Skip to content

release: 0.1.2 - #35

Merged
ehtishammubarik merged 1 commit into
masterfrom
release/0.1.2
Sep 20, 2026
Merged

ehtishammubarik merged 1 commit into
masterfrom
release/0.1.2

Conversation

@ehtishammubarik

Copy link
Copy Markdown
Owner

PyPI has been on 0.1.0 since 2026-07-29 while master carried nine merged changes. This unblocks that.

What happened to 0.1.1

Tagged 2026-09-04. The release run failed at test-wheel on 3.10, 3.11 and 3.12 and published nothing, not to TestPyPI and not to PyPI. The gating worked exactly as designed; the thing it gated on was wrong.

The cause was in the workflow, not the package:

assert "publicly_accessible = false" in prod    # release.yml:170, single space

create_db_subnet_group is 22 characters, joined that block in #13, and terraform fmt realigned the attribute to four spaces. The workflow had been red since 2026-07-31 and nothing noticed, because it runs only on a tag push and no tag was cut in that window. #33 fixed it in 4e862a1.

The generated output was correct throughout: the database was private, prod deletion_protection was true.

Why 0.1.2 rather than re-tagging 0.1.1

v0.1.1 stays where it is, as a git tag pointing at a commit that was never published. It records that a release was attempted and refused, which is a true thing that happened. Moving or deleting it would make the history tidier and less accurate, and this repo is not in the business of that trade.

The changelog says so plainly, under 0.1.1 own heading:

## [0.1.1] - 2026-09-04 Tagged, but never published. See the 0.1.2 entry above. Everything described below shipped in 0.1.2.

Package contents are identical to what 0.1.1 would have published, plus the workflow fix.

Verification

pytest -q      91 passed
coverage       90.90%, floor 88 met
ruff check     All checks passed
secret-scan    PASS
git-hygiene    PASS

The real test is the release run itself, which is what failed last time and which this PR exists to let succeed.

🤖 Generated with Claude Code

https://claude.ai/code/session_012fMM2HhoJ2XzDkfMR2HwSa

0.1.1 was tagged on 2026-09-04 and never published. Its release run failed at
test-wheel on all three Pythons and correctly published nothing, to TestPyPI or
PyPI, so PyPI has sat on 0.1.0 for sixteen days while master carried nine
merged changes.

The cause was in the workflow, not the package. It asserted
`publicly_accessible = false` with a single space; create_db_subnet_group
joined that block in #13 and terraform fmt realigned it to four. Red since
2026-07-31, unnoticed, because a workflow that runs only on a tag push has no
feedback loop between tags. #33 fixed it in 4e862a1.

Package contents are identical to what 0.1.1 would have published, plus the
workflow fix.

v0.1.1 is left in place as a git tag rather than deleted or moved. It records
that a release was attempted and refused, which is a true thing that happened,
and the changelog says so under its own heading rather than quietly
renumbering.

Verified: pytest 91 passed, coverage 90.90% against the 88% floor, ruff clean.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012fMM2HhoJ2XzDkfMR2HwSa
@ehtishammubarik
ehtishammubarik merged commit 6b5d832 into master Sep 20, 2026
10 checks passed
@ehtishammubarik
ehtishammubarik deleted the release/0.1.2 branch September 20, 2026 16:43
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant