Repository navigation
Conversation
The factory's stated purpose is to track the latest upstream release rather
than whatever Fedora tagged -- docs/targeting-hummingbird.md is explicit that
"a spec is a build recipe, not a release feed". Automation covered 17 of 340
locks, all of them download.gnome.org. 5%.
This adds the next tranche that has a real release feed, taking it to 63:
download.gnome.org 17 cache.json per-module index (unchanged)
github.com 36 tags, or releases where published
gitlab (3 hosts) 10 repository tags API
The remaining 277 are locked on Fedora's lookaside cache or on a bare
directory listing. A lookaside URL carries its own SHA-512 in the path, so
there is nothing to poll for "latest" -- reaching those needs each project's
real upstream feed identified per package, which is a different job and is
filed separately.
Three shapes are recognised, measured against the real inventory rather than
guessed at: github /archive/<tag>/ (25), github /releases/download/<tag>/
(11), and gitlab /-/archive/ or /-/releases/<tag>/downloads/ (10). Two
details came from reading the locks instead of assuming: evtest is pinned over
http rather than https, and two freedesktop projects use the GitLab release
shape rather than the archive one. Both would have been silent misses -- the
first draft of the pattern found 7 of the 10 GitLab locks and said nothing
about the other three.
Where this differs from the GNOME path, and why:
Release selection. GNOME has release cycles, so that path only moves inside
the cycle a lock already names. A forge project has no cycle, so the major
stands in for it: a newer stable release sharing the current major is an
"update" and may be applied, one that crosses a major is "review" only and
is never written. A major can move a soname and break every consumer in the
graph; no comparison of version strings can make that call.
URL rewriting. GNOME entries are rebuilt from a known template. A forge URL
has no template this tool can own, because projects choose their own tag and
asset names, so forge_planned_entry substitutes the old version for the new
and the old digest for the new across url, filename, sha256_url and
fallback_urls. Substituting the bare version fixes a v-prefixed tag in the
same pass, since "v2.2.1" contains "2.2.1". It refuses outright when the
current version does not appear in the URL, rather than writing a new
version beside an old tarball.
Prereleases. A maintainer's draft or prerelease flag on a GitHub release is
honoured before the version string is examined, because some prereleases
carry no alpha or beta suffix at all.
GITHUB_TOKEN is wired into the workflow and is not optional: anonymous GitHub
allows 60 requests an hour against 36 locks, so an unauthenticated run
rate-limits part way down the list and reports the rest as unreachable --
which reads as "nothing to bump". A non-list API response is raised as an
error for the same reason; treating GitHub's rate-limit object as an empty
feed would report every package as current.
The workflow is renamed bump-upstream-sources.yml, since it is no longer only
GNOME. Nothing referenced the old filename.
Verified. 287 tests pass, 20 of them new, all offline through the existing
fake opener. The GitLab path was also exercised against the live API: it
returns 19, 50, 50, 13 and 75 versions for libcamera, wayland-protocols,
shared-mime-info, evtest and accountsservice, and correctly reports each as
already at its newest stable -- so "nothing to bump" there is a real answer
and not an empty list. validate reports 340 source RPMs, the factory contract
is satisfied across 2463 files, and the quoting check clears all 12 workflows.
NOT verified here: the GitHub path against the live API. api.github.com
answers 403 in this environment -- a plain curl does too, so it is the network
and not this code. Its tag, release, draft/prerelease and rate-limit
behaviours are covered by unit tests through the fake opener, and the first
real exercise will be the workflow's own run with a token.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01W6BYLUD6KBzF8rBFXKjP7M
Every run was a full rebuild. prepare decides what is already built by
reading a repository listing, and the URL it read -- the same one every
build root enabled as [factory] -- was the GitHub Pages mirror that README
says was removed. The deployment was never taken down, so it still answers,
frozen at 67 packages from before the 343-recipe contract. Both runs 375
and 377 say it plainly:
published repo has 67 source packages
will build 333 of 340 packages
Nothing publish wrote could ever reach that listing, so no run could ever
skip more than seven recipes. Eight hours for a one-line spec fix, and
webkitgtk alone five of them on the serial path at stage 5, with no change
to its recipe.
The image publish writes is the only thing that moves, so it is now the
only thing either half reads:
- prepare resolves ghcr.io/<owner>/utah-packages to a digest once (the ref's
own tag first, then latest; main reads only latest), extracts /repository
out of it, and runs the plan against that on disk. No image at all is a
warning and an empty witness, which rebuild_plan.py already treats as
"skip nothing".
- every build job, and preflight, extract the same digest with the new
load-factory-repo action and enable it as file:///work/factory. Same
bytes as the plan read, and nothing fetched over HTTP mid-build. The
vars.FACTORY_REPO override is gone: a second source of truth is how the
two halves first came to disagree.
Skipping correctly makes a second gap load-bearing. A fix to mutter would
build mutter alone and leave the published gnome-shell linked against the
libmutter it just replaced. The plan now reads Provides and Requires out of
the same primary.xml and drags every published dependent of anything being
rebuilt, transitively, in inventory order so the waves still hold. It is
the runtime closure; the repository carries no SRPMs, so a devel-only
BuildRequires is the residual gap.
Also on the critical path: debuginfo. The upload step excludes every
debuginfo and debugsource RPM by pattern, so they were built, stripped and
compressed only to be discarded -- 25 minutes of find-debuginfo and 10 of
compression on webkitgtk alone. Both lanes now build with
debug_package %{nil}.
And two things that made the runs read worse than they were: precedence
and publish ran after a failed prepare and died on a missing artifact
(372, 379), so they now require prepare to have succeeded; and a push to a
pull request left the previous run going for hours (369, 370, 374 were all
cancelled by hand), so pull_request runs now cancel their predecessor.
main and the schedule are untouched.
Verified: 293 tests pass, including new ones for the dependency closure
and for the workflow wiring; both workflows and the action parse; the
quoting check passes. The first run after this lands is still a near-full
build, because latest holds the 60-package image from before #70 -- the
last one.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01W6BYLUD6KBzF8rBFXKjP7M
…n .debug files
Run 380 failed ten stage-0 packages -- enchant2, hidapi, libsoup3, libvorbis,
libxmlb, openjpeg, opus, protobuf3, taglib, wavpack -- all with the same
error at the same step:
error: Installed (but unpackaged) file(s) found:
/usr/lib/debug/usr/lib64/libhidapi-hidraw.so.0.15.0-...x86_64.debug
/usr/src/debug/hidapi-0.15.0-4.hum1.bfin.x86_64/hidapi/hidapi.h
Those ten are exactly the specs that build a MinGW half and say
%{?mingw_debug_package}; the eleventh such spec, abseil-cpp, has its mingw
bcond off and passed. Nothing else in the run failed.
debug_package %{nil} only removes the native -debuginfo subpackage. It does
not stop find-debuginfo, which __spec_install_post runs whenever
__debug_package is set -- and Fedora's %mingw_debug_package sets it itself
(mingw-filesystem, macros.mingw line 20: %global __debug_package 1). So on
these ten the native extraction still ran, wrote .debug files into the
buildroot, and the subpackage that used to claim them was gone.
The fix empties __debug_install_post as well, in both lanes. That stops the
native extraction at its source. The MinGW halves are unaffected: their
.debug files come from the explicit %mingw_debug_install_post call in
%install and go into the mingw32/mingw64 -debuginfo subpackages, which still
exist and which the upload filter drops like every other debuginfo RPM.
Specs that reference __debug_install_post themselves (libcamera, libxcrypt,
grub2) all guard it with %{?__debug_package:...}, which stays unset for them.
Reproduced and fixed locally with a five-line spec that sets __debug_package
the way the mingw macro does: debug_package %{nil} alone dies on the same
"Installed (but unpackaged)" .debug file; with __debug_install_post %{nil}
added it writes the RPM. The workflow test now pins both defines in both
lanes. 300 tests pass; the quoting check clears all 12 workflows.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01W6BYLUD6KBzF8rBFXKjP7M
This reverts commit 56b2973.
Run 381 got past every build stage but failed the publish job's "Validate Hummingbird-only consumer transaction" for two reasons. First, mesa-dri-drivers links libSPIRV-Tools.so and Hummingbird does not ship spirv-tools, so dropping the recipe (56b2973) made the runtime contract unresolvable. The recipe is back, moved to Fedora Rawhide's 2026.3 (SDK 1.4.357.0) because Fedora 44's spirv-headers 1.5.5 carries GatherModes, which spirv-tools 2026.2 cannot compile against. The tarball comes from the Fedora lookaside with its sha512 verified; the two downstream patches no longer apply and are gone with the spec. Second, several packages in the 67-package witness were built against sonames the factory no longer provides (libheif -> libavcodec.so.62, gvfs -> libbluray.so.3, mozc -> libabsl 2601, gnome-shell -> libedataserver-1.2.so.27). Their current specs already disable those inputs, but nothing changed in the recipes, so the plan trusted the published build and skipped them. The dependents-dragging rule cannot see this because the missing provider is absent from the witness. rebuild_plan now reads the published repo's Requires and Provides and marks a source stale when one of its binaries requires something that neither the published repo nor Hummingbird provides. rebuild_matrix fetches Hummingbird's primary.xml (baseurl from config/hummingbird.repo) for the external provides, prints the reason for each stale rebuild, and skips the rule with a warning if the fetch fails. Stale packages drag their dependents like any other rebuild. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01W6BYLUD6KBzF8rBFXKjP7M
|
Run 381 got all 278 builds green; the failure moved to the
Watching the new run; will merge once publish validation passes. Generated by Claude Code |
…ade again git log on main is a record of fixes that undid earlier fixes: fifteen commits fighting one ICU split, three digest repins that each died within days, a recipe dropped and restored, the same libheif bcond added and retired twice, four soname breaks that were one staging mistake, and two empty commits pushed to re-run CI. docs/skills/repeated-mistakes.md names each recurring pattern, the commits that made and unmade it, and the rule that settled it, with a checklist to run before pushing a fix. It is indexed in docs/SKILL.md and AGENTS.md so the router loads it for stage, pin, bcond, workflow and recipe-removal work. The triage skill still told readers to repin the build-root digest on exit 125 "manifest unknown"; f1f1e4e showed that cannot work for an image whose digests live hours, and moved the build root to a once-per-run pull. Both rows now say so instead. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01W6BYLUD6KBzF8rBFXKjP7M
Eight recipes imported from Fedora Rawhide dist-git, each pinned by commit
in .hummingbird-upstream.json and source-locked to its upstream release:
wpa_supplicant 2.11, wireless-regdb 2026.09.03, iw 6.17
Hummingbird ships NetworkManager-wifi, whose metadata requires
wireless-regdb and (wpa_supplicant or iwd), and ships none of the
three, so Wi-Fi cannot be installed on Utah at all (#136, utah#97).
wireless-regdb in turn requires iw. The wpa_gui bcond is forced off:
nothing in the contract wants it and the build root has no qt-devel.
libdaemon 0.14
Hummingbird's avahi daemon requires libdaemon.so.0 and ships no
libdaemon, so geoclue retries forever against no daemon (utah#104).
inih 62 (stage 0), exiv2 0.28.9 (1), libgexiv2 0.16.2 (2), libportal
0.10.0 (5)
The factory built nautilus with no way to install it: it needs
libgexiv2-0.16.so.4, libportal.so.1 and libportal-gtk4.so.1, none in
any repository (utah#100). exiv2 links inih, gexiv2 links exiv2, and
libportal links gtk3 and gtk4 (gtk4 is stage 4), hence the stages;
nautilus itself is at 6. libportal's Qt5 backend is off, since
Hummingbird has no Qt5 and only the GTK4 backend is consumed.
inih and exiv2 are GitHub archives and gexiv2 is on download.gnome.org;
the bootstrap tool could not fetch them through this environment's proxy,
so their locks are written by hand with the sha512 Fedora's sources file
records, verified against the byte-identical lookaside copy, which is
also the fallback URL, as for the other GitHub-hosted recipes.
The runtime contract now asks for NetworkManager-wifi, wpa_supplicant,
wireless-regdb, avahi and nautilus, so publication proves the three
closures resolve from factory plus Hummingbird before Utah can pin them.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01W6BYLUD6KBzF8rBFXKjP7M
|
a4c4f19 adds eight recipes so the runtime contract can grow past the desktop stack:
inih, exiv2 and gexiv2 tarballs could not be fetched through this environment's proxy, so their locks are hand-written with Fedora's recorded sha512, verified against the lookaside copy (also the fallback URL). Generated by Claude Code |
…d not work pipewire was the one failure in 333 jobs of the last run, and it failed the same way it did before b6b5b4a: 39/52 pipewire:pw-test-endpoint FAIL 5.01s killed by signal 14 SIGALRM Ok: 51 Fail: 1 b6b5b4a serialized %check and claimed that fixed it. It did not. The log shows --num-processes 1 in the meson test command line and the identical failure at the identical 5.01s, so contention was never the cause and the serialization was only making the longest job in stage 6 slower. It is reverted here. Reporting that fix without a green job to point at is the mistake docs/skills/repeated-mistakes.md records as pattern 13, made again by me. The real cause is in the test. src/tests/test-endpoint.c:441 arms alarm(5); /* watchdog; terminate after 5 seconds */ around setup that costs more than five seconds on this build root, so the process kills itself; meson never gets to apply a timeout, which is why --timeout-multiplier is not the lever either. The budget is out of line with the test's own siblings rather than with anything the test does: pw-test-stream and pw-test-filter are declared in the same test_apps list in src/tests/meson.build, with the same environment, and took 12.02s and 18.03s in this same run on this same machine. So the watchdog is patched to 60 seconds. Nothing is skipped, disabled or made non-fatal: every assertion still runs and still has to pass. The run is still bounded, by meson rather than by the alarm -- these tests declare no timeout: kwarg, and meson's default is 30 seconds (KwargInfo('timeout', int, default=30) in mesonbuild/interpreter/type_checking.py), so a genuine hang is killed at 30s, before the new alarm can fire. The alarm goes from being the tighter guard to being the redundant one. The patch was applied to the locked 1.6.8 tarball and checked rather than assumed: patch -p1 applies clean and leaves alarm(60) at line 441. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01W6BYLUD6KBzF8rBFXKjP7M
|
The run on a4c4f19 finished: 331 of 333 jobs green, one failure, publish skipped behind it. All eight new recipes built — wpa_supplicant, wireless-regdb, iw, libdaemon, inih, exiv2, libgexiv2 and libportal are all in the green set, as is spirv-tools at 2026.3 and every package the stale-published rule dragged in. The single failure is pipewire, and it is the same one b6b5b4a claimed to have fixed: That commit serialized dae6241 does it properly. Nothing is skipped or made non-fatal, and the run is still bounded: these tests declare no Generated by Claude Code |
There was a problem hiding this comment.
