Fix Ruby 4 CGI flaky test

[MikeMcQuaid]

• Ignore the compatibility warning emitted when plist lazily loads CGI.
• Keep unexpected warnings visible to stderr expectations.

---

• Have you followed our Contributing guidelines?
• Have you checked for other open Pull Requests for the same change?
• Have you explained what your changes do? Performance claims (e.g. "this is faster") must include Hyperfine benchmarks.
• Have you explained why you'd like these changes included, not just what they do?
• For bug fixes, have you given step-by-step brew commands to reproduce the bug?
• Have you written new tests (excluding integration tests)? Here's an example.
• Have you successfully run brew lgtm (style, typechecking and tests) locally?

---

• I did not use AI/LLM to create this PR, or I disclosed the tool/model below and reviewed its output; I did not attribute commits to AI and will answer maintainer questions and review comments myself without AI/LLM.

OpenAI GPT 5.6 sol xhigh with local review and testing.

---

[Copilot]

Pull request overview

This PR aims to stabilise the test suite under Ruby 4 by filtering out the specific CGI removal compatibility warning that can be emitted when plist ends up loading CGI, while keeping other unexpected warnings visible so stderr-based expectations still catch regressions.

Changes:

• Add a :cgi warning category pattern to Warnings::COMMON_WARNINGS.
• Add a unit test asserting the Ruby 4 CGI warning can be ignored.
• Wrap each RSpec example run in Warnings.ignore(:cgi) in spec_helper.rb to prevent flaky failures caused by that warning.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

File	Description
Library/Homebrew/warnings.rb	Adds a new common warning category (:cgi) to be ignored via Warnings.ignore.
Library/Homebrew/test/warnings_spec.rb	Adds a spec ensuring the Ruby 4 CGI warning is suppressed when ignored.
Library/Homebrew/test/spec_helper.rb	Ignores the :cgi warning around each example execution to avoid Ruby 4 flakiness.

---

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

[dduugg]

This is a genuine narrowly-scoped fix rather than a papered-over incompatibility, which is what I wanted to check given the branch name. Reviewing the current head, which no longer touches warnings.rb, so the earlier comment on that file looks superseded.

I traced the cause: vendor/portable-ruby/4.0.6/lib/ruby/4.0.0/cgi.rb is now a shim doing warn <<-WARNING, uplevel: Gem::BUNDLED_GEMS.uplevel if $VERBOSE, routed through Warning.warn, so the prepended Warnings::Filter is the only thing that can suppress it. The regex is anchored to that exact message rather than being a broad filter, so it will not swallow unrelated warnings, and warnings_spec.rb keeps its existing coverage. Production is unaffected either way, since brew.sh runs with HOMEBREW_RUBY_WARNINGS=-W1 and the shim only fires when $VERBOSE is set.

Two optional notes, neither blocking.

The suppression is thread-local

Library/Homebrew/test/spec_helper.rb:353

Warnings uses Thread.current.thread_variable_set, and thread variables are not inherited by child threads (confirmed against the vendored 4.0.6). So a first require "plist" reached from a thread spawned by Utils.parallel_map, via the SystemConfig.dump_verbose_config to host_software_config to MacOS::Xcode.detect_version route, would not be covered.

Nothing fails today, because every spec that reaches dump_verbose_config stubs the section that would pull in plist (test/cmd/config_spec.rb:86, :104 and :115). So this only bites if someone later adds a spec exercising an unstubbed dump_verbose_config on macOS.

The load-time idiom already used at spec_helper.rb:21 would close it for free:

Warnings.ignore(/CGI library is removed from Ruby 4\.0\./) { require "cgi" }

The require marks cgi loaded, so plist's later require "cgi" is a no-op and the warning cannot fire again.

The new example is a tautology with respect to the regex

Library/Homebrew/test/warnings_spec.rb:8

The example emits its own copy of the message, so by construction it only asserts that the literal and the regex agree. If the upstream shim reworded the text, spec_helper.rb:353 would stop matching the real cgi.rb heredoc while this example stayed green.

It is not worthless, since it fails loudly if someone deletes or reorders the Warnings.ignore wrapper, which is the realistic regression for an 8-line harness change. And the drift risk is bounded, because the message lives in pinned portable-ruby and can only change on a deliberate bump.

If you want it to check wording against pattern, shelling out to HOMEBREW_RUBY_PATH -w -e 'require "cgi"' and asserting its stderr matches the same regex would do it. Otherwise it might belong outside RSpec.describe Warnings, since it asserts spec_helper behaviour rather than the Warnings API and is the only example in the file not touching described_class.