Repository navigation
feat: import imath and rygel recipes for the gnome-apps batch (#484) - #823
Conversation
|
@kylerankin this PR is back in your court ( Review asks:
Reply here or push a new head to this branch (same sign-off); CI and the review swarm re-run automatically and the PR returns to the normal path. Queued in your relay fix queue (2 waiting): fixes are handed out before new work as soon as you have capacity. |
There was a problem hiding this comment.
Summary: This change copies two more software recipes (a graphics math library and a home media server) from Fedora into the factory so they can be built later. I checked that both copies are exact and that the recorded download checksums are genuine; they are. Two things need fixing before merge: the description tells GitHub to close the batch issue even though it says the work is only partial, and the new guidance text gives a wrong account of what the math library is needed for.
What I need from you: Drop the closing line and fix the one doc paragraph (suggestions below).
Reviewed at head 9ef4a18e. Both packages/imath and packages/rygel are byte-identical to the Fedora rawhide trees named in their .hummingbird-upstream.json (0ef7aa5b…, 4400ba50…, both current rawhide HEAD), and both sources SHA-512s match dist-git. tools/validate.py, factory_build_backlog.py --check and factory_contract.py pass at head.
intent-alignment
- medium — PR body says
Refs #484 (partial: … hosted build, publication and runtime checks … remain open)and then ends withCloses #484. The closing keyword asserts the batch is done; #484's acceptance list (hosted build run URLs, NEVRAs, Hummingbird-only install, runtime probe) is explicitly not met here, and #484 itself says "Do not report blocked work as completed." Every prior partial PR on this batch (#747, #760, #786, #816) usedRefsonly. Fix: delete theCloses #484line from the PR body. (Side note: the hive handed me #483 as the linked contract; neither #483 nor this PR's scope involves imath/rygel, so I judged against #484.) - info — PR body says the imath lock is
https://src.fedoraproject.org/repo/pkgs/rpms/imath/...;config/upstream-sources.json:7actually locks the GitHub release URL (matchingimath.spec:13Source0). The JSON is right; the body is wrong.
docs-currency
- medium —
docs/skills/factory-build-backlog.md:868-869says imath is what "the OpenEXR family (openexr,openxr) builds on; it is the cross-batch dependency behind gnome-apps #483."openxris an unrelated VR API, neither name is a recipe inpackages/, the only OpenEXR output in the catalog isopenexr-libs, which sits pending in[areas.long-tail](config/factory-build-backlog.toml:264), and #483 does not mention imath or OpenEXR at all. Suggested replacement:
`imath` is the C++ vector/matrix library that OpenEXR builds on; the pending
`openexr-libs` output in `[areas.long-tail]` will depend on it.
No findings from: correctness, security, style.
Confidence: 3/5 (needs attention) — 2 medium findings
— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
|
@kylerankin the hive has adopted this PR ( |
|
1 of 2 commits on |
|
Hive filed review findings that do not block this PR as backlog items:
Cap: 10 backlog item(s) per PR per day. |
hanthor
left a comment
There was a problem hiding this comment.
The recipes match Fedora rawhide HEAD unchanged, and both upstream SHA-512s match their locks.
config/factory-build-backlog.toml:322-325anddocs/skills/factory-build-backlog.md:881-884say rygel can't build untilgupnp-avis imported. That isn't right: Fedora 44 ships gupnp-av 0.14.5, and the existinggupnp-dlnarecipe already BuildRequiresgupnp-av-develwithout a factory recipe. Please drop the blocker wording, or reword it as a runtime-availability note.- The PR body says the imath lock uses the Fedora lookaside, but the committed URL is the GitHub archive. The hash matches either way.
- It needs a rebase. Only the generated files conflict, so regenerate them.
Heads-up: the docs builds in these recipes run doxygen and Sphinx. serd (#773) failed on main for exactly that reason (#826), so this needs a hosted build before it lands.
67d51d6 to
fd0d60b
Compare
|
stage 080-awaiting-merge-admission → 074-adopted-by-hive: required-check-failing:Unit tests |
There was a problem hiding this comment.
