Repository navigation
ci(extensions): bump pinned GNOME extensions - #1290
mergeraptor[bot] wants to merge 2 commits into
Conversation
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: danathar The full list of commands accepted by this bot can be found here. Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
Danathar
left a comment
There was a problem hiding this comment.
Stand-in review (the hive's reviewers are out of quota). Approving. All four submodule bumps are fast-forwards of their tracked branches from the current pins: blur-my-shell +20, bluefin-bling +28, copyous +31 and fuzzy-application-search +4, with nothing rewritten or behind. As the PR itself says, shell-version metadata doesn't prove GNOME 51 runtime behaviour, so the smoke suite's extension scenarios on the next image are the real check. copyous in particular has its own stress feature in the testsuite.
There was a problem hiding this comment.
Summary: This automated update moves four add-ons for the desktop (GNOME extensions) to newer upstream versions. Three of them really do change what gets built into the system. The fourth, the Copyous clipboard manager, does not: the system installs Copyous from a fixed, checksum-verified release file, so this bump only changes the version the build records as the Copyous source. That record now names code that is not what users get, and the Copyous fixes listed in the PR description will not reach users.
What I need from you: Decide whether to merge with the Copyous line as it is, or revert the Copyous pin on main until a matching Copyous release can be pinned with it.
correctness (medium, in-scope)
The Copyous bump is recorded but does not ship, so the build's provenance record (the file that lists which source each shipped extension came from) becomes wrong.
scripts/build-gnome-extensions.sh:133-139installs Copyous from the pinnedv2.1.2release zip, checked by sha256. It never reads thetmp/copyoussubmodule.scripts/verify-gnome-extensions.py:107-110("The gitlink is reference source; the verified archive is what ships") writes the submodule commit into provenance assource_commit, next to thev2.1.2archive.- Upstream tag
v2.1.2resolves to21ad86435b0f…, which is exactly the old pin (compare v2.1.2...21ad864→ identical). The new pind31a4c7fc405is 31 commits past it, and no newer release exists. After merge,/usr/share/utah/provenance.jsonwould pair the source commitd31a4c7with an archive built from21ad864. Shell-version checks also read metadata fromtmp/copyous/resources/metadata.json(verify-gnome-extensions.py:33), which is no longer the metadata that ships. - The PR body lists Copyous as "proposed". Its upstream range includes GNOME Shell freeze fixes (#186, #191), but none of them reach the image.
Fix on main (the bot will regenerate this PR from it): keep the Copyous gitlink at 21ad86435b0f05a0e5e576c6fbf971a1124d99d6 until upstream cuts a release, then bump the gitlink, copyous_sha256 and the release URL together. A longer-term fix is to make the bump bot skip any extension listed in ARCHIVES (verify-gnome-extensions.py:70-76), or bump it only together with its archive. Alternatively, a maintainer can accept the mismatch, because the gitlink is documented as reference only.
Other three bumps: blur-my-shell (built from source, build-gnome-extensions.sh:19-21) and bluefin-bling (upstream changed only a CI workflow file) look fine. All image builds and the contract check passed on this head (run 43805518442, head_sha 796d83e). I could not diff fuzzy-application-search, which is hosted on Codeberg. Its review relies on the passing contract check and build.
No findings from: security, intent-alignment, style, docs-currency.
Confidence: 3/5 (needs attention) — 1 medium finding; 1 perspective requires a human decision
— hive: agent=reviewer backend=claude model=claude-opus-5-5 effort=medium claude=2.1.325
There was a problem hiding this comment.
Summary: This automated update moves four GNOME desktop add-ons (small programs that extend the desktop) to newer upstream versions. Three of them will really be built from the new version. The fourth, the Copyous clipboard manager, is listed as updated, but the system image still downloads the older release, so the record of what ships would be wrong. The newest change here only merged in work from the main branch and did not fix that.
What I need from you: Fix the Copyous issue on the main branch, either by keeping its old version pin or by updating its download to match.
Re-reviewed at head cd14b505. Changes since the last review at 796d83ed: the bot merged main into this branch (the Merge branch 'main' commit). The PR's own change is still the same four gitlink bumps. None of the merged files change how those four extensions are built, except one comment wording in scripts/build-gnome-extensions.sh:132.
Previous finding: still present.
correctness
-
The Copyous version bump is recorded but never shipped (
scripts/build-gnome-extensions.sh:136). The image does not build Copyous from thetmp/copyoussubmodule. It downloads the fixed release zipv2.1.2/copyous@boerdereinar.dev.zipand checks it against a hard-coded sha256 (lines 133–140). Thenrm -rf /usr/share/gnome-shell/extensions/tmpat line 142 deletes the submodule. The old pin21ad86435b0fis the upstreamv2.1.2tag. The new pind31a4c7fc405is 31 commits afterv2.1.2and has no release. After merge, the repo would claim one Copyous version and ship a different one, and the PR table's "proposed" row for Copyous would be misleading. The fix belongs in the bump bot onmainbecause the bot regenerates this branch. In.github/scripts/prepare-gnome-extension-bumps.pyprepare()(around line 102), skip submodules whose shipped artifact is a pinned release:RELEASE_PINNED = {"system_files/shared/usr/share/gnome-shell/extensions/tmp/copyous"} ... for name, path in submodules(root): if path in RELEASE_PINNED: pinned = run_git(root, "-C", path, "rev-parse", "HEAD").stdout.strip() bumps.append(Bump(path, pinned, pinned, "—", "skipped (shipped from pinned release zip)")) continue
The other option is to update the release URL and sha256 in
build-gnome-extensions.shwhenever the gitlink moves. That only works once upstream tags a release that matches the new head. The next bot run will then drop the Copyous gitlink from this PR, and the other three bumps can land.
