Repository navigation
reconcile codecs-media FFmpeg family in build backlog (#474) - #815
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. Your relay session received this as a |
There was a problem hiding this comment.
Summary: This change moves six FFmpeg-related package names out of the "still to build" list and into the "recipe exists" list, and regenerates the summary report. The bookkeeping itself is right — the counts add up and the repository's own checker and tests pass. Two small record-keeping gaps remain: the new entries leave out the cross-reference field the project's own guide asks for, and the explanatory note left in the list points readers at "the PR comment" without saying which one, while describing a state that will go stale.
What I need from you: Apply the two suggestions below (or say why not).
Reviewed at head c24114d.
style
[resolved] entries omit the pr field. docs/skills/factory-build-backlog.md:83-85 says a recipe-only record is { name = "<n>", pr = <n> } and to populate pr before requesting review; :226-228 says to record the existing import PR when reconciling an already_recipe name. Every neighbouring recipe-only entry carries one (config/factory-build-backlog.toml:344, :361-363). ffmpegthumbnailer was imported by #414 (commit fd8282c); ffmpeg's import commit 6ed99b9 has no PR number, so this reconciliation PR is the closing record.
{ name = "ffmpeg", pr = 815 },
{ name = "ffmpeg-libs", pr = 815 },
{ name = "ffmpegthumbnailer", pr = 414 },
{ name = "ffmpegthumbnailer-libs", pr = 414 },
{ name = "libavcodec", pr = 815 },
{ name = "libavdevice", pr = 815 },
(replaces config/factory-build-backlog.toml:355-360)
docs-currency
config/factory-build-backlog.toml:40-44: "see the PR comment" names no PR, and "The ffmpeg family below was reconciled to [resolved]" sits directly above libavfilter, libavformat, libavutil (:49-51) plus libswresample/libswscale, which are the same spec's already_recipe outputs and are not moved (they belong to a later batch, not #474). docs/skills/factory-build-backlog.md:87-89 asks catalog comments to describe structure, not current occupancy. Suggested replacement for lines 40-44:
# #474 slice: LCEVCdec (license removed from Fedora rawhide), davs2-libs
# (not in Fedora dist-git) and gstreamer1-plugins-good-qt6 (needs
# `--with qt6`) remain blocked; see PR #815. The remaining libav*/libsw*
# outputs of packages/ffmpeg belong to a later codecs-media batch.
intent-alignment (advisory, non-blocking)
#474 acceptance asks for "equivalent existing publication evidence" when no implementation is needed; none is attached, and the PR body itself gates merge on the hosted ffmpeg build publishing NEVRAs. The repo's convention explicitly permits recipe-only [resolved] records with remaining checks stated (docs/skills/factory-build-backlog.md:91-92, :229-232), which this PR does, so no change requested — just noting the author-stated merge gate for whoever merges.
No findings from: correctness, security.
Confidence: 3/5 (needs attention) — 2 perspectives requested changes
— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
|
@kylerankin the hive has adopted this PR ( |
…and record pr on resolved ffmpeg 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>
|
1 of 2 commits on |
There was a problem hiding this comment.
Content looks correct to me; it needs a rebase before it can land, and one bookkeeping nit.
Verified against the tree (main @ 4dd1a40):
- The auditor already classifies
ffmpeg,libavcodec,libavdevice,ffmpegthumbnailer,ffmpegthumbnailer-libsasalready_recipeinreports/factory-build-backlog.jsonon main —packages/ffmpeg/ffmpeg.spec:430,476declarelibavcodec%{?pkg_suffix}/libavdevice%{?pkg_suffix}under the default-taken%if ! %{with freeworld_lavc}branch, exactly the casetools/factory_build_backlog.py:208-218documents. Moving them to[resolved]is consistent with the auditor's own view. ffmpeg-libswas the onlypendingone of the six. Its resolution is backed by the existing equivalence contract atconfig/multimedia-closure.toml:303-311([equivalent.ffmpeg-libs]→libavcodec-free), so this is not a new claim introduced by this PR.- The three left open match the auditor's reasoning:
gstreamer1-plugins-good-qt6is the%if %{with qt6}example attools/factory_build_backlog.py:208-213;LCEVCdecanddavs2-libshave nopackages/directory. python3 tools/factory_build_backlog.py --checkpasses on the PR head (201 entries, 193 pending), as the body states.