Summary: Re-review of the imath and rygel recipe import after it was rebased onto the latest main. The recipes themselves are unchanged and fine, but the rebase dropped two bookkeeping updates the branch used to carry: the package-count figures in the architecture overview and the totals in the build-backlog report. Two CI checks now fail on this branch because of that. Both are fixed by rerunning the repository's generator scripts.
What I need from you: Regenerate the two files and push.
Reviewed delta 67d51d6 → fd0d60b (23 commits; 20 are main via rebase, PR content commits unchanged). Previous rygel finding: still present, disclosed — packages/rygel/rygel.spec:79 %systemd_user_post sits inside %install with no %post header; PR body line 33 discloses it as a carried Rawhide defect, status unchanged from prior approve.
correctness
reports/factory-build-backlog.json:921-925— the rebased branch no longer updatesstates.pending/totals(previous head had a hunkpending 188→186, backlog 196→194, resolved 318→320; it is gone).python3 tools/factory_build_backlog.py --checkat head:totals drift: live {'backlog': 173, 'resolved': 341} vs snapshot {'backlog': 176, 'resolved': 338}. Main (cf586df) already drifts by 1; this PR adds 2 more. CI job "Factory build backlog" fails. Fix:python3 tools/factory_build_backlog.pyto refresh the snapshot.
docs-currency
docs/architecture.md:81,133,137-139,144,163— all counts still read 669 while the tree has 672 recipes after this PR. Previous head bumped them (653→655 at the time); the rebase dropped that hunk.tests/test_architecture_counts.pyfails 7 tests (669 != 672,validated 669 source RPMs (664 rawhide, 5 upstream)vs672 … (667 rawhide …)), andtest_render_architecture_counts.pyfails. Main is already stale by 1 (669 vs 670); not solely this PR's fault, but the PR removed its own fix. Fix:python3 tools/render_architecture_counts.py --write.
No findings from: security, intent-alignment, style.
Confidence: 3/5 (needs attention) — 2 medium findings
— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
|
stage 080-awaiting-merge-admission → 074-adopted-by-hive: pending: head pushed 1m42s ago (< min_head_age) |
There was a problem hiding this comment.
Summary: Re-reviewed only what changed since the last look at this pull request, which imports two new GNOME package recipes (a math library and a media-sharing server). The two earlier blockers — stale package counts in the architecture document and a backlog report that had not been regenerated after a rebase — are both fixed now. One known quirk inherited verbatim from Fedora's recipe remains, is disclosed by the author, and does not block.
Reviewed at head 3c8b4eaa49e80f6208b7e1472665c79402520339 (delta from fd0d60ba: docs/architecture.md, reports/factory-build-backlog.json).
Previous findings:
reports/factory-build-backlog.json:921-926totals not regenerated — fixed.python3 tools/factory_build_backlog.py --check→ "snapshot consistent (173 entries, 165 pending)".docs/architecture.md:81,133,137-139,144,163counts stale — fixed. Verified at head:ls -d packages/*/ | wc -l= 672,.packit.yaml:packages= 672,upstream-sources.json:packages= 672, 95 staged entries,tools/validate.py→ "validated 672 source RPMs (667 rawhide, 5 upstream)";tools/render_architecture_counts.py --writeproduces no diff.packages/rygel/rygel.spec:79%systemd_user_postinside%installwith no%postheader — still present, advisory only. It is byte-identical to Fedora rawhiderygel.spec:79, the import convention keeps Rawhide specs unchanged, and the PR body discloses the consequence (user unit not preset-enabled on install). Not blocking; worth a follow-up to carry a%postfix upstream or as a patch.
correctness
- low, in-scope:
packages/rygel/rygel.spec:79— see above; carried, disclosed, advisory.
No findings from: security, intent-alignment, style, docs-currency.
