fix(ext/node): add hasColors() to process.stdout and process.stderr by ddmoney420 · Pull Request #31985 · denoland/deno

ddmoney420 · 2026-01-29T15:51:10Z

Summary

Adds hasColors() method to process.stdout and process.stderr
Fixes Node.js compatibility issue where packages like Commander.js check color support

Fixes #27278

Changes

Modified ext/node/polyfills/_process/streams.mjs to add hasColors property
Added tests in tests/unit_node/process_test.ts

Implementation

The hasColors method mirrors the logic from tty.WriteStream.prototype.hasColors():

When called with no arguments or an object, defaults to checking for 16 colors
Uses op_bootstrap_color_depth() to get the current color depth
Returns true if count <= 2 ** depth

Test plan

Added unit tests for process.stdout.hasColors
Added unit tests for process.stderr.hasColors
Verify Commander.js color detection works correctly

🤖 Generated with Claude Code

CLAassistant · 2026-01-29T15:51:19Z

All committers have signed the CLA.

bartlomieju · 2026-01-29T17:34:55Z

@Tango992 can you review this one? Ideally instead of adding a custom unit test we could enable test-tty-color-support.js test instead.

…tion This adds the missing `hasColors()` method to `process.stdout` and `process.stderr` streams, matching Node.js behavior. The implementation includes proper validation that throws `ERR_INVALID_ARG_TYPE` for invalid input types and `ERR_OUT_OF_RANGE` for invalid count values. Additionally, this implements the full `getColorDepth(env)` function from Node.js in tty.js, which detects color support based on environment variables like FORCE_COLOR, COLORTERM, TERM, and various CI environments (GitHub Actions, GitLab CI, etc.). Note: The Node.js test file `test-tty-color-support.js` cannot be enabled yet because it tries to redefine `process.platform`, which is not configurable in Deno. A TODO comment has been added to track this. Fixes denoland#27278 Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

ddmoney420 · 2026-01-29T19:42:30Z

Thanks for the suggestion! I looked into enabling test-tty-color-support.js but there's a blocker noted in the config:

// TODO: Enable once process.platform is configurable.
// The test tries to redefine process.platform which Deno doesn't support.
// "pseudo-tty/test-tty-color-support.js": {},

The test at lines 109-118 does:

Object.defineProperty(process, 'platform', { value });

Since process.platform is not configurable in Deno, this section would fail. The rest of the test (hasColors/getColorDepth behavior with env vars) would pass, but we'd need to either:

Make process.platform configurable in Deno (separate change)
Fork/modify the test to skip the OS settings section

Would you prefer I take one of these approaches, or is the unit test approach acceptable for now?

bartlomieju · 2026-01-29T20:21:02Z

Changing process.platform to be writable sounds good. You can do it in this PR.

- Make process.platform configurable and writable so tests can override it via Object.defineProperty() - Enable test-tty-color-support.js test which was previously disabled pending this change - Remove custom unit tests in favor of the Node.js test Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

ddmoney420 · 2026-01-29T21:05:42Z

Done! I've made process.platform configurable and writable, enabled test-tty-color-support.js, and removed the custom unit tests in favor of the Node.js test.

Changes:

Added set(value) and configurable: true to the process.platform property definition
Enabled pseudo-tty/test-tty-color-support.js test (removed the TODO comment)
Removed the custom hasColors unit tests

Replace raw RegExp literals with SafeRegExp from primordials to fix the prefer-primordials lint error. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

- Use SafeMapIterator for CI_ENVS_MAP iteration - Replace 'in' operator with ObjectPrototypeHasOwnProperty - Replace remaining RegExp literals with SafeRegExp Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

Tango992

Thanks, great work! Only some minor reviews. Also please address the lint errors so we could run the tests on CI

Tango992 · 2026-01-30T10:26:14Z

ext/node/polyfills/tty.js

@@ -1,19 +1,246 @@
 // Copyright 2018-2026 the Deno authors. MIT license.
+// Copyright (c) Sindre Sorhus <sindresorhus@gmail.com> (sindresorhus.com). MIT license.


While we're at it, let's copy the whole license as defined here:
https://github.com/nodejs/node/blob/70f6b58ac655234435a99d72b857dd7b316d34bf/lib/internal/tty.js#L5-L21

Tango992 · 2026-01-30T10:28:23Z

ext/node/polyfills/tty.js

+    if (+OSRelease[0] >= 10) {
+      const build = +OSRelease[2];
+      if (build >= 14931) {
+        return COLORS_16m;
+      }
+      if (build >= 10586) {
+        return COLORS_256;
+      }
+    }


I feel like we should add this comment to give people some context:
https://github.com/nodejs/node/blob/70f6b58ac655234435a99d72b857dd7b316d34bf/lib/internal/tty.js#L160-L162

ddmoney420 · 2026-02-01T18:35:14Z

Good catch on the license header. Added the full MIT license text.

ddmoney420 · 2026-02-02T17:31:29Z

Closing this PR as part of a cleanup - I opened too many PRs across projects and want to be more respectful of maintainer time. I also used AI assistance without proper disclosure. Apologies for the noise, and thank you for your patience.

bartlomieju · 2026-02-02T17:38:12Z

@ddmoney420 this PR looks okay, I think we could land it with no problem

ddmoney420 · 2026-02-02T17:44:10Z

Thanks @bartlomieju! Reopening - I closed this prematurely as part of cleaning up some other noisy PRs. Happy to address any remaining lint errors if needed.

