Support ncurses extended colors API by shugo · Pull Request #130 · ruby/curses

shugo · 2026-03-06T03:08:05Z

Support ncurses extended colors API

Use init_extended_pair/init_extended_color/extended_color_content/
extended_pair_content when available (ncurses 6+), removing the 256
color pair limitation. Window#color_set is upgraded to use wattr_set
internally when available, allowing pair numbers > 255.

New methods:

Curses.support_extended_colors? — runtime check for extended support
Curses.reset_color_pairs — reset all pairs to undefined (ncurses 6.1+)

Use init_extended_pair/init_extended_color/extended_color_content/ extended_pair_content when available (ncurses 6+), removing the 256 color pair limitation. Window#color_set is upgraded to use wattr_set internally when available, allowing pair numbers > 255. New methods: - Curses.support_extended_colors? — runtime check for extended support - Curses.reset_color_pairs — reset all pairs to undefined (ncurses 6.1+) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

When Curses.support_extended_colors? returns true, display up to 512 color pairs (capped at color_pairs) using stdscr.color_set instead of attrset(color_pair()), since COLOR_PAIR() cannot encode pair numbers > 255. Pairs 256-511 use a dark grey background to distinguish them from the first 256. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Copilot

Pull request overview

This PR updates the Ruby curses extension to take advantage of ncurses’ extended colors API (ncurses 6+) when available, lifting the historical 256 color-pair limitation and updating Window#color_set behavior to better support large pair numbers.

Changes:

Prefer init_extended_pair / init_extended_color and extended content getters when available.
Add Curses.support_extended_colors? and Curses.reset_color_pairs.
Update sample + README to reflect/support extended color pairs usage.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 7 comments.

File	Description
`sample/colors.rb`	Demonstrates extended color-pair usage and avoids `color_pair()` for pairs > 255.
`ext/curses/extconf.rb`	Detects extended-color-related ncurses functions/macros at build time.
`ext/curses/curses.c`	Implements extended-color-aware wrappers and adds new Ruby-facing APIs.
`README.md`	Updates limitation docs to reflect extended color support on ncurses 6+.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-03-06T03:12:48Z

ext/curses/curses.c

+static VALUE
+curses_reset_color_pairs(VALUE obj)
+{
+    reset_color_pairs();


reset_color_pairs() returns a status code (OK/ERR). The wrapper currently ignores the return value and always returns nil, making failures silent. Consider returning true/false (consistent with other wrappers) or raising on ERR.

Suggested change

reset_color_pairs();

int ret = reset_color_pairs();

if (ret == ERR) {

rb_sys_fail("reset_color_pairs");

}

Copilot · 2026-03-06T03:12:48Z

ext/curses/curses.c

+	extended_color_content(NUM2INT(color), &r, &g, &b);
+	return rb_ary_new3(3, INT2FIX(r), INT2FIX(g), INT2FIX(b));
+    }
+#else
+    {
+	short r, g, b;
+	color_content(NUM2INT(color), &r, &g, &b);


extended_color_content/color_content return OK/ERR, but the code ignores the return value and then returns potentially uninitialized r/g/b when the call fails. Handle the return code and raise/return nil on ERR to avoid undefined behavior.

Suggested change

extended_color_content(NUM2INT(color), &r, &g, &b);

return rb_ary_new3(3, INT2FIX(r), INT2FIX(g), INT2FIX(b));

}

#else

{

short r, g, b;

color_content(NUM2INT(color), &r, &g, &b);

int rc = extended_color_content(NUM2INT(color), &r, &g, &b);

if (rc == ERR) {

return Qnil;

}

return rb_ary_new3(3, INT2FIX(r), INT2FIX(g), INT2FIX(b));

}

#else

{

short r, g, b;

int rc = color_content(NUM2INT(color), &r, &g, &b);

if (rc == ERR) {

return Qnil;

}

Copilot · 2026-03-06T03:12:48Z

ext/curses/curses.c

+	extended_pair_content(NUM2INT(pair), &f, &b);
+	return rb_ary_new3(2, INT2FIX(f), INT2FIX(b));
+    }
+#else
+    {
+	short f, b;
+	pair_content(NUM2INT(pair), &f, &b);


extended_pair_content/pair_content return OK/ERR, but the code ignores the return value and then returns potentially uninitialized f/b when the call fails. Handle the return code and raise/return nil on ERR to avoid undefined behavior.