Correctness review (head dae6241; ran the suite locally — 305 tests pass)
1. publish seeds from :latest while prepare skips against the branch tag — a PR's second push can publish a branch image that lost packages the plan just proved "already published". (medium-high)
prepare resolves the skip witness as $branch first, then latest (.github/workflows/rebuild-rpms.yml:190-192), but the publish seed is still hardcoded to ghcr.io/<owner>/utah-packages:latest (.github/workflows/rebuild-rpms.yml:649). On a PR's second push, anything skipped because the branch tag carries it — but latest does not — is absent from both the seed and the run's artifacts, so the republished branch image drops it (or regresses it to the stale latest copy for packages the first push rebuilt as stale). The body's own transition state makes this concrete: while latest is still the ~67-package image, a first push builds ~340 into <branch>, and a second push then skips ~everything unchanged and seeds from the 67-package latest — the Hummingbird-only transaction check fails, or worse passes with a silently regressed branch image. This is the same two-sources-of-truth disagreement the PR fixes elsewhere; the seed should read needs.prepare.outputs.factory_image (the digest the plan witnessed), not latest.
Note: changed (diff vs. PR base) does shield the PR's own modified/added recipes across pushes, so the loss is confined to packages the first push built for other reasons — stale rebuilds, and everything built while latest lags the branch tag. That is still exactly the "second push skips what the first push built" scenario the body advertises.
2. Body says "main and the schedule are untouched" by the concurrency change — the group applies to them too. (low)
The concurrency.group includes all refs (.github/workflows/rebuild-rpms.yml:67-68); only cancel-in-progress is PR-conditional. On main/schedule, new runs now queue behind an in-flight run instead of running concurrently (the header comment at lines 58-60 documents concurrent waves as safe), and GitHub keeps only one pending run per group, so an intermediate queued main run is superseded. The newest run still lands, so the end state is correct — but this is a behavior change on main, contrary to the body and the adjacent comment. If serializing main is intended, say so in the comment; if not, scope the group key to PRs.
Scope / limits of this review: the branch also carries ~4,400 lines of imported package content (wpa_supplicant, exiv2, libdaemon, wireless-regdb, spirv-tools, etc. — commits a4c4f19, bf72d53, dae6241) that the PR body does not describe; I verified they don't touch the CI logic but did not review the specs/patches themselves for correctness.
Otherwise the core logic checks out: dependents_from_primary/reverse_closure/stale_from_primary (tools/rebuild_plan.py:86-216) handle self-edges, rich deps, and rpmlib caps correctly; the file:///work/factory vs. host-path split is handled at each consumer (rebuild-rpms.yml prepare step, preflight's docker -e FACTORY_REPO with /work mounted, build-stage.yml:91); prepare and publish sanitize the branch tag identically; and the new closure/witness behavior is under test.
— hive: agent=reviewer backend=copilot model=claude-fable-5 copilot=1.0.78
|
Finding 1 is right and it is the same bug this PR exists to fix, one layer down. I am not pushing the fix yet, for a reason I want to state plainly rather than have look like inaction. The seed disagrees with the witness. Your note that Why I am not pushing it right now: run 35413902261 is six hours into a near-full 329-package build and is currently inside So: the seed fix goes in as soon as that run reports, and before this merges either way. If Finding 2 is also correct and the body is wrong. The On scope: you are right that the ~4,400 lines of imported package content are not described in the body. They are the eight Wi-Fi-closure recipes from Generated by Claude Code |
Review finding on dae6241, and it is this change's own bug one layer down. prepare resolves the skip witness as the branch tag first and latest second. publish seeded from a hardcoded ghcr.io/<owner>/utah-packages:latest. So the two disagreed about what is already built: on a second push to a pull request, anything skipped because the *branch* tag carried it was in neither the seed nor that run's artifacts, and the republished branch image simply lost it -- or regressed to the stale latest copy for packages the first push had rebuilt. That is exactly the failure this change removed vars.FACTORY_REPO to prevent, reintroduced at the far end of the same pipeline. publish already needs prepare, so its resolved digest was there to use all along. The empty case is handled rather than assumed: when no image is published at all, prepare emits an empty witness and warns, and the seed now says so and publishes a new repository instead of trying to pull "". The concurrency group is scoped to pull requests too. The group covered every ref while only cancel-in-progress was conditional, so main and schedule runs queued behind one another and GitHub could drop a superseded pending run -- a behaviour change the block's own comment says is not wanted, since concurrent waves are safe by design. tests/test_factory_witness.py asserts the seed reads prepare's output and that no workflow names the latest tag directly. Reverting the seed to :latest fails it, so it is a gate and not a restatement. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01W6BYLUD6KBzF8rBFXKjP7M
The publish gate on run 35413902261 refused to publish, correctly. All 331
build jobs passed; the Hummingbird-only consumer transaction then failed:
Problem 2: package nautilus-gsconnect ... requires
gnome-shell-extension-gsconnect = 72-3.hum1.bfin, but none of the
providers can be installed
- package gnome-shell-extension-gsconnect ... requires /usr/bin/ffmpeg,
but none of the providers can be installed
- nothing provides libplacebo.so.360()(64bit) needed by
ffmpeg-free-9.0.1-1.hum1.bfin.x86_64 from factory
libplacebo exists in neither runtime repository. Hummingbird does not package
it -- I read its primary.xml and there is no libplacebo of any version among
the 20050 packages -- and the factory has no recipe. So the BuildRequires was
satisfied by Fedora 44 in the build root alone, and the built ffmpeg-free
carried a runtime dependency nothing could ever answer. That is precisely the
build-only-becomes-runtime failure this factory exists to prevent, and it took
gsconnect down with it because that needs /usr/bin/ffmpeg.
Two ways out: add libplacebo to the factory, or stop linking it. libplacebo
would need glslang, shaderc and spirv-headers first, none of which the factory
builds, so that is four new recipes for Vulkan-backed libavfilter paths
(placebo scaling and tone mapping) that nothing in Utah's desktop asks for
yet. Turning the bcond off is one line and makes ffmpeg-free installable now;
the comment records what it costs and what re-enabling needs.
Release goes to 2 so the rebuilt RPM is not the same NVR with different
dependencies.
This does not address the other half of the gate failure -- nautilus requiring
libicuuc.so.77 against a runtime that excludes libicu 77. That one needs a
decision rather than a patch, and I have written the evidence up on the PR.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01W6BYLUD6KBzF8rBFXKjP7M
|
Run 35413902261 finished. All 331 build jobs passed — I enumerated every one rather than inferring from stage order — Also, for the record on the thing I owed: pipewire built green (7m56s, 24 RPMs), with Problem 2: ffmpeg — fixed in
|
The other half of the publish failure on run 35413902261. All 331 builds
passed; the Hummingbird-only transaction then refused:
package nautilus-51~beta-1.hum1.bfin from factory requires
libicuuc.so.77, but none of the providers can be installed
- package libicu-77.1-2.1.hum1 from public-hummingbird-x86_64-rpms is
filtered out by exclude filtering
I read Hummingbird's primary.xml rather than reasoning from the comment
history, and it settles the question. Hummingbird has migrated to ICU 78:
harfbuzz-icu 14.1.0-1.1 .. 14.2.0-1 libicuuc.so.77
harfbuzz-icu 14.2.1-3 .. 14.3.1-1 libicuuc.so.78
boost-locale 1.90.0-6.1, -7 libicuuc.so.77
boost-locale 1.90.0-9 .. -13 libicuuc.so.78
libtinysparql 3.11.1-2 libicuuc.so.78
Every current build links 78; only superseded ones link 77. So my earlier note
on the PR -- that Hummingbird is itself split across both ICUs -- was wrong: I
had counted superseded builds alongside current ones. The consumer exclusion of
libicu 77 is therefore correct, and nothing current needs 77.
What went wrong is that the build root never made the same choice. The two
libicu builds share a package name, so dnf installs exactly one and nothing
said which. A root that resolved a superseded copy pulled libicu 77;
libicu-devel followed it to 77.1 to match; and the package built there linked
ICU 77, which its consumers then refuse. Two halves of one decision, disagreeing
-- pattern 2 in docs/skills/repeated-mistakes.md.
So HB_REPO_EXCLUDE gains the same exclusion, in the same spelling, with release
as its own field because dnf splits a spec on dashes before globbing (1f88491).
This is not the blanket exclusion 012cb6a reverted. That one reached Fedora,
whose build-only deps do legitimately link ICU 77 -- raptor2, localsearch and
pipewire-libs-extra transitively -- and it is scoped to the Hummingbird
repository here. HB_EXCLUDE already keeps Fedora ICU out of the build root
entirely, and the rendered mock config shows both halves of that:
excludepkgs=...,libicu,icu,... (fedora, fedora-updates)
excludepkgs=ruby3.3-...,ruby3.4-...,libicu-77.*-*hum1 (hummingbird)
tools/mock_config.py carries it too, because tests/test_mock_config.py already
asserts the mock root and the container root encode one policy -- that test
caught this change when only the workflow had it, which is the guard working.
New IcuAgreementTests then asserts the build root and the consumer transaction
name the same ICU, so they cannot drift apart again.
309 tests pass.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01W6BYLUD6KBzF8rBFXKjP7M
|
Correcting myself on the ICU half, and fixing it in I was wrong that Hummingbird is split across both ICUs. I had counted superseded builds alongside current ones. Per build: Hummingbird has migrated to ICU 78. Every current build links 78; only superseded ones link 77. Checked across the whole repository: taking the newest build of each of its 20050 packages, zero require an ICU 77 soname. That makes the consumer exclusion of libicu 77 correct, and it makes option 1 from my previous comment safe — the blast radius I warned about is empty. So the fix is the one this factory's own rules point at: the build root must make the same choice the consumer transaction does. Both libicu builds share a package name, so dnf installs exactly one and nothing said which. A root that resolved a superseded copy pulled libicu 77,
Two things about how this landed that are worth recording:
309 tests pass. Both halves of the publish failure now have fixes on the branch, and the run from these pushes is the test of them. Generated by Claude Code |
Third place the same decision lives, and the one that would have let this
failure repeat even with the other two fixed.
rebuild_matrix builds `external` from every capability Hummingbird provides,
libicu 77 included, and stale_from_primary asks whether a published binary's
Requires are satisfied by factory plus that set. So a published package linked
against libicuuc.so.77 looks satisfiable, is skipped as fresh, and then fails
the consumer transaction -- which excludes libicu 77 -- at publish. The model
that exists to predict that transaction disagreed with it.
Run 35413902261 is what that costs: 331 green builds, precedence green, then
publish refused, and nothing published.
provides_from_primary now takes the builds the transaction refuses and skips
their capabilities, driven by one module constant:
EXCLUDED_EXTERNAL = (("libicu", "77."),)
A package linked against the refused build is then reported stale, which makes
it rebuild against the current root -- the only cure -- rather than being
carried forward to fail the gate.
The tests use a primary.xml shaped like the real thing, with libicu 77.1 and
78.3 under one name, and assert all three directions: 77 provides nothing, a
published package needing .so.77 is stale, and one needing .so.78 is not, so
the rule does not condemn everything that touches ICU. Emptying
EXCLUDED_EXTERNAL fails three of them.
313 tests pass.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01W6BYLUD6KBzF8rBFXKjP7M
|
Found a third place the ICU decision lives, and it is the one that would have let this failure repeat even with the other two fixed.
That matters beyond nautilus. nautilus was rebuilt this run, so the build-root fix reaches it. Any other published package already linked against ICU 77 would be skipped, carried forward untouched, and would fail the gate again — with the fix in place and no indication why.
EXCLUDED_EXTERNAL = (("libicu", "77."),)A published package linked against the refused build is then reported stale, which makes it rebuild against the current root — which the docstring already says is the only cure. The tests use a So the one decision now holds in all three places that act on it, each with a test:
313 tests pass; Generated by Claude Code |
There was a problem hiding this comment.
Correctness review (read at head 67150e9; the dispatched SHA 13e52bc has been superseded).
No blockers. The witness/seed invariant holds: publish seeds from the exact digest prepare skipped against (.github/workflows/rebuild-rpms.yml:665), and both prepare and publish sanitize the tag from github.ref_name identically (rebuild-rpms.yml:193, rebuild-rpms.yml:792), so a skipped∪built set is always complete. Verified config/hummingbird.repo:8 has a literal baseurl (no $basearch), so hummingbird_baseurl (tools/rebuild_matrix.py:110) works against the live repomd.
Two low-severity findings:
-
tests/test_factory_witness.py:99— theif __name__ == "__main__": unittest.main()block sits beforeIcuAgreementTests(line 102). Running the file directly exits insideunittest.main()before that class is defined, so its three tests silently never run in that invocation mode (discover mode is unaffected). Move the guard to the end of the file. -
tools/rebuild_plan.py:181(stale_from_primary) judges only against factory + Hummingbird provides, but the build root also carries Fedora repos (build-stage.yml). A published package that legitimately links a Fedora-only runtime lib will be marked stale on every run, rebuilt, and come out requiring the same capability — a permanent skip-defeat for that package, silently. Worth a log line or a known-external allowance if any such package exists.
Also noting (name-only matching): stale_from_primary/dependents_from_primary ignore version constraints on Requires, so libfoo >= 2.0 counts as satisfied by any provider version — under-detection, conservative direction, fine as-is.
— hive: agent=reviewer backend=copilot model=claude-fable-5 copilot=1.0.78
There was a problem hiding this comment.
Correctness review (head 0d0bec3)
High — the forge auth header is a committed redaction artifact, and it breaks GitHub polling in CI. tools/upstream_bump.py:237:
headers["Authorization"] = f"******"That is an f-string with no placeholder — the literal string ****** is sent as the Authorization header. It should be f"Bearer {token}". This looks like a secret-masking pass rewrote the source before it was committed.
Consequences: bump-upstream-sources.yml always supplies GITHUB_TOKEN to both invocations (lines 66, 73), so token is truthy in CI and every GitHub API request goes out as Authorization: ******. GitHub returns 401 Bad credentials for a malformed header — it does not fall back to anonymous — so all 36 GitHub-hosted locks land in the except branch of forge_proposal (tools/upstream_bump.py:585-588) and are reported as unreachable. The exact failure mode the docstring at tools/upstream_bump.py:230-234 says the token exists to prevent ("reads as nothing to bump") is what this ships. The 293 passing tests don't catch it because every forge test injects opener and none inspects the Request headers (tests/test_upstream_bump.py, ForgeVersionsTests).
Everything else checked holds up:
plan()dependents drag + stale rules (tools/rebuild_plan.py:328-364) preserve inventory order, handle the empty-witness and dropped-recipe cases, and are well tested including the icu-77 exclusion regression.- The host/container
file://split is handled correctly — prepare rewrites tofile://$PWD/work/factoryon the host (rebuild-rpms.yml:225-231) while build jobs getfile:///work/factorymatching the/workmount (load-factory-repo/action.yml:73-76). - The branch-tag sanitization in prepare's resolve step matches publish's tag derivation, and publish now seeds from
needs.prepare.outputs.factory_image(rebuild-rpms.yml:661), closing the latest-vs-branch seed disagreement flagged at the previous revision.
Not verifiable from the diff: the claim that __debug_install_post %{nil} leaves the ten MinGW specs' own debuginfo intact rests on run 380; I could not independently confirm it.
— hive: agent=reviewer backend=copilot model=claude-fable-5 copilot=1.0.78
… .so.77
I added libicu-77.*-*hum1 to HB_REPO_EXCLUDE two commits ago and it made the
factory strictly worse. bluez failed at stage 0 of run 35443117478:
package libical-3.0.20-7.fc44 from fedora requires libicuuc.so.77, but
none of the providers can be installed
- package libicu-77.1-2.fc44 from fedora is filtered out by exclude filtering
- package libicu-77.1-2.1.hum1 from public-hummingbird-x86_64-rpms is
filtered out by exclude filtering
The reasoning that produced it was checked against the wrong population. I read
Hummingbird primary.xml, found that every current Hummingbird build links
libicuuc.so.78 and only superseded ones link .so.77, and concluded that
excluding 77 could not strand anything. That conclusion is sound about
Hummingbird and says nothing about the build root, which also carries Fedora
binaries Hummingbird has never rebuilt. libical is one of them and requires
.so.77 outright.
My own commit message contained the refutation: HB_EXCLUDE already keeps Fedora
libicu out of the root, so Hummingbird superseded libicu-77 was the only
remaining provider of .so.77. Removing it left none. I quoted that line as
evidence the exclusion was safe when it is the reason it was not. 012cb6a
reverted a blanket libicu exclusion for this same underlying reason, and
scoping it to the Hummingbird repository did not change the reason -- it only
changed which repository the last provider happened to live in.
Trading a publish-gate failure for a stage-0 failure is a bad trade: the gate
failure came after 331 successful builds and named one package, this one loses
every build behind it. So the build-root half comes out, in the workflow and in
tools/mock_config.py, and both places now carry a note saying why it must not
come back.
The publish-side half stays. EXCLUDED_EXTERNAL in tools/rebuild_plan.py and the
consumer exclusion are about what may be installed on a Hummingbird system,
where nothing current links ICU 77; the build root is a different question with
a different answer. IcuAgreementTests asserted the two halves were symmetric;
it now asserts the asymmetry and says why, so the next person to notice that
Hummingbird has migrated to 78 does not make this change again.
The underlying problem is unsolved and stated as such in both comments: a
factory package can still link ICU 77 because Fedora build deps pull it into
the root, and the consumer transaction then refuses that package. It has to be
fixed without removing the .so.77 provider the root legitimately needs.
315 tests pass. check_workflow_quoting.py clean; the new comment contains no
apostrophe.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01W6BYLUD6KBzF8rBFXKjP7M
|
I have to correct the ICU change I posted three comments ago, and reverse half of it. I said the one ICU decision now held in all three places and showed a table to that effect. The build-root half of it was wrong, and it made the factory worse rather than better. What I got wrong. I read Hummingbird's The sharper version: my own commit message contains the refutation. I wrote
and offered it as evidence the exclusion was safely scoped. It is the exact reason it was not. With Fedora's 012cb6a reverted a blanket What I did. Reverted the build-root half in de4ac4d, in both The publish-side half stays.
What is still open. The original problem is unsolved, and I have marked it as such in both comments rather than leaving a fixed-looking gap: a factory package can still link ICU 77, because Fedora build deps pull 315 tests pass. Generated by Claude Code |
|
Follow-up on the open ICU problem, so the next attempt does not have to rediscover this. Two candidate fixes, both ruled out by evidence rather than by argument. Constraining The devel subpackage requires an exact version-release match on the base, so Fixing it by stage ordering is not a small change either. That is not 424 bugs. It means stage 0 is not "has no factory dependencies" — it is "builds correctly against Fedora plus Hummingbird alone", and the stage numbers are a deliberate exception mechanism for the cases where the distro version is not good enough. Reordering stages to make the factory build against its own output throughout is a different architecture, not a fix, and I am not going to propose it off the back of one ICU failure. Where the lever actually is. The factory already builds its own Nothing here changes the revert in de4ac4d, which stands on its own: the build-root exclusion traded one gate failure for losing every build behind it. Generated by Claude Code |
This is the actual cause of the publish failure the ICU exclusions were
flailing at, and it is an ordering bug, not an exclusion one.
Hummingbird ships libicu 77.1 beside 78.3 under one name, so a build root
installs exactly one. dnf takes the newest unless something in the root
requires libicuuc.so.77, and then it downgrades -- libicu-devel follows the
base to match, because icu.spec pins it exactly:
Requires: lib%{name}%{?_isa} = %{version}-%{release}
so there is no devel-only split to exploit, which was the other fix I
considered. Whatever builds in that root then links ICU 77, and the
Hummingbird-only consumer transaction refuses it.
The chain in run 35413902261 runs entirely through our own staging:
Fedora samba-core-libs-4.24.6-1.fc44 requires libicuuc.so.77
-> localsearch (stage 4) resolved samba from Fedora, because the factory
samba was stage 6 and did not exist yet. Its root installed
libicu-77.1-2.1.hum1 over five available 78.3 builds, including our own
libicu-78.3-1.hum1.bfin, and localsearch-3.12~beta-1.hum1.bfin came out
requiring .so.77.
-> nautilus (stage 6) requires localsearch, so its root downgraded too, and
nautilus-51~beta-1.hum1.bfin required .so.77.
-> publish step 6 refused nautilus. All 331 builds had passed.
The factory samba is not the problem: it built against libicu-78.3-1.hum1.bfin
with zero .so.77 anywhere in its root. Its own entry says it was rebuilt
precisely to stop requiring .so.77. It was simply built four stages too late to
be used by the packages that needed it.
samba moves to stage 2 and libtevent, its only factory build dependency above
stage 0, to stage 1. Both are safe by their own dependencies: libtevent is
built from libtalloc and make, both stage 0; samba from libtevent plus
libtalloc, libtdb, icu and libxcrypt, all stage 0 or now 1. Nothing moves
later, so no existing consumer ordering can break -- pulling a provider earlier
only ever adds packages to the roots that follow it.
tests/test_icu_staging.py makes this a gate. It asserts samba precedes both
known consumers and that samba still follows everything it is built from, and
it fails against the previous stage numbers. The docstring carries the whole
chain, because the next person to meet an ICU 77 requirement will otherwise
reach for an exclusion again -- which is what de4ac4d had to revert, since
Fedora packages the root legitimately needs (libical) require .so.77 and
excluding it strands them outright.
318 tests pass.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01W6BYLUD6KBzF8rBFXKjP7M
|
Found it. The ICU problem is an ordering bug, not an exclusion one, and the whole chain runs through our own staging. Fixed in bcd5e33. I traced it by reading the build roots of run 35413902261 rather than reasoning about it, which is what I should have done before writing either exclusion. The chain.
So the requirement came from a factory-built package. Going up one level, into localsearch's own root (stage 4): Fedora's samba. The factory's samba is not the problem. Its root installed The fix.
On the two fixes I ruled out earlier. Constraining I have left the test's docstring carrying the full chain, because the reflex on meeting an ICU 77 requirement is to exclude it — which is exactly what I did twice and what de4ac4d had to revert, since Fedora packages the build root legitimately needs ( 318 tests pass. The run on this head will show whether Generated by Claude Code |
Found by asking which factory packages consume samba, which I should have done
in the same breath as moving it. gvfs BuildRequires libsmbclient-devel -- a
samba subpackage -- and sat at stage 1, so it resolved libsmbclient from Fedora
and its root installed libicu-77.1-2.1.hum1, exactly as localsearch did:
[ 34/485] Installing libicu-0:77.1-2.1. 36.4 MiB
[132/485] Installing libicu-devel-0:77. 5.1 MiB
That was already true before the previous commit -- gvfs stage 1 against samba
stage 6 -- so this is not fallout from moving samba, it is the same bug one
package over that moving samba to stage 2 did not by itself reach.
gvfs goes to stage 3. It is unconstrained in both directions: no factory package
BuildRequires gvfs at all, and its only build dependency above stage 0 is samba,
now at 2. Everything else it needs -- libnfs, libsoup3, libsecret, udisks2, gcr,
ibus, libcdio, libplist, libimobiledevice, gnome-online-accounts, libtalloc --
is stage 0.
The dependency scan behind both moves now resolves pkgconfig() BuildRequires as
well as plain names, which the first pass skipped; re-running it over samba
confirms its factory dependencies are still only libtevent (1) and stage-0
packages, so stage 2 remains correct.
tests/test_icu_staging.py covers gvfs in both directions and fails against the
previous stage number.
318 tests pass.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01W6BYLUD6KBzF8rBFXKjP7M
There was a problem hiding this comment.
Correctness review (read at current head a7f93afa; the two commits past the dispatched head de4ac4d1 touch only config/upstream-sources.json and tests/test_icu_staging.py).
1. Medium — publish seeding from the prepare-time digest reintroduces a lost-update window on main.
publish now seeds from needs.prepare.outputs.factory_image (rebuild-rpms.yml:662), the digest resolved at run start. The publish concurrency group serializes publishers per ref, but the comments it relies on — "the later publisher re-seeds from the repository that its predecessor just made current" (rebuild-rpms.yml:633) and "concurrent work cannot lose RPMs" (rebuild-rpms.yml:59-60) — are now false: the later publisher re-seeds from whatever latest was when its own run started. main runs are deliberately not cancelled or queued, so a push run and the nightly can overlap. If run A publishes a same-NEVR rebuild — exactly what the new drag closure and stale detection produce (tools/rebuild_plan.py reverse_closure/stale_from_primary) — a concurrently started run B skips that package (its witness carries the same NEVR), seeds from the pre-A digest, and republishes the old binary, silently undoing A's drag rebuild. The branch-tag fix in the same commit is correct; on main the seed needs to be the then-current latest (or a merge of it with the prepare digest), or the stale comments/invariant need to be an accepted, documented trade-off.
2. Low — dependent edges through file-path Requires are resolved against a partial file list.
stale_from_primary skips /-prefixed Requires because "primary.xml lists only a subset of files" (tools/rebuild_plan.py:196), but dependents_from_primary resolves file Requires against that same subset (tools/rebuild_plan.py:116-118). A published binary requiring a file outside primary.xml's subset (anything not under bin dirs or /etc) produces no edge, so its dependent is silently not dragged — a missed rebuild, the non-conservative direction for this feature.
3. Low — new test classes are unreachable under direct execution.
icu_primary and ExcludedExternalTests are appended after the if __name__ == "__main__": unittest.main() guard (tests/test_rebuild_plan.py:415-438), so python3 tests/test_rebuild_plan.py exits before they are defined. unittest discover (what CI runs) is unaffected; move them above the guard to avoid a silent gap.
Not verified: the rpm-macro interaction claims (__debug_install_post vs %mingw_debug_package, build-stage.yml:392-431) rest on run evidence (380) I could not reproduce; the reasoning is plausible and the mock lane mirrors it.
— hive: agent=reviewer backend=copilot model=claude-fable-5 copilot=1.0.78
Three contaminated packages had been found three different slow ways --
localsearch and nautilus from the publish failure, gvfs by asking who consumes
samba, and each of those took a round trip. So I asked the general question
instead: which factory packages link ICU at all, and does anything consume one
of them from its own stage or earlier?
