Skip to content

fix: check that installed video decoders can load - #919

Closed
arteeh wants to merge 5 commits into
projectbluefin:mainfrom
arteeh:hive/issue-12-multimedia
Closed

arteeh wants to merge 5 commits into
projectbluefin:mainfrom
arteeh:hive/issue-12-multimedia

Conversation

@arteeh

@arteeh arteeh commented Oct 25, 2026 •

Copy link
Copy Markdown
Contributor

Summary: Utah now checks that its installed video software can load the decoders for H.264 and AV1, two common video formats. This catches broken video support that a package-presence check would miss. The contributor guidance also reflects the media packages that have already arrived.


Refs #12 — partial: the complete factory multimedia closure, full FFmpeg, sample playback, and hardware decoding proof remain open under #12 and #383.

Changes:

Validation:

  • python3 -m pytest tests/test_verify_rpm_contract.py -q: 204 passed.
  • Pre-commit hooks for all changed files: passed.
  • pre-commit run --all-files: ran; existing end-of-file issues in two unrelated files required automatic fixes before a clean rerun. Those unrelated edits were restored and are excluded from this PR.
  • just check: attempted after initializing the pinned extension submodules. It reached the full unittest suite, reported three failures, then stalled in a screenshot test; the run was terminated. The focused RPM verifier suite passes independently.
  • No local image build or real decoder probe was possible because this host lacks Podman. Image composition must confirm the decoder probes against the pinned factory; these checks establish element availability, not successful playback or GPU support.

— hive: backend=codex model=gpt-6.1-sol effort=medium

Refs #12


🐝 Hive Agent: contributor | SHA: 7360b37

Probe the contracted H.264 and AV1 GStreamer elements during image verification with a fresh registry. Correct stale media publication guidance.

Assisted-by: GPT-6.1 via GitHub Copilot
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Hive-Run: projectbluefin#12
Hive-Plan: multimedia-capability-verification
Hive-Spec: issue-12#runtime-verification
Signed-off-by: arteeh <arteeh@users.noreply.github.com>
@github-actions

github-actions Bot commented Oct 25, 2026 •

Copy link
Copy Markdown
Contributor

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by:
To complete the pull request process, please assign ahmedadan after the PR has been reviewed.
You can assign the PR to them by writing /assign @ahmedadan in a comment when ready.

Needs approval from an approver in each of these files:

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

@hivecommons-hive hivecommons-hive Bot added author/clanker A contributor's own hive authored this PR under their login dco All commits are DCO signed-off role/trusted Author is trusted for unattended merge (trusted-author role or registry/claim tier in require_tiers) stage/050-review-pending Review swarm requested; awaiting a verdict for the current head labels Oct 25, 2026

@hivecommons-hive hivecommons-hive Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Summary: This change makes the image build actually try to load the two installed video decoders (H.264 and AV1) instead of only checking that their packages exist, and it refreshes the contributor notes to say those packages have arrived. The code and tests look right. One problem: the pull request is marked as fully closing the long-running "ship Bluefin's media stack" issue, while its own text and the docs it edits say that issue must stay open for the remaining work. Merging as-is would auto-close that issue prematurely.

What I need from you: Change Closes #12 to Refs #12 in the PR body (or confirm #12 should close now).


intent-alignment

  • medium — closing keyword contradicts the PR's own scope. The body says "Refs #12 — partial: the complete factory multimedia closure, full FFmpeg, sample playback, and hardware decoding proof remain open under #12 and #383", and the doc this PR edits says "Keep #12 and #383 open for the remaining factory closure, full FFmpeg and runtime playback/hardware proof" (docs/skills/package-contract.md:127-129). Yet the body also ends with Closes #12, which GitHub will honour on merge against main. #12's acceptance list still has "The multimedia contract is included in Utah's package transaction" (full FFmpeg and its codec tail not yet consumed) and only the software-decoder half of "representative codec capabilities" lands here (scripts/verify-rpm-contract.py:983-986 probes avdec_h264/dav1ddec only, by design). Fix in the PR body:
    -Closes #12
    +Refs #12

No findings from: correctness, security, style, docs-currency.

Confidence: 3/5 (needs attention) — 1 medium finding; 1 perspective requested changes

— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88

@hivecommons-hive hivecommons-hive Bot added stage/072-returned-to-contributor Contributor PR: CI red, changes requested or conflicting; diagnosis posted, author must act and removed stage/050-review-pending Review swarm requested; awaiting a verdict for the current head labels Oct 25, 2026
@hivecommons-hive

Copy link
Copy Markdown
Contributor

@arteeh this PR is back in your court (stage/072-returned-to-contributor). The hive will not merge it until the head changes.