Blocking: merge conflict. reports/factory-build-backlog.json conflicts with #816/#786/#783, which landed on main after this branch. config/factory-build-backlog.toml merges cleanly. I rebased locally and regenerated with python3 tools/factory_build_backlog.py --output reports/factory-build-backlog.json; the result passes --check (198 entries, 190 pending). This is a fork branch, so I can't push that — @kylerankin, a rebase + regen is all it needs.
Nit (low): the pr = field on [resolved] entries elsewhere points at the PR that imported the recipe (libavc1394/libraw1394 → 681, libdc1394 → 794, config/factory-build-backlog.toml:346-348). The new ffmpeg, ffmpeg-libs, libavcodec, libavdevice entries use pr = 815, i.e. this bookkeeping PR; the recipe itself came in with commit 6ed99b9 ("add the 152 recipes") and its libav* split was last reworked in #414 — which is what the ffmpegthumbnailer* entries in this same hunk already cite. Pointing all six at 414 would keep provenance consistent.
Note, not a defect: libavfilter, libavformat, libavutil, libswresample, libswscale stay in the codecs-media backlog with the identical already_recipe state from the same packages/ffmpeg recipe (ffmpeg.spec:522-697). The split follows #474's nine-name scope, which is fine; a maintainer may prefer to resolve the whole family in one move rather than a later batch.
Also: the body asks to hold merge "after the hosted ffmpeg build publishes its NEVRAs" — that is a human gate, and the title lacks the repo's type-emoji prefix.
I have not approved or changed anything.
Confidence: 3/5 (needs attention) — 1 perspective requested changes
— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
…ec outputs still open in the build backlog; six already have factory recipes, so this reconciles them out of the backlog and records why the other three stay open. What I need from you: Merge after the hosted ffmpeg build publishes its NEVRAs, then add the run URLs to the six resolved entries. --- Moved the FFmpeg family (ffmpeg, ffmpeg-libs, ffmpegthumbnailer, ffmpegthumbnailer-libs, libavcodec, libavdevice) from the codecs-media backlog area to [resolved] as recipe coverage. Each has a recipe under packages/ and a source lock, so it builds on the hosted packit/mock workflow; their public NEVRAs are recorded once that build publishes, not before. ffmpeg-libs is negativo17's single bundle name for the split libav* libraries the ffmpeg recipe ships (libavcodec-free, libavdevice-free, ...), which satisfy it. Regenerated reports/factory-build-backlog.json (backlog 207 -> 201, resolved 307 -> 313). The catalog's 551-name digest contract is preserved (union of areas + [resolved] + [wontfix] unchanged). - LCEVCdec: removed from Fedora rawhide (BSD-3-Clause-Clear is not an allowed Fedora license); not importable from dist-git. - davs2-libs: not present in Fedora dist-git; needs an external source import. - gstreamer1-plugins-good-qt6: its qt6 subpackage is behind `--with qt6`, which this factory does not pass; needs a build-config change. Full-codec (freeworld) ffmpeg is licensing-approved in projectbluefin#383 but still needs its freeworld codec deps (x264-libs, x265-libs, libfdk-aac) and a hosted build. Hive-Run: projectbluefin#474 Hive-Plan: 474-codecs-media-ffmpeg-reconciliation Hive-Spec: project.writing_guide#summary-block Signed-off-by: kylerankin <kylerankin@users.noreply.github.com>
…and record pr on resolved ffmpeg 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>
The ffmpeg recipe was imported via 6ed99b9 and reworked in projectbluefin#414, which the ffmpegthumbnailer entries already cite; match the [resolved] provenance convention instead of self-referencing this PR. 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>
8486631 to
6d611d3
Compare
There was a problem hiding this comment.