Suggested change

extended_pair_content(NUM2INT(pair), &f, &b);

return rb_ary_new3(2, INT2FIX(f), INT2FIX(b));

}

#else

{

short f, b;

pair_content(NUM2INT(pair), &f, &b);

int rc = extended_pair_content(NUM2INT(pair), &f, &b);

if (rc == ERR) return Qnil;

return rb_ary_new3(2, INT2FIX(f), INT2FIX(b));

}

#else

{

short f, b;

int rc = pair_content(NUM2INT(pair), &f, &b);

if (rc == ERR) return Qnil;

Copilot · 2026-03-06T03:12:49Z

sample/colors.rb

+    addstr extended ? " (extended).\n" : ".\n"
+
+    (extended ? [512, color_pairs].min : colors).times { |i|
+      Curses.init_pair(i, i%256, i>=256 ? 8 : 0)


In the extended branch, init_pair(i, i%256, ...) can pass invalid color numbers when colors < 256 (e.g., a 16-color terminal built against an extended-capable ncurses). Consider using i % colors (and choosing a background within range) so the sample works on low-color terminals too.

Suggested change

Curses.init_pair(i, i%256, i>=256 ? 8 : 0)

fg_color = i % colors

bg_color = (i / colors) % colors

Curses.init_pair(i, fg_color, bg_color)

Copilot · 2026-03-06T03:12:49Z

sample/colors.rb

+    addstr "This Terminal supports #{colors} colors, #{color_pairs} pairs"
+    addstr extended ? " (extended).\n" : ".\n"
+
+    (extended ? [512, color_pairs].min : colors).times { |i|


The loop initializes color pair 0 (Curses.init_pair(0, ...)). In ncurses, pair 0 is the default pair and typically cannot be redefined (init_pair(0, ...) returns ERR). Consider starting the loop at 1 (or skipping 0) to avoid silent errors in this sample.

Suggested change

(extended ? [512, color_pairs].min : colors).times { |i|

(extended ? [512, color_pairs].min : colors).times { |i|

next if i == 0

Copilot · 2026-03-06T03:12:49Z

ext/curses/curses.c

+#ifdef HAVE_WATTR_SET
+    return (wattr_set(winp->window, 0, NUM2INT(col), NULL) == OK) ? Qtrue : Qfalse;
+#else
+    return (wcolor_set(winp->window, NUM2INT(col), NULL) == OK) ? Qtrue : Qfalse;


Window#color_set now uses wattr_set(..., 0, pair, ...) when available. Passing attrs = 0 will clear any existing window attributes (bold/underline/etc.), which changes behavior compared to wcolor_set (and to the method’s intent of only changing the color pair). Preserve the current attributes when switching pairs (e.g., fetch current attrs via wattr_get if available, or prefer wcolor_set when it exists and only fall back to wattr_set when needed).

Suggested change

#ifdef HAVE_WATTR_SET

return (wattr_set(winp->window, 0, NUM2INT(col), NULL) == OK) ? Qtrue : Qfalse;

#else

return (wcolor_set(winp->window, NUM2INT(col), NULL) == OK) ? Qtrue : Qfalse;

#ifdef HAVE_WCOLOR_SET

/* Prefer wcolor_set when available, as it only changes the color pair. */

return (wcolor_set(winp->window, NUM2INT(col), NULL) == OK) ? Qtrue : Qfalse;

#elif defined(HAVE_WATTR_SET)

# ifdef HAVE_WATTR_GET

/*

* When wcolor_set is unavailable but wattr_set/wattr_get are,

* preserve existing attributes and only change the color pair.

*/

attr_t attrs;

short current_pair;

if (wattr_get(winp->window, &attrs, &current_pair, NULL) == ERR) {

return Qfalse;

}

return (wattr_set(winp->window, attrs, NUM2INT(col), NULL) == OK) ? Qtrue : Qfalse;

# else

/*

* Fallback: wcolor_set and wattr_get are unavailable; use wattr_set

* with attrs = 0 (may clear attributes, but there is no way to

* query and preserve them portably here).

*/

return (wattr_set(winp->window, 0, NUM2INT(col), NULL) == OK) ? Qtrue : Qfalse;

# endif

Copilot · 2026-03-06T03:12:49Z

ext/curses/curses.c

+#ifdef HAVE_RESET_COLOR_PAIRS
+static VALUE
+curses_reset_color_pairs(VALUE obj)
+{


Curses.reset_color_pairs should follow the pattern of other color APIs by calling curses_stdscr() before invoking ncurses functions. As written, calling it before init_screen can behave inconsistently vs. the rest of the API.