Review asks:

  • the review swarm requested changes

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 fix-pr task and launched a parallel fix session; your current work continues.

Use Refs for partial media work in PR descriptions and commits. PR projectbluefin#919 now references issue projectbluefin#12 without closing its remaining acceptance criteria.

Assisted-by: GPT-6 via GitHub Copilot
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: arteeh <arteeh@users.noreply.github.com>
@hivecommons-hive hivecommons-hive Bot added stage/050-review-pending Review swarm requested; awaiting a verdict for the current head and removed stage/072-returned-to-contributor Contributor PR: CI red, changes requested or conflicting; diagnosis posted, author must act labels Oct 25, 2026

@hivecommons-hive hivecommons-hive Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Summary: This change makes the image build actually try to load the two video decoders (H.264 and AV1) it installs, instead of only checking that their packages are present. I read the code and ran the new tests; the logic is sound and safe. Two small notes below: the real decoder probe has only been exercised by mocked tests so far, and the verifier's own top-of-file description does not yet mention the new check.


correctness

  • (advisory, in-scope) The new tests (tests/test_verify_rpm_contract.py:2958-3002) mock subprocess.run, so /usr/bin/gst-inspect-1.0 avdec_h264 / dav1ddec has not yet been run against a real image; the PR body says no Podman was available. The two on-image calls (Containerfile:274, Containerfile:389) in the pending build_main/build_kernel jobs are the first real exercise. Gate on those builds going green before merge. Element/package pairing (scripts/verify-rpm-contract.py:983-986) and the --check early return (:1350) that keeps the probe off-image both read correctly.

docs-currency

  • (low, in-scope) The module docstring (scripts/verify-rpm-contract.py:2-9) enumerates what the on-image run attests and now omits the decoder probe, while the success line (:1465-1467) and docs/skills/package-contract.md:114-118 were updated. Suggested:
    resolve from an unapproved Fedora repository; the contracted GStreamer
    software decoders (`avdec_h264`, `dav1ddec`) load on-image; the composed
    image exposes only the runtime repositories the manifest allows; and the
    resolved package-origin/NEVRA set is retained as a report with build
    provenance.
    
    (replaces scripts/verify-rpm-contract.py:6-9.)

No findings from: security, intent-alignment, style.

Confidence: 5/5 (safe)

— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88

@hivecommons-hive hivecommons-hive Bot added stage/060-reviewed Review swarm verdict approve; a gate is still pending (CI running, soak not started) stage/072-returned-to-contributor Contributor PR: CI red, changes requested or conflicting; diagnosis posted, author must act and removed stage/050-review-pending Review swarm requested; awaiting a verdict for the current head stage/060-reviewed Review swarm verdict approve; a gate is still pending (CI running, soak not started) labels Oct 25, 2026
@hivecommons-hive

Copy link
Copy Markdown
Contributor

@arteeh this PR is back in your court (stage/072-returned-to-contributor). The hive will not merge it until the head changes.

Failing checks:

  • check-failure

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 fix-pr task and launched a parallel fix session; your current work continues.

@hivecommons-hive

Copy link
Copy Markdown
Contributor

@arteeh the hive has adopted this PR (stage/074-adopted-by-hive): CI stayed red for 2h after the nudge and the relay nudge went unanswered. The hive will rebase this branch on main and fix CI here; your commits and sign-off stay. Push a new head yourself at any time and the hive yields it back to you.

@hivecommons-hive hivecommons-hive Bot added stage/074-adopted-by-hive Contributor PR adopted: author unavailable after red CI; the hive is finishing it (takeover path) and removed stage/072-returned-to-contributor Contributor PR: CI red, changes requested or conflicting; diagnosis posted, author must act labels Oct 25, 2026
@hivecommons-hive

Copy link
Copy Markdown
Contributor

The hive adopted this PR (stage/074-adopted-by-hive) but its head dc276d4c2995 has not moved for 3h1m and the hive has nothing it can push: adoption-stalled:check-failure.

Blocker: lane reason check-failure on this head and nothing changed.

Moving to stage/085-needs-decision so a person sees it. This is not a merge-approval wait: the lane has not admitted this head.

@hivecommons-hive hivecommons-hive Bot added stage/085-needs-decision A human decision is owed: sentinel alert, hold/needs-decision label, human-only rule or parked issue stage/080-awaiting-merge-admission CI green, verdict approve, no modifiers; the automerge lane has not admitted it yet and removed stage/074-adopted-by-hive Contributor PR adopted: author unavailable after red CI; the hive is finishing it (takeover path) stage/085-needs-decision A human decision is owed: sentinel alert, hold/needs-decision label, human-only rule or parked issue stage/080-awaiting-merge-admission CI green, verdict approve, no modifiers; the automerge lane has not admitted it yet labels Oct 26, 2026
@hivecommons-hive hivecommons-hive Bot added stage/080-awaiting-merge-admission CI green, verdict approve, no modifiers; the automerge lane has not admitted it yet stage/085-needs-decision A human decision is owed: sentinel alert, hold/needs-decision label, human-only rule or parked issue and removed stage/085-needs-decision A human decision is owed: sentinel alert, hold/needs-decision label, human-only rule or parked issue stage/080-awaiting-merge-admission CI green, verdict approve, no modifiers; the automerge lane has not admitted it yet labels Oct 28, 2026
hivecommons-hive Bot pushed a commit to hivecommons/hive that referenced this pull request Oct 28, 2026
…eason (#15043)

The adoption escalation marked the head noted before posting its note, so
when GitHub answered the adoption-stalled note with a 500 the PR sat at
stage/085-needs-decision with no comment on that head ever
(projectbluefin/utah-packages#650 and #652, 2026-10-26). The same
mark-before-post pattern was in the three workflow-path refusal paths and
in escalateUnadoptableFork.

- stagelabels: MarkAdoptionEscalated records the escalation only; the new
  MarkAdoptionEscalationNoted records (head, reason) after the post
  succeeds. AdoptionEscalationNoted(head, reason) checks both, so a new
  blocker on the same head gets its own note (mirrors EntryCommentExplains,
  #13139). A record from before reasons were kept (noted
  head, no reasons) reads as noted for every reason on that head, so the
  upgrade reposts nothing.
- automerge: all five call sites post through postAdoptionEscalationNote.
  The held-escalation branch of escapeStalledAdoption retries an unposted
  workflow-path refusal note, and hands an unposted or changed stall reason
  to the stall path instead of holding silently.
- Hold an active escalation against a weak opinion: an errored lane on an
  approved head reads 080 at role rank, which the adoption pin does not
  claim, so the pass dropped the escalation for a sweep and
  projectbluefin/utah#919 flipped 085 -> 080 -> 085 on one head (10-26,
  10-28). stageengine.HoldsEscalation is shared by the live pass and
  passDecision so enforce mode derives the same stage.


Claude-Session: https://claude.ai/code/session_01BQYnPGFJx81c8JXR3o1sg1

Signed-off-by: Danathar <Danathar@users.noreply.github.com>
Co-authored-by: Danathar <Danathar@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

@hanthor hanthor left a comment

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Blocking

  • All four image builds fail on the new check: Media decoder 'avdec_h264' (gstreamer1-plugin-libav) failed to load (run 40600999027). The check is right; the image really can't load H.264, probably because the libav plugin is built against ffmpeg-free. Either ship an avdec_h264 that loads, or drop the H.264 probe or make it non-fatal, and then change the docs, which currently say H.264 is covered.
  • The branch conflicts with the base branch and needs a rebase.

Nits

  • Remove GST_PLUGIN_PATH* from the environment with env.pop rather than setting it to an empty string.
  • The assertFalse(registry.exists()) test doesn't show that a broken load is caught.

@hivecommons-hive

Copy link
Copy Markdown
Contributor

/hive decision hive's fixer: rebase onto the base branch, make CI green without widening scope, address the changes @hanthor requested on GitHub on #919; then it merges

Rationale: the hive adopted #919 in place and owns its branch, so the reviewer's finding is the hive's own fixer's to address — not the contributor's and not a person's (hivecommons/hive#15722).

Hive decided: Told the author agent what to change.
Why: The hive adopted this PR; its own fixer addresses the finding.
What happens next: The author agent pushes a new commit with that change — expected within 30 min.
If it doesn't happen: Re-run the decision, give different direction, or close it.

Decided by the hive at @Danathar's request as the owner's delegate under the owner-delegate policy (rule adopted-fixer).

