library INSTALL #49, 2ff9553
docs/contributor/install/configure-options/with-m-z.md listed in the the “.map” files listed in the “.map” files INSTALL #49, 2ff9553
docs/contributor/install/mingw.md –enable-database given twice in the configure command –enable-database given once README.MinGW #49, 2ff9553
docs/contributor/announce/portability.md attemping attempting announce.html.in #52, 54088dc
docs/client-library/getting-started/hello-world.md the string is printed at the left hand corner of the window the string is printed at the upper left-hand corner of the window NCURSES Programming HOWTO, Hello World #54, 74a7f5a
docs/client-library/output/characters-and-strings.md These attribute macros can be found in the header file ncurses.h; getmaxyx(), a macro defined in ncurses.h curses.h in both places, the header Hello World includes NCURSES Programming HOWTO, Output #54, 74a7f5a
docs/client-library/forms/fields.md Typos and slips across the Forms guide: mist, from, an hidden, by of, declare of, of course., locale’s, than, have do, types, our own, unclosed parentheses, a sentence starting ‘As with other field attributes, Also’, the menu driver loop, free_form without () Corrected text Writing Programs with NCURSES, The Forms Library #56, e117d45
docs/client-library/forms/form-driver.md Form requests are represented as integers above the curses value greater than KEY_MAX and less than or equal to the constant MAX_COMMAND Form requests run from MIN_FORM_COMMAND (KEY_MAX + 1) to MAX_FORM_COMMAND; other values return E_UNKNOWN_COMMAND Writing Programs with NCURSES, The Forms Library #56, e117d45
docs/client-library/hints/cautions.md [Windows and refresh] [Windows and the screen], the page’s title This edition #56, e117d45
docs/contributor/install/installation-procedure.md Two tables with ragged columns Columns aligned by the formatter INSTALL #57, 769b14b
docs/contributor/hackguide/terminfo-compiler.md the AIX box1 capability get translated gets translated A Hacker’s Guide to NCURSES #63, 3cc6b89
docs/client-library/ Function and variable names in six headings and captions written as plain prose (getstr and printw examples, chgat usage example, stdscr) Code spans, with () on functions; page titles stay plain because the sidebar shows them as written NCURSES Programming HOWTO #68, 9a17eb9
docs/contributor/install/compatibility/ Function, type and variable names in the change lists written as plain prose (getbegx through getpary, wgetparent, tiparm, vwprintw, WINDOW, SCREEN, chtype, ospeed, slk_* and others) Code spans, like the names around them that were already marked INSTALL #68, 9a17eb9
docs/contributor/install/configure-options/ Names written as plain prose or in quotes (chtype, ospeed, SCREEN, SP, LINES, tparm, wattr*, tputs(“50”), “beep” and “flash”) Code spans INSTALL #68, 9a17eb9
site/src/styles/layout.css On a wide screen the columns spread to the window’s edges: the text drifted away from the sidebar (298px at 1920, 602px at 2560) and Text mode’s On this page box stretched across the right side From 1440px up the sidebar, text and On this page column sit in one centred 1440px frame with equal side margins; the On this page column is capped at 16.25rem and the text is centred in its column This edition #74, 469e011
site/src/styles/styles.css The retro styles drew their scrollbars only when a page was long enough to scroll, so the layout shifted by the bar’s width between short and long pages The page’s and the sidebar’s scrollbars always show in the retro styles, as part of the interface This edition #75, bf42bee
site/src/components/MigrationProgress.astro API cost was shown without saying it wasn’t paid API cost has an asterisk and a note that the work ran on a flat-rate subscription that includes this usage This edition #75, 51b390f
site/src/scripts/token-violins.js The violins’ tooltip showed the file path as code, which the site styles restyled: invisible in Manual page, boxed in Workstation and Text mode The path is plain monospaced text in the tooltip’s own colours, at 10:1 contrast or more in every style This edition #77, 4f8a03f
site/src/scripts/token-violins.js The tooltip sat above the lane, over the track before it, and ranked a file as 50 of 50 by size The tooltip sits beside the file, centred on its lane, and says the largest of 50 files This edition #77, 4f8a03f
site/src/components/MigrationProgress.astro Before and Now violins differed only by colour, and the two colours are close in lightness in every style; Text mode’s grid lines were bright white Before is dashed and lighter, so the two differ without colour; grid lines are a faint grey in every style This edition #77, 4f8a03f
site/scripts/man-render.mjs curs_add_wch(3X) and curs_addch(3X) showed a header stacked over two rows (Unicode over Default) as one line with the upper word in parentheses: Default (Unicode), Default (ACS), char (acsc) Each header cell keeps both lines as the printed page does: Unicode / Default, ACS / Default, acsc / char, with Symbol and Glyph Name on the bottom line; the tables stay markdown. A header that spans columns keeps the parenthesised group label This edition #81, 074dd1b
Where Was Now Source Change
docs/client-library/forms/index.md The HOWTO Forms chapter and ncurses-intro The Forms Library covered the same ground, with 30 paragraphs copied between them One guide of 10 pages, each crediting its sources, with no duplicate paragraphs NCURSES Programming HOWTO and ncurses-intro.html #1, f4e21cd
docs/client-library/forms/field-validation.md TYPE_ALPHA … TYPE_REGEXP were stray top-level <h1> headings in the middle of a chapter Subsections of Field Validation NCURSES Programming HOWTO (DocBook output) #1, f4e21cd
docs/man/common/background-attributes.3x The background attributes paragraph written out in both curs_bkgd and curs_bkgrnd One troff fragment that both pages include; stitch.sh still produces each page whole ncurses manual pages #6, 6bb4c7e
docs/man/common/background-color-ignored.3x The paragraph on color pairs without color support written out in both curs_bkgd and curs_bkgrnd One troff fragment that both pages include; stitch.sh still produces each page whole ncurses manual pages #6, 6bb4c7e
docs/man/common/svr4-immedok-return.3x The SVr4 immedok() return value remark (each page keeps its own SVID citation) written out in both curs_border and curs_clear One troff fragment that both pages include; stitch.sh still produces each page whole ncurses manual pages #6, 6bb4c7e
docs/client-library/menus/index.md Two menu chapters, in two HTML guides Nine shards merging Writing Programs with NCURSES and the NCURSES Programming HOWTO, so each fact appears once; the HOWTO’s asides (You want Colors !!!) are dropped This edition #9, d03f995
docs/client-library/panels/index.md The HOWTO Panel Library chapter and ncurses-intro The Panels Library covered the same ground, with one paragraph quoted from one in the other One guide of 7 pages, each crediting its sources, with no duplicate paragraphs NCURSES Programming HOWTO and ncurses-intro.html #10, e57bdd2
docs/client-library/getting-started/index.md HOWTO chapters 1–4 and the intro’s Introduction and Overview each introduced curses, compiling and the standard windows One guide of 7 pages, each crediting its sources, with no duplicate paragraphs. The intro’s Updating the Screen stays for the windows guide; the HOWTO’s empty chapter 3 is dropped NCURSES Programming HOWTO and ncurses-intro.html #11, eb995ee
site/src/content/docs/credits.md HOWTO sections 1.5–1.9 (the example tarball, other formats, credits, wish list, copyright) described the HOWTO as a separate document Its sources, example tarball and acknowledgements are on the credits page; the wish list is dropped NCURSES Programming HOWTO #11, eb995ee
docs/client-library/windows/index.md HOWTO chapter 5, chapter 9 and the intro’s Updating the Screen each explained windows and refresh, and chapter 5 repeated the naming convention from the intro One guide of 4 pages; the naming convention is linked to Getting started instead of repeated NCURSES Programming HOWTO and ncurses-intro.html #12, 2b29d57
docs/client-library/output/index.md HOWTO chapter 6 and the intro’s Output and Using Forms Characters sections each described addch() and the ACS characters One guide of 3 pages; the intro’s refresh and touchwin paragraph joins the output page, and the naming convention is linked to Getting started NCURSES Programming HOWTO and ncurses-intro.html #13, eea2d03
docs/client-library/input/index.md HOWTO chapter 7 and the intro’s Input section each described getch(), echo and the line-reading functions; the intro’s keypad paragraph pointed at an example earlier in that document One guide of 2 pages; the keypad paragraph links to Initialization, where keypad() is explained NCURSES Programming HOWTO and ncurses-intro.html #15, d341cd3
docs/client-library/attributes/index.md HOWTO chapters 8 and 10 and the intro’s Character Attributes and Color each explained highlights and color pairs; the intro referred to sample code earlier in that document One guide of 4 pages; the intro’s eight-pair example is described in words NCURSES Programming HOWTO and ncurses-intro.html #16, 282ca13
docs/client-library/keyboard/index.md HOWTO chapter 11, one page with two sections One guide of 2 pages: Reading keys and A simple key usage example NCURSES Programming HOWTO, chapter 11 #17, 8aff0ee
docs/client-library/mouse/index.md HOWTO chapter 12 and the intro’s Mouse Interfacing each described mousemask(), getmouse() and clicks One guide of 2 pages; each fact appears once NCURSES Programming HOWTO and ncurses-intro.html #19, 02ebd3a
docs/client-library/screen/index.md Chapters 13 and 14 of the HOWTO, one page each A Screen and other features guide of six pages, with its own sidebar group This edition #24, dbe947c
docs/client-library/init-and-debugging/index.md Finishing Up and Function Descriptions, two sections of one long page An Initialization, wrapup and debugging guide of four pages, with its own sidebar group; Finishing Up merges into the endwin() section This edition #26, 808b628
docs/client-library/hints/index.md Hints, Tips, and Tricks, Compatibility with Older Versions and XSI Curses Conformance were three sections of one long page One guide of 6 pages: Tuning for Speed joins the cautions, Special Features becomes ncurses extensions, and XSI conformance joins compatibility Writing Programs with NCURSES #32, c812489
docs/client-library/more/index.md Other Libraries (ch. 15), Tools and Widget Libraries (ch. 19) and Just For Fun (ch. 20) named the programs’ files in the example archive One guide: a widget libraries page, and each Just For Fun program as its own page with its listing and a Run button NCURSES Programming HOWTO #33, bec4c6d
docs/appendix/references.md References was the last chapter of the HOWTO A page in the appendix, beside the errata NCURSES Programming HOWTO, References #34, c43feed
docs/man/common/insert-character-return-value.3x The RETURN VALUE section written out word for word in both curs_insch and curs_ins_wch One troff fragment that both pages include; stitch.sh still produces each page whole ncurses manual pages #35, 6811c12
docs/man/curs_getch.3x The 19 input pages (curs_get_wch to keyok) each written as one troff file Each page a skeleton of .so lines, one fragment per section; stitch.sh still produces each page whole ncurses manual pages #36, 8d05a01
docs/man/curs_attr.3x The 6 attribute, color and soft-label pages (curs_attr to new_pair) each written as one troff file Each page a skeleton of .so lines, one fragment per section; stitch.sh still produces each page whole ncurses manual pages #37, ce3da71
docs/man/curs_window.3x The 9 window, pad and refresh pages (curs_move to wresize) each written as one troff file Each page a skeleton of .so lines, one fragment per section; stitch.sh still produces each page whole ncurses manual pages #39, e747608
docs/man/curs_initscr.3x The 18 initialization, terminal and low-level pages (curs_border_set to term_variables) each written as one troff file Each page a skeleton of .so lines, one fragment per section; stitch.sh still produces each page whole ncurses manual pages #40, e238b12
docs/man/form.3x The 47 form, menu and panel pages (form to form_win, menu to mitem_visible, and panel) each written as one troff file Each page a skeleton of .so lines, one fragment per section; stitch.sh still produces each page whole ncurses manual pages #41, 2d57d2c
docs/man/common/ncurses-specific-portability.3x The PORTABILITY section written out word for word in form_requestname, menu_requestname and menu_spacing One troff fragment that the three pages include; stitch.sh still produces each page whole ncurses manual pages #41, 2d57d2c
docs/man/common/default-windows-stdscr.3x The paragraph saying both windows default to stdscr written out word for word in form_win and menu_win One troff fragment that both pages include; stitch.sh still produces each page whole ncurses manual pages #41, 2d57d2c
docs/man/tic.1m The 8 program pages (captoinfo to tset) and scr_dump(5), term(7) and user_caps(5) each written as one troff file Each page a skeleton of .so lines, one fragment per section; stitch.sh still produces each page whole ncurses manual pages #42, 6051a69
docs/man/infocmp/03-description.1m infocmp’s DESCRIPTION section, 2539 words, as one fragment One fragment per subsection (Default Options to Other Options); stitch.sh still produces the page whole ncurses manual pages #42, 6051a69
docs/man/common/no-size-restrictions.1m The description of the -T option written out word for word in infocmp and tic One troff fragment that both pages include; stitch.sh still produces each page whole ncurses manual pages #42, 6051a69
docs/contributor/hackguide/ doc/html/hackguide.html, one HTML page Twelve shards, one per section, published under Contributing A Hacker’s Guide to NCURSES #45, 6e167a6
docs/contributor/hackguide/reporting-bugs.md A link to ncurses-intro.html#debugging A link to the Debugging page of this edition’s guides A Hacker’s Guide to NCURSES #45, 6e167a6
docs/contributor/install/bug-reports.md The Hacker’s Guide in the doc directory A link to the Reporting bugs page of this edition’s hacker’s guide INSTALL #49, 2ff9553
docs/contributor/install/index.md INSTALL, README.MinGW and README.emx, three plain-text files One guide of 13 pages, with the configure options on five pages by prefix and the compatibility notes on four pages by release INSTALL, README.MinGW and README.emx #49, 2ff9553
docs/contributor/ada/packages.md Links to the gnathtml pages in doc/html/ada Links to the specifications in Ada95/src and Ada95/gen; the generated pages stay generated Ada95 Binding for ncurses #50, e3fa730
docs/contributor/ada/index.md Ada95/README: The documentation is provided in HTML format in the ./html subdirectory. The main document is named index.html This guide (Ada95 has no html subdirectory; the document was doc/html/Ada95.html) Ada95/README #50, e3fa730
docs/contributor/ada/samples.md The samples README and the demo help text, explain.txt A sample programs page with the menu and form keys as tables Ada95/samples/README and explain.txt #50, e3fa730
docs/contributor/announce/index.md One page from announce.html.in, a template with @VERSION@, @HOMEPAGE@ and @WEBSITE@ Eleven pages by section, with the placeholders filled in for 6.6 and the in-page table of contents left to the site announce.html.in #52, 54088dc
docs/client-library/getting-started/introduction.md Links to the Mouse Interfacing, Debugging, Compatibility and Hints sections of ncurses-intro.html on invisible-island.net Links to this site’s mouse, debugging, compatibility and hints pages This edition #54, 74a7f5a
docs/client-library/getting-started/get-ncurses.md A link to INSTALL.html on invisible-island.net A link to the installation guide on this site This edition #54, 74a7f5a
docs/client-library/forms/overview.md Links into doc/html/ncurses-intro.html for the menu library, panels and menu input handling (overview, form-windows, form-driver) Links to this site’s Menus and Panels pages This edition #56, e117d45
docs/client-library/forms/field-options.md The paragraphs on changing options of the selected field and on combining option bits appeared twice Once, under Field Option Bits This edition #56, e117d45
docs/appendix/index.md The references page was reachable only from the sidebar, so merlon check reported it as an orphan The appendix introduction links it NCURSES Programming HOWTO #59, 8e5de3e
docs/appendix/index.md The appendix introduction did not point to a meta-verification process The introduction links Errata meta-verification and the current handoff report This edition #63, 184d54b