Suggested change

{

{

curses_stdscr();

- curses_color_content, curses_pair_content: check return value of extended_color_content/color_content/extended_pair_content/pair_content and return nil on ERR instead of returning uninitialized values - curses_reset_color_pairs: add curses_stdscr() call for consistency with other color APIs - window_color_set: use wattr_get to preserve existing window attributes when calling wattr_set; use NCURSES_PAIRS_T for the pair argument to match the ncurses header type in both standard and extended modes; fall back to wattr_set without attr preservation if wattr_get is unavailable, and to wcolor_set if wattr_set is also unavailable - extconf.rb: add have_func check for wattr_get - sample/colors.rb: skip pair 0 (cannot be redefined); use i%colors instead of i%256 to handle terminals with fewer than 256 colors Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

NCURSES_PAIRS_T is ncurses-specific and not defined in PDCurses. Fall back to short when it is not available. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

shugo and others added 2 commits March 6, 2026 11:42

shugo requested a review from Copilot March 6, 2026 03:08

Copilot started reviewing on behalf of shugo March 6, 2026 03:09 View session

Copilot AI reviewed Mar 6, 2026

View reviewed changes

shugo and others added 2 commits March 6, 2026 12:25

Fix build on Windows/PDCurses: guard NCURSES_PAIRS_T

c408e11

NCURSES_PAIRS_T is ncurses-specific and not defined in PDCurses. Fall back to short when it is not available. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

shugo merged commit ffcc80f into master Mar 6, 2026
8 checks passed

shugo deleted the feature/extended-colors branch March 6, 2026 04:14

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Support ncurses extended colors API#130

Support ncurses extended colors API#130
shugo merged 4 commits intomasterfrom
feature/extended-colors

shugo commented Mar 6, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Mar 6, 2026

Uh oh!

Copilot AI Mar 6, 2026

Uh oh!

Copilot AI Mar 6, 2026

Uh oh!

Copilot AI Mar 6, 2026

Uh oh!

Copilot AI Mar 6, 2026

Uh oh!

Copilot AI Mar 6, 2026

Uh oh!

Copilot AI Mar 6, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

-    reset_color_pairs();
+    int ret = reset_color_pairs();
+    if (ret == ERR) {
+        rb_sys_fail("reset_color_pairs");
+    }

-      Curses.init_pair(i, i%256, i>=256 ? 8 : 0)
+      fg_color = i % colors
+      bg_color = (i / colors) % colors
+      Curses.init_pair(i, fg_color, bg_color)

	(extended ? [512, color_pairs].min : colors).times { \|i\|
	(extended ? [512, color_pairs].min : colors).times { \|i\|
	next if i == 0

-#ifdef HAVE_WATTR_SET
-    return (wattr_set(winp->window, 0, NUM2INT(col), NULL) == OK) ? Qtrue : Qfalse;
-#else
-    return (wcolor_set(winp->window, NUM2INT(col), NULL) == OK) ? Qtrue : Qfalse;
+#ifdef HAVE_WCOLOR_SET
+    /* Prefer wcolor_set when available, as it only changes the color pair. */
+    return (wcolor_set(winp->window, NUM2INT(col), NULL) == OK) ? Qtrue : Qfalse;
+#elif defined(HAVE_WATTR_SET)
+# ifdef HAVE_WATTR_GET
+    /*
+     * When wcolor_set is unavailable but wattr_set/wattr_get are,
+     * preserve existing attributes and only change the color pair.
+     */
+    attr_t attrs;
+    short current_pair;
+    if (wattr_get(winp->window, &attrs, &current_pair, NULL) == ERR) {
+        return Qfalse;
+    }
+    return (wattr_set(winp->window, attrs, NUM2INT(col), NULL) == OK) ? Qtrue : Qfalse;
+# else
+    /*
+     * Fallback: wcolor_set and wattr_get are unavailable; use wattr_set
+     * with attrs = 0 (may clear attributes, but there is no way to
+     * query and preserve them portably here).
+     */
+    return (wattr_set(winp->window, 0, NUM2INT(col), NULL) == OK) ? Qtrue : Qfalse;
+# endif

Conversation

shugo commented Mar 6, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Mar 6, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 6, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 6, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 6, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 6, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 6, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 6, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants