Repository navigation
[bot] Cut a semsql release including the semanticsql.berkeleybop.io default #114
Description
Activity
Summary
This is a straightforward "cut a release" ask, already assigned to @caufieldjh. Context on the linked issues, plus a couple of things worth flagging for the release itself:
Status of linked issues
- Provide sqlite via a different URL #110 (Provide sqlite via a different URL) — closed, completed via PR Download SQLite databases from the semanticsql CDN instead of S3 #113, which repointed the
semsql downloaddefault tohttps://semanticsql.berkeleybop.ioand added a non-urllibUser-Agent (the CDN 403s on the defaultPython-urllibUA). - Eliminate S3 bucket access #112 (Eliminate S3 bucket access) — open, tracks turning off public access to the
bbop-sqliteS3 bucket once enough traffic has migrated to the CDN. This release is a prerequisite:semsqlcan't drive traffic off S3 until the CDN default actually ships to PyPI. - Reduce S3 egress costs for bbop-sqlite database downloads ontology-access-kit#880 (Reduce S3 egress costs for bbop-sqlite database downloads) — open, OAK-side companion effort. Its "Phase 1" (decouple the download URL into a configurable base) is conceptually the same move Provide sqlite via a different URL #110 just made in semsql. Worth keeping OAK's phase-1 rollout in sync with this release so both libraries agree on the CDN as default.
Verified current PyPI state:
semsqlon PyPI is indeed still 0.4.0, uploaded 2025‑02‑05 — confirming the CDN-default commit (merged via #113, currently onmasterat8350307) has not gone out yet.How the release mechanism works here (from
.github/workflows/pypi-publish.yaml): publishing is triggered by creating a GitHub Release (on: release: types: [created]); the workflow then derives the package version fromgit describe --tags --abbrev=0and overwritespyproject.toml'sversionbefore building/publishing. So cutting the release is just: tagmaster(e.g.v0.4.1) and publish a GitHub Release from it — no manual version bump needed in the repo itself.One thing to consider bundling: #97 (Add graph column to main schema) was explicitly deferred with "put it into the schema for the next release in the minor series" — if that's still desired, it may be worth checking whether it should land before/alongside this cut rather than in a follow-up release.
Since this requires pushing a tag / creating a GitHub Release (repo-admin/PyPI-trust action), I'm leaving that step to a maintainer rather than acting on it here.
- Provide sqlite via a different URL #110 (Provide sqlite via a different URL) — closed, completed via PR Download SQLite databases from the semanticsql CDN instead of S3 #113, which repointed the
- added this to Software essential and proactive maintenance and removed this from Software essential and proactive maintenance
on Aug 18, 2026 [bot] Posted by a Claude Code agent on behalf of @kltm.
Update: with raw access retired 2026-08-28 (#112), this release request became a bug fix —
semsql==0.4.0's built-in download URL now returnsAccessDenied, sosemsql downloadis broken for all PyPI installs until a release ships the CDN default.— Posted by Claude Code agent on behalf of @kltm.
[bot] Opened by a Claude Code agent on behalf of @kltm.
The
builder.pydownload default was repointed tohttps://semanticsql.berkeleybop.io(with a non-urllibUser-Agent, since the CDN returns 403 toPython-urllib) — closing #110 — but that change is onmasterand not yet released: PyPIsemsqllatest is still 0.4.0. Until a release is cut,semsql downloadand anything depending onsemsqlstill hit the raw S3 bucket.Ask
Cut a new
semsqlrelease to PyPI containing the CDN default so downstream consumers pick it up. This unblocks the migration tracked in #112.Related: #110, #112, INCATools/ontology-access-kit#880.
— Posted by Claude Code agent on behalf of @kltm.