docs/appendix/meta-verification.md The page did not point to an adjudication of the outside pass Its summary and artifact table link the adjudication This edition #63, 82580d2
docs/appendix/meta-verification/adjudication.md Counts as rows confirmed, rows wrong and can’t tell, with verdicts in the errata convention Counts as true and false positives and negatives, type I and type II errors, for the errata rows and the checkers’ Bo3 majority, with verdicts in the prompt’s convention This edition #63, 58da765
docs/client-library/menus/attributes.md The sentence about text-mode DOS games and its link were removed The sentence is back, with the HOWTO’s URL as plain text marked as a dead link This edition #63, bc49d13
docs/contributor/hackguide/reporting-bugs.md Join from the subscription page, or mail bug-ncurses-request@gnu.org with the line subscribe @<host.domain> Join from the Mailman subscription page only This edition #63, bc49d13
docs/appendix/meta-verification/adjudication.md L179 and L286 left as can’t tell; nine type I errors L179 settled as a type I error and L286 settled from the repository; ten type I errors and 8 unsettled rows, with the introduction and lessons learned updated to match This edition #63, 981621b
docs/glossary/ The sidebar listed the terms in the overview’s topic order (window, stdscr, subwindow, pad …) The sidebar lists the terms A to Z; the overview keeps its topic groups This edition #63, 8141bf1
site/astro.config.mjs Credits first; Attributes and colors between Input and Keyboard; Other libraries before Forms, Menus, Panels; Hacker’s guide before Installing ncurses; meta-verification last in the Appendix; sidebar labels unlike their page titles; first pages named Introduction or Overview Guides in reading order: Output, Attributes and colors, Input, Keyboard, Mouse; Panels, Menus, Forms, then Other libraries; Installing before the hacker’s guide; Credits before the Appendix; meta-verification after the errata; labels match titles; every first page is Introduction This edition #63, 6b6c4d6
docs/appendix/meta-verification/adjudication.md L55 marked as a dead link pending a check of the archived copy L55 linked to the archived copy the owner confirmed This edition #63, c659c30
site/astro.config.mjs Credits moved to just before the Appendix Credits first in the sidebar, since this edition is a tribute and its history comes first This edition #63, 99cee2b
docs/client-library/output/line-drawing-characters.md Under Screen manipulation and other features Under Output, after the forms-drawing characters it illustrates This edition #63, 99cee2b
docs/client-library/getting-started/initialization.md The page ended without pointing to the fuller Initialization and wrapup page A closing sentence links Initialization and wrapup for the calls before initscr() and endwin() This edition #63, 99cee2b
docs/contributor/install/ The sidebar followed the first links in the introduction’s prose, so the installation procedure came fifth The sidebar follows the introduction’s Guide list, so the installation procedure comes first This edition #63, 99cee2b
site/astro.config.mjs About twenty sections side by side in the sidebar, with only Installing ncurses nested Four areas under Credits, as MDN groups its docs: Learn curses (tutorials), Guides, Reference (manual pages and glossary) and About this edition (release announcement and appendix) This edition #63, 694e609
docs/appendix/meta-verification/adjudication.md L195 said the front page now credits Dickey with writing several pages L195 says the owner kept the sentence as first written and an open row lists the twelve pages This edition #63, 8f32c73
site/src/components/run-example.ts A runnable example showed only its Run button and options until it was run The terminal stands blank under the listing, with the Run button and options below it This edition #71, c0d2d5f
site/src/components/boot-screen.js The easter eggs’ dates computed in the browser: the Hebrew and Chinese calendars from Intl, the solstice from tables in the code Every date read from site/data/calendar.json, 2025–2036, each with its source (Hebcal, the Hong Kong Observatory, the US Naval Observatory, 5 U.S.C. 6103 or a fixed date); a year past the file gets no occasion This edition #71, cf1c038
site/src/components/MigrationProgress.astro The time figures recomputed from a timeline of every session event (site/data/session/timeline.csv) The time figures are the sums of the session’s time totalled by day (site/data/session/daily.csv) This edition #71, 67de6da
site/src/components/MigrationProgress.astro Done was counted by documents and by words, as two percentages Done gives half for the documents sharded and half for the tokens reviewed, in the tokenizer the dropdown picks; the words bar is gone This edition #75, 51b390f
site/data/migration.json The session figures’ reviewer fields were named maintainerSeconds and maintainerMessages They are reviewerSeconds and reviewerMessages, in migration.json, session/daily.csv and migration-stats.mjs This edition #76, 9eeec79
Where Was Now Source Change
docs/client-library/forms/fields.md — The Form Attributes example runs in the page, compiled to WebAssembly This edition #4, c48d0f7
docs/client-library/forms/fields.md — Try it: make Backspace and Delete work, a tutorial step with a second runnable example This edition #5, e10fc82
docs/client-library/forms/form-windows.md Only the two listings in fields.md had a Run button, from a hand-written list Every complete program in the guides is built and run in CI; each gets a Run button, or a notice with its build log when it doesn’t run yet. 5 of 5 run This edition #7, 880098f
docs/appendix/lessons-learned.md — A lessons-learned page: what went wrong in the migration, why, and what now catches it This edition #7, e384c3c
docs/client-library/menus/compiling.md — All seven menu programs build to WebAssembly and run on the site This edition #9, d03f995
docs/client-library/panels/compiling.md — All four panel programs build to WebAssembly and run on the site This edition #10, e57bdd2
docs/client-library/panels/compiling.md — panel.h includes curses.h, so a panel program needs only one include man/panel.3x NOTES #10, e57bdd2
site/src/content/docs/credits.md — The copyright and permission notices of both guides, which their license requires in every copy ncurses-intro.html header; NCURSES Programming HOWTO section 1.9 #11, eb995ee
docs/client-library/getting-started/initialization.md — Both programs build to WebAssembly and run on the site This edition #11, eb995ee
docs/client-library/windows/creating-windows.md — Both programs build to WebAssembly and run on the site; the text says what happens when the box goes off screen This edition #12, 2b29d57
docs/client-library/output/printw.md — The printw example builds to WebAssembly and runs on the site This edition #13, eea2d03
docs/client-library/menus/compiling.md — A sentence explains that new_item() takes a name and a description, and that the example passes the same text for both, so each choice shows twice This edition #13, eea2d03
site/src/content/docs/credits.md — GitHub accounts linked for Thomas E. Dickey, Alexander V. Lukyanov and G. Branden Robinson, each tied to the person by the account’s own public repositories This edition #13, 038aeea
docs/client-library/windows/border-functions.md — A sentence says what print_win_params() prints and that a Run with _DEBUG defined button runs that build This edition #14, 16fc371
site/ — Examples that test a macro of their own get a second build with it defined; the terminal can be 80x24, 100x30 or 60x20, and resizing it during a run sends the program SIGWINCH and KEY_RESIZE This edition #14, 16fc371
docs/client-library/input/characters-and-strings.md — The getstr example runs on the site; the page says how to end it and that F1 is ignored while a line is read; a paragraph explains getnstr() and the LINE_MAX limit This edition #15, d341cd3
docs/client-library/attributes/ — All three programs run on the site; the pager reads its own listing, embedded in its build as example.c; the terminal bridge reads real files instead of treating every non-stdin read as end of file This edition #16, 282ca13
docs/client-library/keyboard/key-usage-example.md — The example runs on the site; the page says that Enter arrives as 10 because ncurses maps carriage return to newline while nl() is on, that F1 does not end this program, and that the closing message disappears on terminals with an alternate screen This edition #17, 8aff0ee
docs/client-library/mouse/mouse-menu-example.md — The example runs on the site and responds to clicks; ncurses for the site is built with a working poll(), so click resolution and ESCDELAY wait as on a terminal; the page explains BUTTON1_CLICKED and the coordinates This edition #19, 02ebd3a
docs/client-library/keyboard/key-usage-example.md The page said the closing message disappears with the menu The page says xterm loses it at endwin(), and that the terminal on this page keeps the last screen in view, so the choice can be read This edition #20, 351d87c
docs/client-library/screen/coordinates.md getparyx() gets the beginning coordinates of the subwindow relative to the main window getparyx() gives -1 for both if the window isn’t a subwindow man/curs_getyx.3x #24, dbe947c
docs/client-library/screen/saving-and-copying.md Later it can be restored by scr_restore function scr_restore() reads the dump into the virtual screen, and the next doupdate() or refresh() puts it on the terminal man/curs_scr_dump.3x #24, dbe947c
docs/client-library/screen/cursor-visibility.md The parameter values 0, 1 and 2 curs_set() returns the previous visibility, or ERR if the terminal can’t show the one asked for man/curs_kernel.3x #24, dbe947c
docs/client-library/screen/leaving-curses-mode.md — The page says the example starts a shell, so it can’t run on the site, and the site shows that note in place of a Run button This edition #24, dbe947c
docs/client-library/init-and-debugging/initialization-and-wrapup.md A few functions may be called before initscr (slk_init(), filter(), ripoffline(), use_env(), newterm()) The list adds use_tioctl() and says what each function is for man/curs_initscr.3x #26, 808b628
docs/client-library/init-and-debugging/initialization-and-wrapup.md — newterm() returns a null pointer on an error instead of exiting; delscreen() comes after endwin(), which doesn’t free the SCREEN man/curs_initscr.3x #26, 808b628
site/src/content/docs/credits.md — This edition explains the errata and lessons learned, and why it exists: an example of AI agent skills reviewing documentation sharded into guides This edition #28, 8ae3d2d
docs/client-library/hints/resizing.md — Call resizeterm() from the main loop, not from a SIGWINCH handler, because it allocates memory Writing Programs with NCURSES, Hints, Tips, and Tricks #32, c812489
docs/client-library/hints/resizing.md — resizeterm() can’t resize pads Writing Programs with NCURSES, Hints, Tips, and Tricks #32, c812489
docs/client-library/hints/cautions.md — immedok() makes every change to the window call wrefresh() at once, which can degrade performance considerably when writes are frequent Writing Programs with NCURSES, Hints, Tips, and Tricks #32, c812489
docs/appendix/references.md NCURSES man pages; Writing programs with NCURSES at invisible-island.net, somewhat obsolete Links to this site’s manual pages and to the guides that carry Writing Programs with NCURSES, corrected; the FAQ and the original link were checked and kept NCURSES Programming HOWTO, References #34, c43feed
docs/contributor/hackguide/library-tour.md — A note that lib_gen.c, lib_keyname.c, unctrl.c and comp_captab.c are generated at build, and what the base, tinfo, tty and trace subdirectories hold A Hacker’s Guide to NCURSES #45, 6e167a6
docs/contributor/hackguide/reporting-bugs.md — A link to the bug-ncurses archive A Hacker’s Guide to NCURSES #45, 6e167a6
docs/contributor/hackguide/porting-hints.md — A note on the MinGW port to the Win32 console and its win32con driver A Hacker’s Guide to NCURSES #45, 6e167a6
docs/contributor/install/installation-procedure.md — A note that the wide-character libraries (with the “w”) are the default since ncurses 6.5, and –disable-widec builds the 8-bit ones INSTALL #49, 2ff9553
docs/contributor/install/configure-options/enable.md — An entry for –enable-rgb-color, which INSTALL does not document INSTALL #49, 2ff9553
docs/contributor/install/configure-options/enable.md — A note that –enable-const, –enable-ext-colors, –enable-ext-mouse, –enable-ext-putwin, –enable-interop and –enable-sp-funcs are on by default for ABI 6 INSTALL #49, 2ff9553
docs/contributor/install/system-integrators.md — A note that only linux is a terminfo entry; the database calls the others cons25 (FreeBSD), wsvt25 (NetBSD) and bsdos-pc (BSD/OS) INSTALL #49, 2ff9553
docs/contributor/install/emx.md — A note that this distribution’s configure.in needs autoconf 2.52 with Thomas Dickey’s patches, not the autoconf 2.13 the EMX steps use README.emx #49, 2ff9553
docs/contributor/ada/packages.md — Aux, PutWin, Termcap, Terminfo and Trace added to the package hierarchy, which listed none of them Ada95 Binding for ncurses #50, e3fa730
docs/contributor/ada/implementation-details.md TBD A description of User_Defined_Field_Type, Field_Check, Character_Check and the Choice package Ada95 Binding for ncurses #50, e3fa730
docs/glossary/index.md No glossary: the guides used curses terms such as stdscr, pad, color pair and terminfo without a shared definition A glossary of 22 terms, each checked against the man pages and the library source, with an inclusion bar for new terms. Each guide links a term where it first uses it (103 links in 56 shards). This edition #53, df81beb
docs/client-library/getting-started/naming-and-variables.md — A sentence on newscr, the virtual screen that wnoutrefresh() fills when a program uses several windows This edition #54, 74a7f5a
docs/appendix/meta-verification.md — Errata meta-verification: when to run a best-of-three audit, how a pass works, verdict framing, and pointers to the current outside pass This edition #63, 184d54b
docs/appendix/meta-verification/ — The first outside Cursor meta-verification package: cover note, handoff report, Bo3 findings, eighteen run cards and six batch inputs This edition #63, 184d54b
docs/appendix/lessons-learned.md — Meta-verification belongs in another system: the failed in-session attempt and the outside Cursor pass This edition #63, 184d54b
docs/appendix/meta-verification/adjudication.md — Adjudication of the first outside pass: every misaligned row and split rechecked from primary evidence, with a five-step card per row in findings/adjudication.json This edition #63, 82580d2
docs/appendix/index.md — A paragraph on how a wrong fix (a type I error) is recorded: the row moves to rejected, and a restored text gets an edition-error row This edition #63, cd916c1
docs/appendix/lessons-learned.md — A section on the nine type I errors the adjudication of the first outside pass found, grouped by cause, with the lesson and guard This edition #63, cd916c1
docs/appendix/lessons-learned.md — A section on rows left unsettled because the agent couldn’t open the evidence, such as the Wayback Machine snapshot for L55, and how such rows are handed to a person This edition #63, 8a109b2
docs/areas/ — A landing page for each area: Learn curses, Guides, Reference and About this edition This edition #63, 694e609
docs/client-library/getting-started/get-ncurses.md Nothing on checking the release before unpacking it A paragraph after the install steps gives 6.6’s SHA-256, the .sig and .asc signature files, and the signing key’s fingerprint; the steps themselves are unchanged This edition #66, c8447d2
site/prose-links.mjs — Inline code in the guides that names a library function (with its parentheses), variable, type or constant links to its manual page, with the same hint as in code listings; the hint names the release it quotes, and the compatibility notes leave out the declaration This edition #68, ccc83a4
site/scripts/sync.mjs — The index of linked names gains the panel and form functions declared in each page’s SYNOPSIS, and about 340 named constants (REQ_, O_, KEY_, A_, ACS_, E_ and others) with their meanings from the pages’ tables This edition #68, ccc83a4
docs/appendix/lessons-learned.md — A section on why the 6.6 hints are wrong on the compatibility notes, and how the strict Merlon Docs method avoids it This edition #68, cacd9d9
site/src/components/boot-screen.js — While a runnable example loads, its terminal plays a boot screen: 4 s on a reader’s first run, 1.2 s after, one static line with reduced motion This edition #70, 86ae1ac
site/src/components/run-example.ts — On a slow link the boot screen’s bar follows the bytes of the program that have arrived, and the program starts as soon as they have This edition #70, d915ee8
site/src/components/boot-screen.js One dim “Loading” line with reduced motion With reduced motion the boot report is drawn at once and still, every check ok, and only the bar moves, with the download This edition #70, 9bc2178
site/src/components/boot-screen.js — Boot screen easter eggs, each slowing the boot to 5 s: a verbose boot on Shift+Run, the worm bar, rain on r and an ESCDELAY note on Esc, and on the reader’s local date the ncurses birthday, US federal holidays and winter festivals, drawn still with reduced motion This edition #71, c0d2d5f
site/data/migration.json Session figures as of 15 November 2026, 08:05 UTC Session figures as of 15 November 2026, 22:14 UTC This edition #71, d973a35
site/data/migration.json Session figures as of 15 November 2026, 22:14 UTC Session figures as of 2026-11-15T22:28:47Z, refreshed as this release’s last commit This edition #71, b780b45
site/data/migration.json Session figures as of 15 November 2026, 22:28 UTC Session figures as of 2026-11-15T22:34:32Z, refreshed as this release’s last commit This edition #71, 33be8f8
site/data/migration.json Session figures as of 15 November 2026, 22:34 UTC Session figures as of 2026-11-15T22:47:45Z, refreshed as this release’s last commit This edition #71, 970fa50
docs/client-library/screen/leaving-curses-mode.md A web page has no shell, so this example doesn’t run on the site. A web page has no shell, so when you run it here a small pretend shell answers instead. Type help to see the few commands it knows. (The example runs, with system() answered by site/examples/pty.js) This edition #72, 0769fa6
site/data/migration.json Session figures as of 15 November 2026, 22:47 UTC Session figures as of 2026-11-16T03:51:27Z, refreshed as this release’s last commit This edition #72, da9072f
docs/client-library/screen/leaving-curses-mode.md A web page has no shell, so when you run it here a small pretend shell answers instead. Type help to see the few commands it knows. … Type help to see the few commands it knows, ls to list the other examples on this site, and ./ and a name to run one from inside this program. (The page’s shell, site/src/components/page-shell.js, also answers when a program ends; halt powers the terminal off) This edition #72, dfdfa08
site/data/migration.json Session figures as of 2026-11-16T03:51:27Z Session figures as of 2026-11-17T00:38:21Z, refreshed as this release’s last commit This edition #72, c3fc7c9
site/src/components/page-shell.js ls printed every program at once, so only the end of the list stayed on screen; the shell had no Tab completion ls stops at –More– each screen; Tab completes a name, twice lists the choices; Up brings back the last command This edition #73, ce5a458
site/scripts/examples.mjs The pretend shell ran only the guides’ examples ncurses’s own test programs from test/ (worm, rain, firework, xmas, tclock, knight, gdc, hanoi, blue) are built and kept in the shell’s /usr/games, and vi, nano and emacs open pretend editors This edition #73, ce5a458
site/data/migration.json Session figures as of 2026-11-17T00:38:21Z Session figures as of 2026-11-17T01:48:00Z, refreshed as this release’s last commit This edition #73, e098a8d
docs/client-library/screen/leaving-curses-mode.md … Type help to see the few commands it knows, ls to list the other examples on this site, and ./ and a name to run one from inside this program. … Type help to see the few commands it knows. The site’s other examples sit in a folder per guide: ls and cd find them, and ./ and a name runs one from inside this program. This edition #73, 548852c
site/src/components/page-shell.js Every example sat in one flat list in the home directory, run by its bare name A folder per guide under ~ (~/menus/menu-basics); cd, pwd and a prompt that shows the folder; ls of any folder or path; programs run by path, and a bare example name says which folder it is in; Tab completes paths This edition #73, 548852c
site/data/migration.json Session figures as of 2026-11-17T01:48:00Z Session figures as of 2026-11-17T16:24:08Z, refreshed as this release’s last commit This edition #73, b56178e
site/data/migration.json Session figures as of 2026-11-17T16:24:08Z Session figures as of 2026-11-17T16:26:55Z, refreshed as this release’s last commit This edition #73, 7458fa0
site/src/components/page-shell.js echo printed $PATH and other variables as typed echo expands $NAME, ${NAME} and $? (PATH, HOME, PWD, USER, SHELL, TERM), and env and printenv list them This edition #73, b8eed6f
site/data/migration.json Session figures as of 2026-11-17T16:26:55Z Session figures as of 2026-11-17T16:28:41Z, refreshed as this release’s last commit This edition #73, 4bcab28
site/src/components/page-shell.js The pretend shell knew a handful of commands, each without flags The shell keeps to real Unix command names and their common flags: ls -a -l -1 -F with colours, cd -, echo -n, cat, file, which, type, command -v, history, hostname, id, groups, tty, uptime, sleep, man, uname -snrma, date -u, reset, logout, Ctrl-L, sudo and su (refused), and rm, mkdir, touch, cp, mv and rmdir (a read-only file system) This edition #73, 4697106
site/data/migration.json Session figures as of 2026-11-17T16:28:41Z Session figures as of 2026-11-17T16:36:12Z, refreshed as this release’s last commit This edition #73, 4974ef5
site/scripts/examples.mjs bs (Battleships) from test/ was left out of /usr/games: the smoke test could not get past its first screens bs is in /usr/games; the smoke test presses a key past its title, picks a ship, then quits with Ctrl-Q, and Ctrl-C quits it mid-game This edition #73, 89203d4
site/data/migration.json Session figures as of 2026-11-17T16:36:12Z Session figures as of 2026-11-17T16:50:22Z, refreshed as this release’s last commit This edition #73, 80bb117
site/data/migration.json Session figures as of 2026-11-17T16:50:22Z Session figures as of 2026-11-17T19:23:16Z, refreshed as this release’s last commit This edition #74, 59e2128
site/src/components/MigrationProgress.astro Migration progress counted documents only, and said nothing about the docs’ size Progress is shown by documents and by words, and a table compares the docs’ files, words and tokens before the migration and now, with a choice of four tokenizers This edition #75, c7b7457
site/data/migration.json Session figures as of 2026-11-17T19:23:16Z Session figures as of 2026-11-17T20:08:07Z, refreshed as this release’s last commit This edition #75, 43fc277
site/src/components/MigrationProgress.astro The docs’ size was shown only as totals per track Violins show each track’s tokens per file, before the migration and now, with one dot per file where a side has 12 files or fewer This edition #75, 95b2baa
site/data/migration.json Session figures as of 2026-11-17T20:08:07Z Session figures as of 2026-11-17T20:23:35Z, refreshed as this release’s last commit This edition #75, 7fa26a5
site/data/migration.json Session figures as of 2026-11-17T20:23:35Z Session figures as of 2026-11-17T20:49:58Z, refreshed as this release’s last commit This edition #76, 11669a3
site/data/migration.json Session figures as of 2026-11-17T20:49:58Z Session figures as of 2026-11-17T20:52:28Z, refreshed as this release’s last commit This edition #76, 59274f0
site/data/migration.json Session figures as of 2026-11-17T20:52:28Z Session figures as of 2026-11-17T21:07:14Z, refreshed as this release’s last commit This edition #77, 6d1e36c
.github/workflows/docs-site-build.yml The manual pages’ troff was checked only by the stitch check CI lints every page with mandoc, groff and pandoc’s man reader against a baseline (site/data/man-lint-baseline.json), shows each pull request’s rendered changes as a word diff in the job summary, and GitHub highlights the troff as Roff This edition #78, 4ac670d
site/data/migration.json Session figures as of 2026-11-17T21:07:14Z Session figures as of 2026-11-18T00:35:41Z, refreshed as this release’s last commit This edition #78, fa29ce8
docs/man/mitem_visible/04-return-value.3x The page had no RETURN VALUE section item_visible returns TRUE if item is visible; FALSE if it is not, if item is null, or if item is not connected to a posted menu man/mitem_visible.3x #78, b912aee