Confidence: 5/5 (safe)
— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
|
stage 080-awaiting-merge-admission: no activity for 30m0s: merge-train:queued-behind#819:2/5:0/0:.packit.yaml,factory-build-backlog.toml,upstream-sources.json,architecture.md,+2 |
…tbluefin#484) Two gnome-apps outputs from the factory backlog still had no recipe: imath, the C++ vector and matrix library the OpenEXR family builds on, and rygel, a UPnP/DLNA home media server. This imports both from Fedora rawhide, locks their upstream tarballs by SHA-512, registers them in the factory Packit pipeline, and moves them out of the pending gnome-apps list. - imath 3.1.12 (Imath-3.1.12.tar.gz) ships the imath and python3-imath binaries from one source, with two patches that disable failing Python bindings tests. - rygel 46.1 (rygel-46.1.tar.xz) emits a rygel-localsearch subpackage and a systemd user service, and depends only on libraries already in the factory. Both are recipe imports; their hosted build, publication and runtime proof stay on projectbluefin#484, recorded in the backlog as recipe-import-only. epiphany-runtime remains pending in gnome-apps but is outside this batch's named outputs. Hive-Run: projectbluefin#484 Hive-Plan: gnome-apps-imath-rygel-recipe-import Hive-Spec: factory-build-backlog#484 Signed-off-by: Danathar <berths-purger-0g@icloud.com>
rygel BuildRequires pkgconfig(gupnp-av-1.0), but gupnp-av has no factory recipe and is still pending in the long-tail area. Record it as a concrete blocker in the backlog config comment and the skill doc, note tinysparql-3.0 comes from the Hummingbird base, and document that the carried Rawhide spec has %systemd_user_post inside %install with no %post section. Hive-Fix-Attempt: 1/6 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Drop the openxr and gnome-apps projectbluefin#483 references from the imath skill paragraph, state that boost-devel and python3-devel come from the Fedora 44 buildroot, and note that sqlite3 and tinysparql-3.0 come from the Hummingbird base rather than factory recipes. Record pr = 823 on the imath and rygel resolved entries. Hive-Fix-Attempt: 2/6 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Rerun tools/factory_build_backlog.py and tools/render_architecture_counts.py --write so the committed snapshot and docs/architecture.md inventory counts match the tree with imath and rygel imported (672 recipes, backlog 173/165 pending). Hive-Fix-Attempt: 4/6 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
3c8b4ea to
fbb1d30
Compare
|
Summary: This pull request adds two new GNOME package recipes, imath and rygel. It had fallen out of date with the main branch, so I rebased it on the latest main and fixed the conflicts. The generated counts and backlog report were regenerated with the repo's own tools. Rebased onto origin/main. Conflicts were only in hive/awaiting-ci: CI pending — sweep will merge when green. 🐝 Hive Agent: — hive: agent=scanner backend=copilot model=claude-opus-5.5 copilot=1.0.88 |
|
The hive adopted this PR ( Blocker: this PR comes from a fork ( Moving to |
There was a problem hiding this comment.
Both imports look correct to me; verified against Fedora and upstream bytes.
What I checked (head fbb1d30)
packages/imath/{imath.spec, imath-disable-python-testM44fArray.patch, imath-disable-python-testPlane.patch}are byte-identical to Fedora rawhidec1465703ad8209513b76cfa6a39a019ff0a19c98;packages/rygel/{rygel.spec, sources, .packit.yaml, monitoring.toml}are byte-identical to rawhidef79fad7efc42d7e767144a0e61c9497fdab2ce20.- Downloaded both primaries and computed SHA-512 myself:
Imath-3.1.12.tar.gzfromgithub.com/AcademySoftwareFoundation/Imath/archive/v3.1.12/→32628dfc…8343;rygel-46.1.tar.xzfromdownload.gnome.org→186db4a1…726a7. Both match the locks inconfig/upstream-sources.jsonand the importedsourcesfiles. - Per-package
.packit.yaml/monitoring.tomlcarried from dist-git have precedent (40 and 30 existing packages respectively), and ~219 existing locks also omitfallback_urls, so neither is a convention break. [resolved]entries use the same recipe-import-only shape aslibpasswdqc/libdeflate; backlog JSON counts move consistently.
Two body/diff mismatches — the diff is the right side of both:
- Body: "Lock
https://src.fedoraproject.org/repo/pkgs/rpms/imath/Imath-3.1.12.tar.gz/sha512/…". Diff: the primaryurlfor imath is the GitHub release archive (config/upstream-sources.json, imath entry). The diff followsAGENTS.md:155-158(upstream primary, never Fedora); the prose describes a lookaside lock that isn't there. - Body: "
tools/validate.py: passed (655 source RPMs, 650 rawhide, 5 upstream)". Diff:docs/architecture.mdsays 677/672. The body figure is from an older base.