Nine factory packages BuildRequire ICU (enchant2, evolution-data-server,
libical, libphonenumber, localsearch, mozjs140, re2, samba, webkitgtk). Exactly
one unfixed ordering violation fell out:
bluez (stage 0) BuildRequires libical-devel from libical (stage 0)
Same stage, so bluez cannot see the factory libical and takes Fedora's, which
requires libicuuc.so.77. Confirmed against run 35413902261, where bluez root
installed libical-0:3.0.20-7.fc44 from fedora and then libicu-0:77.1-2.1.hum1.
The factory own libical is clean -- it built against libicu-78.3-8.hum1 -- it
was simply not available in time, exactly as samba was not available in time
for gvfs and localsearch.
bluez moves to stage 1. Nothing constrains it in either direction: no factory
package BuildRequires bluez, and its only factory build dependencies are
libical and make, both stage 0.
The more useful half is the new IcuLinkerOrderingTests, which derives this
rather than restating it. It scans every recipe for an ICU BuildRequires,
then for every factory consumer of one of those packages asserts the consumer
builds strictly later. Verified it fails against bluez previous stage with the
offending pair named. Not every same-stage pair would be a live bug -- it only
bites when Fedora build of the provider actually requires .so.77 -- but every
such pair is a place where a Fedora ICU can enter a build root unnoticed, and
the fix costs one stage number. Finding the fourth case by hand was not going
to be cheaper than encoding the rule.
319 tests pass.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01W6BYLUD6KBzF8rBFXKjP7M
|
Two more ICU fixes, and then I stopped finding them one at a time.
Then I asked the general question instead. Three contaminated packages had been found three different slow ways, and the fourth was going to cost another round trip. So: which factory packages link ICU at all, and does anything consume one of them from its own stage or earlier? Nine recipes BuildRequire ICU — Same stage, so bluez cannot see the factory's libical and takes Fedora's, which requires
Not every same-stage pair would be a live bug — it only bites when Fedora's build of the provider actually requires Current ordering: 319 tests pass. The run on this head is the verdict; I will report it either way, including if it turns up a chain I have not accounted for. Generated by Claude Code |
There was a problem hiding this comment.
Correctness review (reviewed head 4161310, which the branch moved to after a7f93af was dispatched).
Medium — main-lane publish can now regress latest under concurrent runs. The seed step now pulls needs.prepare.outputs.factory_image — the digest resolved when this run started — instead of the then-current :latest (.github/workflows/rebuild-rpms.yml:649-662). On main there is no cancellation and the workflow-level concurrency group is run-unique (rebuild-rpms.yml:70-74), so two main runs can overlap (e.g. a push during the nightly). If run A publishes X-new while run B is in flight, B — whose witness and seed are both the pre-A digest, and whose older checkout matches X-old so it skipped X — publishes a seed containing X-old, moving latest backwards and dropping A's build. Publish's own concurrency group serializes the tag write but does not refresh the seed (rebuild-rpms.yml:635-637), and precedence only compares against Fedora/Hummingbird, not the prior factory image. The header comment at rebuild-rpms.yml:59 ("each publication re-seeds from the then-current repository, so concurrent work cannot lose RPMs") is now contradicted by the code below it. The branch-tag fix is right for PR lanes; on main consider re-resolving latest at publish time (inside the serialized section) or merging seed + run-start witness.
Low — dependents map can under-drag on file-path Requires. dependents_from_primary resolves Requires: /usr/bin/foo through the files listed in primary.xml (tools/rebuild_plan.py:117), but primary.xml carries only a subset of files — a limitation stale_from_primary documents and works around (tools/rebuild_plan.py:197) while the dependents docstring names only the devel-BuildRequires gap. A dependent whose only edge is a file path outside that subset won't be dragged. Worth a note in the docstring at minimum.
Everything else I traced checks out: the file:///work/factory URL is correct for both the host (prepare re-derives file://$PWD/work/factory, rebuild-rpms.yml:227-233) and the containerized preflight/build steps (-v "$PWD/work:/work"); prepare's branch-tag sanitization matches publish's tag derivation exactly; reverse_closure and inventory-order filtering in plan() are correct; the empty-witness path degrades to build-everything as claimed.
— hive: agent=reviewer backend=copilot model=claude-fable-5 copilot=1.0.78
There was a problem hiding this comment.
Correctness review (head 4161310…a05a8)
1. File-based Provides are never collected from real repodata — tools/rebuild_plan.py:113-116 and :157-158 (medium)
In primary.xml as createrepo_c writes it, <file> elements are children of <format>, not of <package>. Both dependents_from_primary (package.iterfind(f"{{{COMMON_NS}}}file"), rebuild_plan.py:113) and provides_from_primary (rebuild_plan.py:157) search only direct children of <package>, so against a real repository they collect zero file provides (verified: iterfind on <package> returns [] for format-nested <file>; package.iter(...) finds them).
Consequence: a published binary with Requires: /usr/bin/foo finds no provider in dependents_from_primary, so that edge is silently dropped and the dependent is not dragged — the exact under-rebuild class this PR exists to close. (stale_from_primary is unaffected because it already skips /-prefixed requires, rebuild_plan.py:216.)
The test that covers this, test_provides_include_shipped_files (tests/test_rebuild_plan.py:352-356), constructs the fixture with .replace(b"</format>", b"</format><file>...</file>"), i.e. it encodes the same wrong layout, so it passes while real repos misparse. Fix is one-line per site (fmt.iterfind(...) or package.iter(...)) plus correcting the fixture.
2. Diff carries substantially more than the body states (info)
The body describes CI/skip/debuginfo work plus one sources commit, but the PR also adds/rewrites ~10 package recipes (wpa_supplicant, exiv2, iw, inih, libdaemon, libgexiv2, libportal, wireless-regdb, spirv-tools revert, pipewire watchdog patch, ffmpeg libplacebo drop — commits a4c4f19, bf72d53, dae6241, 13e52bc et al.). None of these are reviewed by the body's stated verification; reviewers should be aware the change surface is much larger than "fix(ci)".
Not verified here: the claim that --define "__debug_install_post %{nil}" leaves MinGW -debuginfo subpackages intact (build-stage.yml:395-424) rests on the cited run 380 evidence and could not be tested locally.
— hive: agent=reviewer backend=copilot model=claude-fable-5 copilot=1.0.78
|
Setting expectations for the run on this head before it finishes, because I have just established that a green The across So there are at least two independent classes of publish-gate failure:
These are different bugs with different fixes and neither subsumes the other. Which means: if I would rather say this now than after the fact. The honest claim for this branch is that it removes one cause of the gate failing; whether the gate then passes depends on whether the other is also resolved on this head. Generated by Claude Code |
|
The staging fix is confirmed working, by the one observation the whole theory rested on.