- Remove unused import `op_bootstrap_color_depth` - Rename COLORS_16m to COLORS_16M (camelcase lint rule)

Tango992 · 2026-02-03T07:03:38Z

Seems like the windows test still fails on this line:

deno/tests/unit_node/tty_test.ts

Line 43 in 0b38d25

assert(tty.WriteStream.prototype.hasColors({}) === !Deno.noColor);

Would you please take a look @ddmoney420 ? I'll try to investigate it later when I have access to my windows machine

ddmoney420 · 2026-02-03T20:53:23Z

@Tango992 I looked into this — the test on line 43 expects hasColors({}) === !Deno.noColor, but these measure different things:

hasColors({}) defaults to checking for 16 colors
!Deno.noColor just checks if colors aren't disabled

On Windows CI with limited color depth, hasColors({}) returns false (no 16-color support) but !Deno.noColor is true (colors not disabled).

Should I change the test to use hasColors(2) instead? That would check for basic color support rather than 16 colors.

passing empty object as parameter will lead to different result across OS's

Tango992 · 2026-02-04T04:02:16Z

Should I change the test to use hasColors(2) instead? That would check for basic color support rather than 16 colors.

That’s fine, I’ve committed a change for the tty test.

Passing an empty object will return a different result in windows because we got caught on this if statement:

https://github.com/denoland/deno/pull/31985/changes#diff-c73716894df6cd0c109864a4253df672d4dd8ec8463853c50508353c56ef1dbdR161

Given this code and run it with Node.js:

import tty from “node:tty”;
import { release } from “node:os”;

process.env.NO_COLOR = “1”;
console.log(tty.WriteStream.prototype.hasColors(), tty.WriteStream.prototype.hasColors({}))

Linux & macos outputs false false, while on windows it outputs false true

Tango992

LGTM

bartlomieju requested a review from Tango992 January 29, 2026 17:34

ddmoney420 force-pushed the fix/process-stdout-hascolors branch from 1ea9dce to 1f99550 Compare January 29, 2026 18:53

ddmoney420 and others added 2 commits January 29, 2026 17:37

fix: use SafeRegExp for primordials lint compliance

8dd98e6

Replace raw RegExp literals with SafeRegExp from primordials to fix the prefer-primordials lint error. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

fix: use primordials consistently in getColorDepth

e8a2d4c

- Use SafeMapIterator for CI_ENVS_MAP iteration - Replace 'in' operator with ObjectPrototypeHasOwnProperty - Replace remaining RegExp literals with SafeRegExp Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

Tango992 suggested changes Jan 30, 2026

View reviewed changes

chore: add full MIT license header for Sindre Sorhus code

38b6baf

style: fix dprint formatting

542315f

ddmoney420 closed this Feb 2, 2026

ddmoney420 reopened this Feb 2, 2026

fix: address lint errors in tty.js

c7720db

- Remove unused import `op_bootstrap_color_depth` - Rename COLORS_16m to COLORS_16M (camelcase lint rule)

ddmoney420 mentioned this pull request Feb 2, 2026

Resize event on stdout not invoked in TTY on macOS #26671

Closed

fix: update hasColors test to use color depth of 2

7f3ef97

Tango992 added 2 commits February 4, 2026 09:52

test: update hasColors test to check for NO_COLOR environment variable

bd8300a

passing empty object as parameter will lead to different result across OS's

fix: lint

632cb7e

Tango992 approved these changes Feb 4, 2026

View reviewed changes

Tango992 changed the title ~~fix(node): add hasColors() to process.stdout and process.stderr~~ fix(ext/node): add hasColors() to process.stdout and process.stderr Feb 4, 2026

Tango992 merged commit 9aca4f6 into denoland:main Feb 4, 2026
22 checks passed

		@@ -1,19 +1,246 @@
		// Copyright 2018-2026 the Deno authors. MIT license.
		// Copyright (c) Sindre Sorhus <sindresorhus@gmail.com> (sindresorhus.com). MIT license.

Conversation

ddmoney420 commented Jan 29, 2026

Summary

Changes

Implementation

Test plan

Uh oh!

CLAassistant commented Jan 29, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

bartlomieju commented Jan 29, 2026

Uh oh!

ddmoney420 commented Jan 29, 2026

Uh oh!

bartlomieju commented Jan 29, 2026

Uh oh!

ddmoney420 commented Jan 29, 2026

Uh oh!

Tango992 left a comment

Choose a reason for hiding this comment

Uh oh!

Tango992 Jan 30, 2026

Choose a reason for hiding this comment

Uh oh!

Tango992 Jan 30, 2026

Choose a reason for hiding this comment

Uh oh!

ddmoney420 commented Feb 1, 2026

Uh oh!

ddmoney420 commented Feb 2, 2026

Uh oh!

bartlomieju commented Feb 2, 2026

Uh oh!

ddmoney420 commented Feb 2, 2026

Uh oh!

Tango992 commented Feb 3, 2026

Uh oh!

ddmoney420 commented Feb 3, 2026

Uh oh!

Tango992 commented Feb 4, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Tango992 left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

CLAassistant commented Jan 29, 2026 •

edited

Loading

Tango992 commented Feb 4, 2026 •

edited

Loading