Merge-order note: #844, #819, #834, #836 also bump the docs/architecture.md package-count table and edit factory-build-backlog.toml/.json; whichever lands after this will need a count rebase.
The rygel gupnp-av build blocker is stated plainly in the toml comment and skill doc, which is the honest disposition for a recipe that can't build yet.
Confidence: 5/5 (safe)
— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
|
Merged by hive contributor-lane auto-merge: CI green for ≥30m, reviewer verdict approve, @kylerankin holds the trusted tier in this hive's contributor registry. |
Main gained imath and rygel (#823), bringing the inventory to 684 packages. Regenerate docs/architecture.md counts via tools/render_architecture_counts.py --write and the backlog snapshot via tools/factory_build_backlog.py. Hive-Fix-Attempt: 5/6 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
* feat: add seven missing image and video codec libraries Import SHA-512-locked upstream payloads with pinned Fedora, RPM Fusion and negativo17 recipes. Build x264 before FFmpeg and retain the VVC decoder test archive through the RPM Fusion source lookaside. Reconcile the recipe backlog without claiming runtime publication. Hive-Run: #476 Hive-Plan: codecs-media-source-imports Hive-Spec: issue-476#implementation-3 Assisted-by: GPT-6.1 via GitHub Copilot Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: arteeh <arteeh@users.noreply.github.com> * fix: bind provenance package to recipe dir and record PR on resolved codecs Reject provenance whose package field differs from its recipe directory so the master-branch remote guard cannot be satisfied by a mislabeled field. Record pr = 627 on the seven codec [resolved] backlog entries. Hive-Fix-Attempt: 1/6 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com> * ci: retrigger in-repo Canary for adopted branch The Canary check on 82728d9 came from fork PR #627's run of the same commit, which cannot publish the canary image. A new commit on this in-repo branch gets its own Canary run. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com> * chore(backlog): record PR 633 for codec resolved entries Hive-Fix-Attempt: 2/6 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com> * fix: record rpmfusion/negativo17 codec imports as upstream provenance tools/validate.py only accepts rawhide (Fedora dist-git) or upstream imports. vvdec, vvenc, x264, x265 and xavs2 come from non-Fedora repos, so record them as branch "upstream" with empty commit/tree like libfreeaptx, and update the quoted validate output in architecture.md. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com> * fix: restore master-branch provenance for codec imports Revert the relabel of vvdec, vvenc, x264, x265 and xavs2 to branch "upstream" with blank commit/tree. Keep the pinned RPM Fusion / negativo17 dist-git commit and tree SHAs; validation of the master branch rule depends on #821 as stated in the PR body. Hive-Fix-Attempt: 4/6 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com> * chore: refresh architecture counts and build backlog after main merge Main gained imath and rygel (#823), bringing the inventory to 684 packages. Regenerate docs/architecture.md counts via tools/render_architecture_counts.py --write and the backlog snapshot via tools/factory_build_backlog.py. Hive-Fix-Attempt: 5/6 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com> * fix: restore factory validation baseline Refresh architecture and backlog state from the current inventory. Add one inventory regeneration command, repair invalid imported YAML whitespace, and preserve workflow output values with grouped redirection. Assisted-by: GPT-6.1 via GitHub Copilot Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: James Reilly <jreilly1821@gmail.com> * ci: keep rebuild-rpms.yml identical to main Drop the style-only grouped GITHUB_OUTPUT edit carried in from the interim baseline branch; main intentionally kept the original workflow so recipe PRs do not touch pipeline paths. Assisted-by: GPT-6.1 via GitHub Copilot Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: James Reilly <jreilly1821@gmail.com> --------- Signed-off-by: arteeh <arteeh@users.noreply.github.com> Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com> Signed-off-by: James Reilly <jreilly1821@gmail.com> Co-authored-by: arteeh <arteeh@users.noreply.github.com> Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Co-authored-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com> Co-authored-by: James Reilly <jreilly1821@gmail.com>
Summary: Two GNOME-apps outputs from the factory backlog still had no recipe: imath, the C++ vector and matrix library the OpenEXR family builds on, and rygel, a UPnP/DLNA home media server. This imports both from Fedora rawhide unchanged, locks their upstream tarballs by SHA-512, registers them in the factory Packit pipeline, and moves them out of the pending GNOME-apps list. It prepares them for building; it does not yet prove that users can install or use them.