No findings from: security, intent-alignment, style, docs-currency.
Confidence: 3/5 (needs attention) — 1 medium finding; 1 perspective requested changes
— hive: agent=reviewer backend=claude model=claude-opus-5-5 effort=medium claude=2.1.325
hanthor
left a comment
There was a problem hiding this comment.
Maintainer decision on the Copyous question: don't accept the mismatch. Copyous ships from the pinned v2.1.2 zip, so moving its gitlink to d31a4c7 makes provenance.json name source that isn't what ships. The other three bumps (blur-my-shell, fuzzy-application-search, bluefin-bling) are fine.
Fix on main, not here, since the bot regenerates this branch: make .github/scripts/prepare-gnome-extension-bumps.py skip extensions that ship from a pinned archive (the ARCHIVES set in verify-gnome-extensions.py), or bump them only together with their URL and sha256. The next bot run then drops Copyous from this PR and the rest can land. (#1296 removes the Copyous submodule anyway when the RPM extensions land.)
Reviewed on behalf of @hanthor with Claude Code
|
Hive status — updated in place; edits do not notify. Full history is in the Activity Monitor. stage 100-merge-queue → 070-fixing: human-changes-requested |
|
The main-side fix is #1344 (archive-shipped extensions are never bumped alone). Once it lands, the next bot run drops Copyous from this PR. |
|
Heads-up: #1358 (closes #1235) removes the |
GNOME extension bumps
Valid tracked-branch heads are proposed; missing or invalid metadata keeps the existing pin and does not block other extensions.
The proposal uses the repository-scoped App so normal pull-request contract and image-build CI run on its exact head.
Hive review, required checks and the normal merge policy still apply. Source metadata alone does not prove GNOME 51 runtime compatibility.
Heads
system_files/shared/usr/share/gnome-shell/extensions/appindicatorsupport@rgcjonas.gmail.comfbc758a467d8fbc758a467d8system_files/shared/usr/share/gnome-shell/extensions/tmp/bazaar-integration@kolunmi.github.io131388eff9d2131388eff9d2system_files/shared/usr/share/gnome-shell/extensions/blur-my-shell@aunetx840caaa957e00df9022b242dsystem_files/shared/usr/share/gnome-shell/extensions/tmp/caffeineba58454d90ddba58454d90ddsystem_files/shared/usr/share/gnome-shell/extensions/dash-to-dock@micxgx.gmail.com2dfae3474d692dfae3474d69system_files/shared/usr/share/gnome-shell/extensions/gradia-integration@alexandervanhee.github.iof79aa8239fdff79aa8239fdfsystem_files/shared/usr/share/gnome-shell/extensions/gsconnect@andyholmes.github.io61cc76b966f961cc76b966f9system_files/shared/usr/share/gnome-shell/extensions/custom-command-list@storageb.github.comc6e6bdbd3babc6e6bdbd3babsystem_files/shared/usr/share/gnome-shell/extensions/tmp/tilingshell@ferrarodomenico.comaabe5d5806efaabe5d5806efsystem_files/shared/usr/share/gnome-shell/extensions/BudsLink-Companion@maniacx.github.comc11b6cf5626ef76e8665c2c6system_files/shared/usr/share/gnome-shell/extensions/fuzzy-application-search@mkhl.codeberg.pageab335a4542d26f8185b86963system_files/shared/usr/share/gnome-shell/extensions/tmp/quick-settings-audio-paneldf490bff1aefdf490bff1aefsystem_files/shared/usr/share/gnome-shell/extensions/tmp/quicksettings-audio-devices-hider4db155654f1d4db155654f1dsystem_files/shared/usr/share/gnome-shell/extensions/tmp/quicksettings-audio-devices-renamerfac597a0c0fffac597a0c0ffsystem_files/shared/usr/share/gnome-shell/extensions/tmp/tailscale-gnome-qs5307ed116d9b5307ed116d9bsystem_files/shared/usr/share/gnome-shell/extensions/tmp/bluefin-blingca24a3fd2c2b7c615d7b8189system_files/shared/usr/share/gnome-shell/extensions/tmp/copyous21ad86435b0fd31a4c7fc405system_files/shared/usr/share/gnome-shell/extensions/tmp/syncthing-toggle-rehhouarib28698d48842b28698d48842