docs/man/mitem_visible/07-see-also.3x SEE ALSO: curses(3X), menu(3X) SEE ALSO: curses(3X), menu(3X), menu_post(3X), mitem_current(3X), the pages for posting a menu and for its top row man/mitem_visible.3x #78, b912aee
site/data/migration.json Session figures as of 2026-11-18T00:35:41Z Session figures as of 2026-11-18T00:44:18Z, refreshed as this release’s last commit This edition #78, d3403a7
site/src/components/page-shell.js The pretend shell had no terminfo directory, so a reader could not see how the examples’ ncurses finds its terminal description The shell has /usr/share/terminfo/x/xterm-256color: cat and file say it is compiled into the library (–disable-database –with-fallbacks=xterm-256color), and infocmp prints the entry, all generated from the examples’ own ncurses build This edition #79, a09fed7
site/scripts/terminfo.mjs The examples’ ncurses was configured with no –prefix, so its terminfo directory was /usr/local/share/terminfo Configured with –prefix=/usr, so its terminfo directory is /usr/share/terminfo, as in Linux distributions and in the manual pages This edition #79, 26729b5
site/src/components/MigrationProgress.astro Agent working time counted only the main thread’s steps Agent working time counts the subagents’ steps too, with overlapping work counted once This edition #79, 4976450
site/data/migration.json Session figures as of 2026-11-18T00:44:18Z Session figures as of 2026-11-18T16:19:30Z, refreshed as this release’s last commit This edition #79, 311dc57
site/src/content/docs/index.mdx The home page hero had text and buttons only, with an empty right half A terminal in the hero plays ncurses’s test programs firework, xmas, rain and worm from /usr/games, 10 seconds each; a click hands the running program to the reader, and q gives the pretend shell. It sits beside the text on wide screens and below it on narrow ones This edition #80, 16a83bd
site/src/components/home-terminal.ts The home page terminal had a Pause/Play button and a hint line under it, the Workstation style showed a light strip down its right side, and it opened as an empty box while the first program loaded No row under the terminal (status goes to screen readers only); the terminal box fits its 80 columns in every style; a still screen with a cursor holds its place until the script loads, then the boot sequence plays while the first program downloads This edition #80, f5c325c
site/data/migration.json Session figures as of 2026-11-18T16:19:30Z Session figures as of 2026-11-18T17:57:24Z, refreshed as this release’s last commit This edition #80, bac71e2
docs/man/captoinfo/03-description.1m The translation table listed 24 nonstandard capabilities It also lists sb (BSD, scroll_reverse), CF and CO (XENIX, cursor_invisible and cursor_normal), and kq (IBM, key_help), which include/Caps-ncurses translates include/Caps-ncurses #81, 50b61fe
docs/man/menu_requestname/03-description.3x menu_request_by_name searches the name table for a request with the given name. Otherwise E_NO_MATCH is returned. A request’s name is its constant without the REQ_ prefix, such as NEXT_ITEM, and the search ignores case (it uppercases the name first). The return value is left to RETURN VALUE. menu/m_req_name.c #81, 50b61fe
docs/man/form_requestname/03-description.3x form_request_by_name searches the name table for a request with the given name. Otherwise E_NO_MATCH is returned. A request’s name is its constant without the REQ_ prefix, such as NEXT_FIELD, and the search ignores case (it uppercases the name first). The return value is left to RETURN VALUE. form/frm_req_name.c #81, 50b61fe
site/data/migration.json Session figures as of 2026-11-18T17:57:24Z Session figures as of 2026-11-18T18:18:26Z, refreshed as this release’s last commit This edition #81, 14cb1d2
site/data/migration.json Session figures as of 2026-11-18T18:18:26Z Session figures as of 2026-11-18T19:17:14Z, refreshed as this release’s last commit This edition #81, 1dbca3e
site/data/migration.json Session figures as of 2026-11-18T19:17:14Z Session figures as of 2026-11-18T19:37:58Z, refreshed as this release’s last commit This edition #82, 239c662
site/data/migration.json Session figures as of 2026-11-18T19:37:58Z Session figures as of 2026-11-18T20:00:34Z, refreshed as this release’s last commit This edition #82, 368316c
docs/man/tput/05-exit-status.1 — Status 2 also covers the clear operand on a terminal with no clear capability; with several operands tput stops at the first whose status is not 0 man/tput.1 #84, a422f86
docs/man/tput/03-description.1 Before using a value returned on the standard output, the application should test tput’s exit status to be sure it is 0; … to be sure it is 0. The status is 0 for a numeric capability even when it is absent, so test for a value of -1 as well; man/tput.1 #84, a422f86
site/data/migration.json Session figures as of 2026-11-18T20:00:34Z Session figures as of 2026-11-18T20:10:33Z, refreshed as this release’s last commit This edition #84, 7652621
docs/man/define_key/03-description.3x — If definition is already bound to another key code, define_key moves the binding to key-code. man/define_key.3x #85, 72b9115
site/data/migration.json Session figures as of 2026-11-18T20:10:33Z Session figures as of 2026-11-18T20:37:54Z, refreshed as this release’s last commit This edition #85, e00b32a
Where Was Now Source Verdict
site/src/content/docs/index.mdx He has revised every manual page, which Eric S. Raymond first wrote. Credit both: Raymond wrote most of the original pages, and Dickey wrote the newer ones. Kept as written for now, by the owner’s choice This edition matches, high
docs/man/captoinfo/03-description.1m EE me AT&T exit_attribute_mode In the code, the AT&T alias EE→me is commented out and the active alias is EE→mh (XENIX), labelled exit_attribute_mode though mh is enter_dim_mode. Either the table or the alias is wrong; left for the maintainer This edition matches, medium
docs/man/captoinfo/03-description.1m Gr double tee pointing right / Gr double tee pointing left Both double tees are named Gr in the page and in the code; one is probably Gl. Kept as the code has it This edition matches, medium
docs/man/term/04-portability.5 the name field cannot exceed 128 bytes. ncurses allows names up to 512 bytes (MAX_NAME_SIZE); 128 bytes is the X/Open limit the comment cites. Kept, since the sentence is about portability; the maintainer may want both numbers This edition matches, medium
How each was checked
  • He has revised every manual page, which Eric S. Raymond first wrote. (site/src/content/docs/index.mdx)
    1. Claim, as printed: He has revised every manual page, which Eric S. Raymond first wrote.
    2. Opened: man/.3x, man/.1, man/.1m, man/.5, man/*.7: each page’s AUTHORS section (2026-11-15)
    3. It says: “133 pages; 65 have an AUTHORS section: Raymond is named on 53, Dickey alone on 12 (curs_extend, default_colors, define_key, key_defined, keybound, keyok, legacy_coding, new_pair, panel, resizeterm, wresize, user_caps); 68 name no author. All 133 carry Thomas E. Dickey’s copyright”
    4. Verdict: matches, high confidence
    5. One step earlier: Raymond wrote the bulk of the pages, so the sentence reads as true until the twelve pages that name only Dickey are counted
  • EE me AT&T exit_attribute_mode (docs/man/captoinfo/03-description.1m)
    1. Claim, as printed: EE me AT&T exit_attribute_mode
    2. Opened: include/Caps-ncurses:246,258
    3. It says: “#capalias EE me AT&T exit_attribute_mode / capalias EE mh XENIX exit_attribute_mode”
    4. Verdict: matches, medium confidence
    5. One step earlier: The table matches the AT&T origin, so it reads as right until the commented-out alias is noticed
  • Gr double tee pointing right / Gr double tee pointing left (docs/man/captoinfo/03-description.1m)
    1. Claim, as printed: Gr double tee pointing right / Gr double tee pointing left
    2. Opened: include/Caps-ncurses:276-277
    3. It says: “capalias Gr IGNORE XENIX double-ACS_LTEE / capalias Gr IGNORE XENIX double-ACS_RTEE”
    4. Verdict: matches, medium confidence
    5. One step earlier: The page copies the code, so the two agree and nothing flags the duplicate
  • the name field cannot exceed 128 bytes. (docs/man/term/04-portability.5)