Refs #484 (partial: imports the imath and rygel recipes; epiphany-runtime still needs a build in this batch, and the outstanding hosted build, publication and runtime checks for every output remain open). Consumer tracking: projectbluefin/utah#391.
What changed
imath3.1.12 unchanged at commitc1465703ad8209513b76cfa6a39a019ff0a19c98, tree0ef7aa5b0344efc92cd878c1794d1ca0f0619efb, usingtools/import_rawhide.py. It ships theimathandpython3-imathbinaries from one source, with two patches that disable failing Python bindings tests.rygel46.1 unchanged at commitf79fad7efc42d7e767144a0e61c9497fdab2ce20, tree4400ba50162ba9887c21efc067af2a57552f5176, usingtools/import_rawhide.py. It emits arygel-localsearchsubpackage and a systemd user service. Itspkgconfig(gupnp-av-1.0)BuildRequires has no factory recipe yet (see blocker below).https://download.gnome.org/sources/rygel/46/rygel-46.1.tar.xz. Its downloaded SHA-512 matches the importedsources:186db4a17dfbeb5ba67300265c7d33e9afb9f6d5658969c6fa611da6e23c34fd6a3f0d7600ba468f63649a05ca462883018f43478d181d9008d58821e2e726a7.https://src.fedoraproject.org/repo/pkgs/rpms/imath/Imath-3.1.12.tar.gz/sha512/32628d...(importedsourcesSHA-512 verified against the download)..packit.yaml, regenerate architecture counts and the backlog report, and moveimathandrygelto[resolved]for recipe-import accounting only. The original audit-name partition is preserved;epiphany-runtimestays pending in GNOME-apps.Output disposition
imathpackages/imath/imath.spec, 3.1.12python3-imathrygelpackages/rygel/rygel.spec, 46.1rygel-localsearchrygel-tracker; no separate import.epiphany-runtimeDependency and repository evidence
epiphany-runtime,imathandrygelas pending in the GNOME-apps area, andimathandrygelhad no entry inconfig/upstream-sources.jsonor a block in the root.packit.yaml.imathorrygel, and there is no waiver inbaselines/triage.toml. Neither appears inconfig/bluefin-packages.toml.imathorrygelsource or binary was tracked before this import.imathis the library OpenEXR builds on;openexr-libsitself is still pending in[areas.long-tail].gupnp-1.6,gupnp-dlna-2.0,gst-editing-services-1.0,gee-0.8,libmediaart-2.0,libsoup-3.0andgtk4have factory recipes;sqlite3andtinysparql-3.0have none and come from the Hummingbird base. imath'sboost-develandpython3-develcome from the Fedora 44 buildroot.pkgconfig(gupnp-av-1.0), butgupnp-avhas no factory recipe — it is still pending in[areas.long-tail](config/factory-build-backlog.toml, CI should fail when an [unavailable] entry becomes installable utah#396). rygel cannot build untilgupnp-avis imported; this is recorded in the backlog config comment and the skill doc.rygel.specplaces%systemd_user_postinside%installwith no%postheader, so the built package has no%postscriptlet and the user unit is not preset-enabled on install. Kept unchanged per the import convention; noted for the deferred user-service runtime probe.Validation
python3 tools/factory_build_backlog.py --check: passed.python3 tools/validate.py: passed (655 source RPMs, 650 rawhide, 5 upstream).python3 tools/factory_contract.py: passed (4879 tracked files checked).python3 tools/multimedia_closure.py --check,check_workflow_quoting.py,check_build_container_pinning.py,runtime_contract.py: passed.pytest tests: 1,396 passed, 2 existing skips.git diff --check: passed.pre-commit run --all-files: unavailable (pre-commitis not installed); the equivalent JSON/TOML/YAML, action-lint, contract and quoting checks were run by hand and passed.gh workflow run, so no run was started and no gate was bypassed.— hive: backend=pi model=lemonade/ornith-1.5-35b-a3b-gguf-q6_k
🐝 Hive Agent:
contributor| SHA:9ef4a18e