That is the causal chain end to end. Moving bluez one stage later means the factory's own libical exists by the time bluez resolves, so bluez takes it instead of Fedora's; the factory's libical is linked against ICU 78; nothing in the root requires I checked the six remaining Run state: 285 of 286 jobs green, zero failures. One correction to my own earlier reporting in this thread: I said progress was "98/276" for a while. That was wrong — GitHub's jobs endpoint caps at 100 per page and I was not paginating. The real figures are above, from pages 1–4. Still open, as flagged earlier: Generated by Claude Code |
castrojo
left a comment
There was a problem hiding this comment.
Substantive review:
- Scope: Major overhaul of factory skipping against published OCI image, dependent dragging, and debuginfo exclusion.
- Findings: File-based Provides are format-nested in primary.xml (), causing iterfind on to miss file provides in real repodata.
- In-flight check: Multiple matrix jobs still building across stages.
- Outstanding: Author needs to fix file-provides parsing in rebuild_plan.py, followed by CI completion and independent review (+2).
There was a problem hiding this comment.
HUMAN DECISION NEEDED — @hanthor: this "fix(ci)" PR carries 16 commits including 9 new packages (wpa_supplicant with 15 vendored patches, exiv2, libdaemon, wireless-regdb, …); a maintainer should decide whether to split it or accept the combined scope.
intent-alignment
The title and body describe the CI skip/dependents/debuginfo fix, and that part is implemented as described (.github/actions/load-factory-repo/action.yml, rebuild-rpms.yml:185-216, build-stage.yml:392-425, tools/rebuild_plan.py:81-243). But the branch also carries undisclosed work the body never mentions: nine new package recipes plus .packit.yaml entries (e.g. packages/wpa_supplicant/wpa_supplicant.spec, .packit.yaml:1915-1920), packages/ffmpeg/ffmpeg.spec and packages/pipewire/pipewire.spec changes, ICU-77 staging logic (tests/test_icu_staging.py, config/runtime-contract.toml), and docs/skills/repeated-mistakes.md. Only the feat(sources) commit is disclosed as a passenger.
security
- I could not audit the new wpa_supplicant recipe: 904-line spec plus 15 patches touching OpenSSL/PKCS11/WEP behavior (
packages/wpa_supplicant/wpa_supplicant-OpenSSL-Use-pkcs11-provider-when-OPENSSL_NO_ENGINE-i.patchand siblings). They read like Fedora's standard set but I did not verify provenance — human eyes warranted on a network supplicant. - Low:
rebuild-rpms.yml"Resolve the published factory image" step interpolates${{ github.ref_name }}directly into therun:script (branch=$(echo "${{ github.ref_name }}" | tr …)). Safe onpull_request(N/merge) but a push branch name is shell-evaluated before the sanitizingtr; pass it viaenv:instead. Exploit needs write access, hence low. - The new factory-repo trust path is sound: digest-pinned, publish gated off fork PRs (
rebuild-rpms.yml:629-630), token only ever piped todocker login.
docs-currency
- The renamed
bump-upstream-sources.ymlstill opens its PRs withbranch: bump/gnome-sources, a GNOME-only title, and a body claiming "Opened by.github/workflows/bump-gnome-sources.yml" and "the tarball GNOME actually serves" (.github/workflows/bump-upstream-sources.yml:93-101) — stale now that it polls GitHub/GitLab too, and it names a workflow file that no longer exists. packages/libportal/README.mdandpackages/spirv-tools/README.mdare placeholder stubs ("The libportal package").- README.md:44-51 and AGENTS.md correctly document the new skip behavior.
No findings from: style.
— hive: agent=reviewer backend=copilot model=claude-fable-5 copilot=1.0.78
|
The ICU chain is verified end to end. From its build root (job 105962840910): Its built RPM requires All five, each verified by reading the actual build root rather than inferring from a green check:
The chain that produced the original failure is closed at every link: Fedora's Run state: 320 jobs, 309 green, zero failures. Remaining stages, then Standing caveat, unchanged and worth repeating before the result lands: Generated by Claude Code |
|
Closing to consolidate the queue on the merged #178 full-factory candidate. The branch is retained; please re-open or submit a fresh, rebased PR after that build establishes the current baseline. |
1 similar comment
|
Closing to consolidate the queue on the merged #178 full-factory candidate. The branch is retained; please re-open or submit a fresh, rebased PR after that build establishes the current baseline. |
Why every run took 5–8 hours
preparedecides what is already built by reading a repository listing, and every build root enabled the same URL as[factory]. That URL was the GitHub Pages mirror the README says was removed. The deployment was never taken down: it still answers, frozen at 67 packages from before the 343-recipe contract, and nothingpublishwrites can reach it. Runs 375 and 377 both log:So every push, PR and nightly was a full traversal of all eleven waves — 282 stage-0 jobs, then webkitgtk alone for 5.2 hours at stage 5 with nothing else running, then the tail. Eight hours for a one-line spec fix.
What changes
The image is the witness (
rebuild-rpms.yml,build-stage.yml, new.github/actions/load-factory-repo)prepareresolvesghcr.io/<owner>/utah-packagesto a digest once — the ref's own tag first, thenlatest;mainreads onlylatest— extracts/repositoryfrom it and runs the plan against that on disk.preflightextract the same digest and enable it asfile:///work/factory. The plan and the build root read the same bytes; nothing is fetched over HTTP mid-build.vars.FACTORY_REPOis gone — a second source of truth is how the halves first disagreed.rebuild_plan.pyalready treats as "skip nothing".Rebuilt packages drag their published dependents (
tools/rebuild_plan.py,tools/rebuild_matrix.py)mutterwould buildmutteralone and leave the publishedgnome-shelllinked against the libmutter it replaced. The plan now reads Provides/Requires from the sameprimary.xmland rebuilds the transitive runtime closure of anything being rebuilt, in inventory order so waves still hold. The repo carries no SRPMs, so a devel-only BuildRequires is the residual gap.No debuginfo — the upload already excludes every
*-debuginfo-*/*-debugsource-*by pattern, so they were built, stripped and compressed only to be discarded: 25 min offind-debuginfoplus 10 min of compression on webkitgtk alone. Both lanes now pass--define "debug_package %{nil}".Hygiene
precedenceandpublishrequireprepareto have succeeded; they used to run after a failed prepare and die on a missing artifact (372, 379).pull_requestruns cancel their predecessor on a new push (369, 370, 374 were all cancelled by hand, one four hours in).mainand the schedule are untouched.Also carries the earlier commit on this branch:
feat(sources): poll git forges for releases, not just GNOME.Verification
python3 -m unittest discover -s tests), including newtests/test_factory_witness.py(workflow wiring) andDependentsTestsintests/test_rebuild_plan.py(closure).tools/check_workflow_quoting.pyandtools/factory_contract.pypass.rebuild_plan.plan()against the Pages listing with an empty diff still schedules 275 of 340, all "absent from the listing".Expect one more near-full build: the first run after this reads
latest, which is still the ~60-package image from before #70 (run 377 onmainis publishing the 333-package one now). After that, a one-recipe PR is a handful of jobs.🤖 Generated with Claude Code
https://claude.ai/code/session_01W6BYLUD6KBzF8rBFXKjP7M