    1. Claim, as printed: the name field cannot exceed 128 bytes.
    2. Opened: include/tic.h:104
    3. It says: “#define MAX_NAME_SIZE 512 /* maximum legal name field size (XSI:127) */”
    4. Verdict: matches, medium confidence
    5. One step earlier: The 128-byte figure is the portable limit, so it is true for the format if not for ncurses
Where Was Now Source Verdict
man/curs_bkgd.3x The same 51-word paragraph is in curs_bkgd.3x:101 and curs_bkgrnd.3x:92, so one copy is an error Not an error: the pages cover the parallel narrow and wide functions and each stands alone; the shared text becomes one troff fragment both include merlon review –no-compile (duplicate-paragraph) disagrees, medium
man/curs_border.3x The same 50-word paragraph is in curs_border.3x:228 and curs_clear.3x:117, so one copy is an error Not an error: the same SVr4 remark, made about two different sets of functions; the shared text becomes one troff fragment both include merlon review –no-compile (duplicate-paragraph) disagrees, high
man/term.5 term.5 and term.7 share the title “term” Not an error: one name in two sections is the man page convention, and the reader now titles them term(5) and term(7) merlon review –no-compile (similar-titles) superseded, high
man/curs_bkgd.3x The same 25-word paragraph is in curs_bkgd.3x and curs_bkgrnd.3x (If the terminal does not support color …), so one copy is an error Not an error: the same parallel narrow and wide pages; the text becomes one troff fragment both include merlon review –no-compile (duplicate-paragraph), after sharding disagrees, high
docs/client-library/output/characters-and-strings.md If n is negative, then the entire string will be added Not an error: ncurses adds the whole string for any negative n; the guide keeps that and adds that -1 is the documented value NCURSES Programming HOWTO, chapter 6 disagrees, high
docs/client-library/output/printw.md It gives the number of columns and the number of rows in a given window Not an error: the sentence names both values and says nothing about argument order; the guide’s reordered wording stays NCURSES Programming HOWTO, chapter 6 disagrees, medium
docs/client-library/screen/saving-and-copying.md copywin() can be used to copy a window completely onto another window Not an error: copywin() can copy a whole window when the rectangle covers it, and the next sentence gives the rectangle; the guide’s fuller wording stays NCURSES Programming HOWTO, chapter 13 disagrees, medium
docs/client-library/init-and-debugging/terminal-descriptions.md By calling setupterm() for each terminal, and saving and restoring cur_term, a program can use two or more terminals Not an error: saving and restoring cur_term is what term_variables(3X) describes, through set_curterm(); the guide’s set_curterm() wording stays Writing Programs with NCURSES, Function Descriptions disagrees, medium
docs/client-library/more/typing-tutor.md #define KEY_F1 265 Not an error: 265 is ncurses’ own KEY_F(1); the program now uses KEY_F(1), which is the same key NCURSES Programming HOWTO, Just For Fun (example archive) disagrees, high
docs/client-library/more/widget-libraries.md Curses::Form, from the CPAN All Modules page Not an error: Curses::Form is a real module, a package in Curses.pm in the Curses distribution, so the original’s name stands NCURSES Programming HOWTO, Tools and Widget Libraries disagrees, medium
site/src/content/docs/index.mdx Thomas E. Dickey wrote or revised most of the manual pages Not an error: Dickey’s copyright is on every manual page and he wrote several; the fix’s own wording was wrong and is corrected in a later row This edition’s front page disagrees, high
docs/client-library/input/characters-and-strings.md Initially the terminal has echoing enabled and is in ordinary cooked mode, so one or the other has to be changed before calling getch(); otherwise, the program’s output will be unpredictable Not an error: the manual calls getch() with echo and neither cbreak nor raw not well-defined; the warning is restored in a later row Writing Programs with NCURSES, Input disagrees, medium
docs/contributor/hackguide/terminfo-compiler.md This, for example, is where the AIX box1 capability gets translated to an acsc string. Not an error: parse_entry.c translates the AIX box1 capability (box_chars_1) to acsc; the example is restored in a later row A Hacker’s Guide to NCURSES disagrees, high
docs/glossary/wide-character-library.md most are named by adding _w to the 8-bit name, so waddch() becomes wadd_wch() Not an error: ncurses(3X) gives _w as the rule and the addwstr and inwstr families as the exceptions; the rule is restored in a later row This edition (glossary, PR 53, commit df81beb) disagrees, high
site/src/content/docs/index.mdx He has revised every manual page, which Eric S. Raymond first wrote must change because AUTHORS names Dickey on ten pages Not an error for the Maintainer card: Raymond first wrote the man pages and Dickey revised them; review kept that wording This edition’s front page (meta-verification adjudication) disagrees, high
How each was checked
  • The same 51-word paragraph is in curs_bkgd.3x:101 and curs_bkgrnd.3x:92, so one copy is an error (man/curs_bkgd.3x)
    1. Claim, as printed: The same 51-word paragraph is in curs_bkgd.3x:101 and curs_bkgrnd.3x:92, so one copy is an error
    2. Opened: man/curs_bkgd.3x and man/curs_bkgrnd.3x, DESCRIPTION, compared line by line
    3. It says: “Identical except: non-wide-character configuration / wide configuration; character (chtype) / complex character (cchar_t); waddch(3X) and winsch(3X) / wadd_wch(3X) and wins_wch(3X)”
    4. Verdict: disagrees, medium confidence
    5. One step earlier: I would have proposed deleting one copy, taking the text out of a page that is meant to stand alone
  • The same 50-word paragraph is in curs_border.3x:228 and curs_clear.3x:117, so one copy is an error (man/curs_border.3x)
    1. Claim, as printed: The same 50-word paragraph is in curs_border.3x:228 and curs_clear.3x:117, so one copy is an error
    2. Opened: man/curs_border.3x:224-244 and man/curs_clear.3x:113-133, PORTABILITY
    3. It says: “Both: SVr4 documentation says that these functions return OK “or a non-negative integer if immedok() is set”. The source comments differ: ." SVID 4, vol. 3, p. 483 and ." SVID 4, vol. 3, p. 485”
    4. Verdict: disagrees, high confidence
    5. One step earlier: I would have proposed deleting one copy, removing a correct portability note from one page
  • term.5 and term.7 share the title “term” (man/term.5)
    1. Claim, as printed: term.5 and term.7 share the title “term”
    2. Opened: man/term.5:32 and man/term.7:31; Merlon man reader title change (merlon-docs/merlon, commit 3a2f6a71)
    3. It says: “.TH term 5 2025-08-16 … “File formats” and .TH term 7 2025-08-16 … Miscellaneous”
    4. Verdict: superseded, high confidence
    5. One step earlier: I would have proposed renaming one of the pages
  • The same 25-word paragraph is in curs_bkgd.3x and curs_bkgrnd.3x (If the terminal does not support color …), so one copy is an error (man/curs_bkgd.3x)
    1. Claim, as printed: The same 25-word paragraph is in curs_bkgd.3x and curs_bkgrnd.3x (If the terminal does not support color …), so one copy is an error
    2. Opened: man/curs_bkgd.3x:171-174 and man/curs_bkgrnd.3x, DESCRIPTION; docs/man/curs_bkgd/03-description.3x:99 and docs/man/curs_bkgrnd/03-description.3x:97
    3. It says: “Both: If the terminal does not support color, or if color has not been initialized with start_color(3X), ncurses ignores the new background character’s color pair selection.”
    4. Verdict: disagrees, high confidence
    5. One step earlier: The survey reviewed whole pages and did not report this paragraph at all; it surfaced only once the pages were sharded by section
  • If n is negative, then the entire string will be added (docs/client-library/output/characters-and-strings.md)
    1. Claim, as printed: If n is negative, then the entire string will be added
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:1772-1773; ncurses/base/lib_addstr.c:51-79; man/curs_addstr.3x:84-90
    3. It says: “HOWTO: If n is negative, then the entire string will be added. lib_addstr.c (waddnstr): bool explicit = (n >= 0); … if (!explicit) n = INT_MAX; man: If n is -1, waddnstr writes the entire string.”