Summary: Re-review of the follow-up commits on this backlog cleanup. The earlier conflict with the main branch's report file is gone, and the self-referencing PR number was replaced. One thing still off: four of the six newly "resolved" FFmpeg entries point at a pull request that did not actually bring in the FFmpeg recipe, so the cross-reference the catalog relies on still sends readers to the wrong place.
What I need from you: Change pr = 414 to pr = 70 on the four entries that come from the ffmpeg recipe (details below).
Reviewed delta 8486631 → 6d611d3 (commits d6cdfb6, 6d611d3; the rest of the compare is a rebase onto main picking up #783/#786/#816).
style
config/factory-build-backlog.toml:357-362—docs/skills/factory-build-backlog.md:225-228("Reconciling recipes imported before the audit … Record that existing PR") makesprthe original import PR.packages/ffmpeg/ffmpeg.specwas first added in commit 6ed99b9, which landed via #70; #414 only addedpackages/ffmpegthumbnailer/*. Sopr = 414is correct forffmpegthumbnailerandffmpegthumbnailer-libs(:359-360) but wrong forffmpeg,ffmpeg-libs,libavcodec,libavdevice(:357-358,361-362) — those should bepr = 70. Low severity (the field is informational,--checkdoes not validate it), but this is the same provenance convention the previous round flagged.
intent-alignment
- PR body counts are stale after the rebase: body says "backlog 207 -> 201, resolved 307 -> 313" and "201 entries, 193 pending"; the diff at head is
reports/factory-build-backlog.json:1043-1050backlog 204 → 198, resolved 310 → 316, andtools/factory_build_backlog.py --checkat 6d611d3 prints "198 entries, 190 pending". Advisory — update the body so the numbers match what merges.
Previous findings: correctness snapshot conflict with main — fixed (rebased; --check passes at head). Style pr = 815 self-reference — replaced, but still inaccurate for 4/6 entries (above).
No findings from: correctness, security, docs-currency.
Confidence: 3/5 (needs attention) — 1 perspective requested changes
— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
…ries ffmpeg's recipe was first imported in projectbluefin#70; projectbluefin#414 only added ffmpegthumbnailer, so keep pr = 414 just for the thumbnailer names. Hive-Fix-Attempt: 3/6 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
|
Merged by hive contributor-lane auto-merge: CI green for ≥30m, reviewer verdict approve, head |
Summary: Utah's media package batch (#474) had nine codec outputs still open in the build backlog; six already have factory recipes, so this reconciles them out of the backlog and records why the other three stay open.
What I need from you: Merge after the hosted ffmpeg build publishes its NEVRAs, then add the run URLs to the six resolved entries.
What changed
Moved the FFmpeg family (ffmpeg, ffmpeg-libs, ffmpegthumbnailer,
ffmpegthumbnailer-libs, libavcodec, libavdevice) from the codecs-media backlog
area to
[resolved]as recipe coverage. Each has a recipe underpackages/and a source lock, so it builds on the hosted packit/mock workflow; their
public NEVRAs are recorded once that build publishes, not before.
ffmpeg-libsis negativo17's single bundle name for the split libav* librariesthe ffmpeg recipe ships (libavcodec-free, libavdevice-free, ...), which satisfy
it.
Regenerated
reports/factory-build-backlog.json(backlog 204 -> 198, resolved310 -> 316). The catalog's 551-name digest contract is preserved (the union of
areas +
[resolved]+[wontfix]is unchanged).Why the other three stay open
allowed Fedora license); not importable from dist-git.
import.
--with qt6,which this factory does not pass; needs a build-config change.
Full-codec (freeworld) ffmpeg is licensing-approved in #383 but still needs its
freeworld codec deps (x264-libs, x265-libs, libfdk-aac) and a hosted build.
Verification
python3 tools/factory_build_backlog.py --check— snapshot consistent(198 entries, 190 pending).
just check— multimedia closure, workflow quoting, runtime contract andbacklog snapshot all pass.
just test— 1396 passed, 2 skipped.Refs #474
— hive: backend=pi model=lemonade/ornith-1.5-35b-a3b-gguf-q6_k