A tool that turns test results into a report a release can carry - #251
Conversation
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: faaa50a2ae
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| name, _, directory = spec.partition("=") | ||
| if not name or not directory: | ||
| parser.error(f"--module wants NAME=DIR, got {spec!r}") | ||
| modules[name] = Path(directory) |
There was a problem hiding this comment.
When --module is supplied twice with the same name, this assignment silently replaces the earlier directory. The report then counts only the final directory and exits successfully, so a copy/paste error such as two app=... arguments can produce an incomplete release artifact without warning. Reject duplicate names or explicitly combine their results instead of overwriting them.
Useful? React with 👍 / 👎.
The site's new "How Folio is tested" page can state test counts, but only if something measures them. Counts typed into docs go stale (docs/standards/testing.md already says ~476 tests; the notes say well over 800).
What is true now:
Checked: the 5 tests pass. On the old partial results in my working tree it counted 221 tests in 54 suites and exited 0; pointed at a directory that does not exist it refused and exited 1.
Not done on purpose: it is not wired into scripts/release-signed.sh or CI. That is the release path, and it should not change right before the 0.6.8 candidate. After 0.6.8, the release script can run it and attach test-report.json to the release, and the site page will pick it up by itself. It changes nothing that ships, so no changelog line.