Repository navigation
chore: give the fork its own identity and version namespace - #7
Merged
Merged
Conversation
The wheel described itself as upstream's: urls pointed at lilab-bcb, and the version claimed a string upstream can publish. Upstream has 86 .postN tags on PyPI and has already released 1.1.61, so cirrocumulus 1.1.61.post1 was theirs to cut while we had published a wheel claiming it -- same name, same version, different bytes. - Version as a PEP 440 local identifier, 1.1.61+quadbio.N. Local versions are the mechanism for "upstream's release plus our patches", and PyPI must not accept them, so ours cannot collide. tag_regex keeps the local segment that setuptools_scm's default would drop; check_wheel.py fails the build if it ever goes missing. - Point Homepage/Source at quadbio, record upstream under its own key, add a maintainer, and put a fork notice atop README.rst (the wheel's long description). - Drop the stray `dev = [...]` key upstream left inside [tool.hatch.version]. Verified: on the tag -> 1.1.61+quadbio.N; one commit past -> 1.1.62.dev1+quadbio.N.g<sha>. Both keep the marker. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BQnnW5jCKXnefXA89eBDoL
Marius1311
added a commit
that referenced
this pull request
Aug 27, 2026
#7's tag_regex matched only `quadbio-*`, so setuptools_scm raised `Can't parse version from tag '1.1.61'` on upstream's tags -- which are the nearest ones reachable from main, since our two release tags were orphaned by the old force-pushed integration branch. That broke release.yml on every push to main. Make the prefix optional so upstream's tags parse as a fallback, and only require the +quadbio. segment on release builds: a build off an untagged main is a .devN artifact that is never published and has no local segment to carry. Claude-Session: https://claude.ai/code/session_01BQnnW5jCKXnefXA89eBDoL Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Replaces #6, which GitHub auto-closed when its base branch (
chore/branch-model) was deleted onmerge. Same content, rebased onto
main.The wheel currently describes itself as upstream's, and its version claims a string upstream can
publish at any time.
Version → PEP 440 local identifier,
1.1.61+quadbio.N. Upstream has 86.postNtags, shipsthem to PyPI, and has already released
1.1.61— socirrocumulus 1.1.61.post1is theirs to cut,while we have a published wheel claiming it. Same name, same version, different bytes; anything
resolving
cirrocumulusfrom an index can swap our fork for upstream's. Local versions are thedocumented mechanism for "upstream's release plus our patches", and PyPI must not accept them,
which is what makes ours unforgeable.
tag_regexinpyproject.tomlis required: setuptools_scm's default captures[^\+]*and wouldsilently drop the local segment. Verified empirically:
quadbio-1.1.61+quadbio.31.1.61+quadbio.31.1.62.dev1+quadbio.3.gc0cc180Both keep the marker, so
local_schemeneeds no override — settingno-local-versionwould havedropped
+quadbio.3from the tagged build.check_wheel.pynow fails if+quadbio.is ever missing, so a bad derivation breaks CI insteadof shipping.
Identity —
Homepage/Sourcepoint at quadbio, upstream gets its own key, a maintainer isadded, and
README.rst(the wheel's long description) opens with a fork notice. Also drops thestray
dev = ['pre-commit', 'isort', 'black']that upstream left sitting inside[tool.hatch.version].Licence-wise we were already fine: BSD-3 requires retaining the copyright notice, which
license = { file = 'LICENSE' }does, and has no modification-notice clause. This is a routingand collision fix, not a compliance one.