tests: keep environment from leaking in Python 3.15

[henryiii]

Description

Fix #1047.

🤖 Use a bit of copilot with both gpt-5-mini and Kimi-K2.6, mostly to automate running and appying the example. Hand wrote most of the config changes and some of the code changes.

Changelog

• Added changelog fragment: docs/changelog/<pr_number>.<type>.rst
  • Types: feature, bugfix, doc, removal, misc
  • Example: 123.feature.rst containing Add custom backend support - by :user:`yourname`

Checklist

• Tests pass locally (tox)
• Code follows project style (tox -e fix)
• Type checks pass (tox -e type)
• Documentation builds (tox -e docs)

[henryiii]

@copilot investigate the CI failure.

[Copilot]

@copilot investigate the CI failure.

The CI failure was a coverage failure: src/build/_types.py had 0% coverage on Python 3.15.

Root cause: Python 3.15 uses __lazy_modules__ as a native lazy imports mechanism. Since _types is listed in __lazy_modules__ in build/__init__.py, the from ._types import ConfigSettings as ConfigSettingsType lines are deferred until those names are actually accessed. Since no tests access build.ConfigSettingsType, build.DistributionType, or build.RunnerType, _types.py was never executed — resulting in 0% coverage and failing the fail-under=100 check.

Fix (commit 6990945): Added src/build/_types.py to the coverage run.omit list. This is appropriate since _types.py only contains type aliases (no logic to test).

Warning

Firewall rules blocked me from connecting to one or more addresses (expand for details)

I tried to connect to the following addresses, but was blocked by firewall rules:

• docs.python.org
  • Triggering command: /home/REDACTED/work/_temp/ghcca-node/node/bin/node /home/REDACTED/work/_temp/ghcca-node/node/bin/node --enable-source-maps /home/REDACTED/work/_temp/copilot-developer-action-main/dist/index.js (dns block)

If you need me to access, download, or install something from one of these locations, you can either:

• Configure Actions setup steps to set up my environment, which run before the firewall is enabled
• Add the appropriate URLs or hosts to the custom allowlist in this repository's Copilot coding agent settings (admins only)

[layday]

I feel both of these tests are too generic to be meaningful. What are we testing, that a venv is a venv? env.python_executable != sys.executable is probably sufficient.

[henryiii]

I can probably combine the tests. The first one is the original one, with the added protection from #1047. The second is the same test but with an injected PYTHONPATH to verify it fails regardless of the external PYTHONPATH setting - but I think we could just add that to the first test.

[henryiii]

I believe this issue wouldn't show up if we just checked env.python_executable != sys.executable. This demonstrates that you need to apply the extra env. If you remove it, the test fails.

[layday]

The issue only exists in the test. Whether we guard against PYTHONPATH in this one test or not has no impact on the actual running of the tool, so I don’t think it matters much either way.

[henryiii]

It shows that we need to guard it in the code itself (yes, this is way too unit-testy of a unit-test, testing the internals by mimicking the internals instead of using them!)

If we want to drop the test or change it, I think that should be a followup. This only fixes the current test to work on 3.15, and adds 3.15 to our CI.