    4. Verdict: disagrees, high confidence
    5. One step earlier: With only the man page’s -1, the HOWTO looked overbroad; waddnstr() adds the whole string for any negative n
  • It gives the number of columns and the number of rows in a given window (docs/client-library/output/printw.md)
    1. Claim, as printed: It gives the number of columns and the number of rows in a given window
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:1725, 1743-1750; man/curs_getyx.3x
    3. It says: “HOWTO: It gives the number of columns and the number of rows in a given window. The example: getmaxyx(stdscr, row, col); /* get the number of rows and columns */”
    4. Verdict: disagrees, medium confidence
    5. One step earlier: Next to getmaxyx(win, y, x) the order looks swapped; the sentence is true as written
  • copywin() can be used to copy a window completely onto another window (docs/client-library/screen/saving-and-copying.md)
    1. Claim, as printed: copywin() can be used to copy a window completely onto another window
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:3211-3216; man/curs_overlay.3x:72-80
    3. It says: “HOWTO: The function copywin() can be used to copy a window completely onto another window. … according to the rectangle specified, it copies the rectangular region from source to destination window.”
    4. Verdict: disagrees, medium confidence
    5. One step earlier: Read alone, the first sentence sounds like a whole-window copy only; the paragraph goes on to the rectangle
  • By calling setupterm() for each terminal, and saving and restoring cur_term, a program can use two or more terminals (docs/client-library/init-and-debugging/terminal-descriptions.md)
    1. Claim, as printed: By calling setupterm() for each terminal, and saving and restoring cur_term, a program can use two or more terminals
    2. Opened: doc/html/ncurses-intro.html:1080-1084; man/term_variables.3x:126-140; man/curs_terminfo.3x:342-354
    3. It says: “intro: By calling setupterm() for each terminal, and saving and restoring cur_term, it is possible for a program to use two or more terminals at once. term_variables(3X): It is possible to save a value of cur_term for subsequent use. curs_terminfo(3X): set_curterm … returns the old value of cur_term.”
    4. Verdict: disagrees, medium confidence
    5. One step earlier: Read as assignment to cur_term, the advice fails in a reentrant build, where cur_term is a function; the original doesn’t say to assign it
  • #define KEY_F1 265 (docs/client-library/more/typing-tutor.md)
    1. Claim, as printed: #define KEY_F1 265
    2. Opened: include/Caps:439; include/MKkey_defs.sh:151-158
    3. It says: “Caps: key_f0 kf0 str k0 KEY_F(0) 0410. MKkey_defs.sh: KEY_F(n) (KEY_F0+(n)). 0410 octal is 264, so KEY_F(1) is 265.”
    4. Verdict: disagrees, high confidence
    5. One step earlier: A hard-coded key value looks wrong until it is computed
  • Curses::Form, from the CPAN All Modules page (docs/client-library/more/widget-libraries.md)
    1. Claim, as printed: Curses::Form, from the CPAN All Modules page
    2. Opened: doc/html/NCURSES-Programming-HOWTO.html:7385-7405; https://fastapi.metacpan.org/v1/module/Curses::Form and https://fastapi.metacpan.org/v1/module/Curses::Forms (fetched 2026-11-15)
    3. It says: “HOWTO: “The perl module Curses, Curses::Form and Curses::Widgets give access to curses from perl.” MetaCPAN: Curses::Form has “distribution”: “Curses”, “path”: “Curses.pm”; Curses::Forms has “distribution”: “CursesForms”, “abstract”: “Curses Forms Framework””
    4. Verdict: disagrees, medium confidence
    5. One step earlier: The CPAN index lists Curses::Forms as its own distribution, which looked like the module the HOWTO meant; Curses::Form also exists, inside Curses, so the original’s name isn’t wrong
  • Thomas E. Dickey wrote or revised most of the manual pages (site/src/content/docs/index.mdx)
    1. Claim, as printed: Thomas E. Dickey wrote or revised most of the manual pages
    2. Opened: README:171-176; the AUTHORS sections and copyright lines of every file in man/
    3. It says: “Every page carries Dickey’s copyright. AUTHORS: Thomas Dickey in curs_extend.3x, default_colors.3x, define_key.3x, key_defined.3x, keybound.3x, keyok.3x, legacy_coding.3x, new_pair.3x, resizeterm.3x, wresize.3x. README: Eric S. Raymond: the man pages”
    4. Verdict: disagrees, high confidence
    5. One step earlier: README alone supported the change; the AUTHORS sections show Dickey wrote several pages
  • Initially the terminal has echoing enabled and is in ordinary cooked mode, so one or the other has to be changed before calling getch(); otherwise, the program’s output will be unpredictable (docs/client-library/input/characters-and-strings.md)
    1. Claim, as printed: Initially the terminal has echoing enabled and is in ordinary cooked mode, so one or the other has to be changed before calling getch(); otherwise, the program’s output will be unpredictable
    2. Opened: doc/html/ncurses-intro.html:765-772; man/curs_getch.3x:631-633; ncurses/base/lib_getch.c:453-477
    3. It says: “intro: one or the other has to changed before calling getch(); otherwise, the program’s output will be unpredictable. curs_getch(3X): Use of wgetch with echo(3X) and neither cbreak(3X) nor raw(3X) is not well-defined.”
    4. Verdict: disagrees, medium confidence
    5. One step earlier: lib_getch.c alone shows ncurses handles cooked mode, which is how the fix was made; the manual keeps the case undefined
  • This, for example, is where the AIX box1 capability gets translated to an acsc string. (docs/contributor/hackguide/terminfo-compiler.md)
    1. Claim, as printed: This, for example, is where the AIX box1 capability gets translated to an acsc string.
    2. Opened: doc/html/hackguide.html:762-767; ncurses/tinfo/parse_entry.c:1077-1115; include/Caps:1031
    3. It says: “parse_entry.c (postprocess_terminfo): Translate AIX forms characters. if (PRESENT(box_chars_1)) { … acs_chars = _nc_save_str(buf2); _nc_warning(“acsc string synthesized from AIX capabilities”); Caps: box_chars_1 box1 str bx”
    4. Verdict: disagrees, high confidence
    5. One step earlier: A search for the literal box1 finds nothing in parse_entry.c; the C name is box_chars_1
  • most are named by adding _w to the 8-bit name, so waddch() becomes wadd_wch() (docs/glossary/wide-character-library.md)
    1. Claim, as printed: most are named by adding _w to the 8-bit name, so waddch() becomes wadd_wch()
    2. Opened: man/ncurses.3x:518-530; include/curses.wide
    3. It says: “ncurses(3X): where a non-wide function name contains ch or str, prefix it with _w to obtain the wide counterpart. … (Exceptions that add only w comprise addwstr, inwstr, and their variants.) curses.wide: 49 names with _w, 16 with a bare w”
    4. Verdict: disagrees, high confidence
    5. One step earlier: waddstr() and waddwstr() alone made the exception look like the rule
  • He has revised every manual page, which Eric S. Raymond first wrote must change because AUTHORS names Dickey on ten pages (site/src/content/docs/index.mdx)
    1. Claim, as printed: He has revised every manual page, which Eric S. Raymond first wrote must change because AUTHORS names Dickey on ten pages
    2. Opened: README:171-176; the AUTHORS sections of every file in man/; PR 63 review on site/src/content/docs/index.mdx
    3. It says: “README: Eric S. Raymond: the man pages. AUTHORS: Thomas Dickey in curs_extend.3x and nine other pages”
    4. Verdict: disagrees, high confidence
    5. One step earlier: AUTHORS alone made the Raymond clause look wrong; the card’s job is the maintainer’s role relative to the original writer