@hivecommons-hive hivecommons-hive Bot removed the stage/085-needs-decision A human decision is owed: sentinel alert, hold/needs-decision label, human-only rule or parked issue label Nov 5, 2026
@hivecommons-hive

Copy link
Copy Markdown
Contributor

Direction recorded by @Danathar for head f0b5e2f. This PR was opened by @arteeh's hive, so the hive treats it as a changes requested review on that head: the PR is returned to the contributor and @arteeh's hive is nudged with the instruction below verbatim. When a new head is pushed the review swarm re-reviews it; if it approves, the PR comes back for your merge approval.

hive's fixer: rebase onto the base branch, make CI green without widening scope, address the changes @hanthor requested on GitHub on #919; then it merges

Recorded from the hive dashboard.

Signed-off-by: Danathar <berths-purger-0g@icloud.com>

# Conflicts:
#	docs/skills/package-contract.md
@Danathar Danathar added the adoption/supersede Owner-delegate request: adopt now in supersede mode label Nov 5, 2026
@Danathar

Danathar commented Nov 5, 2026

Copy link
Copy Markdown
Contributor

Adopting this into a hive branch: the author's agent has been offline since 10-26 and the hive cannot finish it on the fork. I merged main into the branch (one docs conflict, kept main's text plus this PR's probe paragraph) and reviewed the diff for a trusted run: it only runs gst-inspect-1.0 on two decoders inside the image build. On the hive's copy, please address @hanthor's review: make the avdec_h264 probe non-fatal or drop it, since it fails on the current image.

@hivecommons-hive hivecommons-hive Bot removed the adoption/supersede Owner-delegate request: adopt now in supersede mode label Nov 5, 2026
@hivecommons-hive

Copy link
Copy Markdown
Contributor

The hive adopted this PR (stage/074-adopted-by-hive) but its head b740d8667731 has not moved for 1m and the hive has nothing it can push: adoption-stalled:fork-rehome-refused.

Blocker: this PR comes from a fork (arteeh/utah) and the hive cannot finish it on the fork's own branch. Moving its commits to hive/adopt-919 in this repository would run code no maintainer has reviewed with the repository's trusted CI (secrets, write tokens, no fork guards), so the hive did not. Ask (maintainer): review the change; if it should land, bring it into the repository yourself or follow the repository's process for fork pull requests, otherwise close it.

Moving to stage/085-needs-decision so a person sees it. This is not a merge-approval wait: the lane has not admitted this head.

@hivecommons-hive hivecommons-hive Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Summary: This change teaches Utah's image build to prove that two video decoders (H.264 and AV1) actually load, not merely that their packages are installed. Re-reviewed after the branch was brought up to date with main. The merge itself is clean, but two things now block it: the H.264 probe fails on every real image build because the installed decoder plugin genuinely lacks that element, and the merged contributor guide is five lines over its hard size limit, which fails the test suite.

What I need from you: @arteeh — swap the H.264 probe to the decoder the pinned factory actually ships (or drop it until full FFmpeg lands), and trim the guide back under 500 lines.


Delta reviewed: f0b5e2f → b740d86 (merge of up/main; the PR's own files at this head are unchanged apart from the merge resolution in docs/skills/package-contract.md). Prior notes — GST_PLUGIN_SYSTEM_PATH not cleared (scripts/verify-rpm-contract.py:996) and the module docstring omitting the probe (scripts/verify-rpm-contract.py:4) — are still present, unchanged, and remain advisory.

correctness

build_main fails at the new probe on every run of this branch (runs 40227347356, 40600999027; the head run 42112470988 never reached it because contract failed first):

ERROR: 1 supply-chain / repository contract violation(s):
  - Media decoder 'avdec_h264' (gstreamer1-plugin-libav) failed to load: No such element or plugin 'avdec_h264'

MEDIA_DECODERS maps gstreamer1-plugin-libav → avdec_h264 (scripts/verify-rpm-contract.py:983-985). The pinned factory ships libav built against ffmpeg-free, which strips the H.264 decoder, so that element cannot exist until the "full FFmpeg" delivery the guide itself lists as still open (docs/skills/package-contract.md:132-134). main already records this exact failure (docs/skills/package-contract.md:124). As written the PR cannot produce an image. The pin's working H.264 path is openh264dec from gstreamer1-plugin-openh264, which packages/utah.toml:194-195 already requests and whose release identity is asserted:

 MEDIA_DECODERS = {
-    "gstreamer1-plugin-libav": "avdec_h264",
+    "gstreamer1-plugin-openh264": "openh264dec",
     "gstreamer1-plugin-dav1d": "dav1ddec",
 }

and update tests/test_verify_rpm_contract.py:2980,2984,2988 to the new element name. Move avdec_h264 back in with the full-FFmpeg switch (#383). Also: when gst-inspect-1.0 <element> fails, the message hides why the plugin did not load (scripts/verify-rpm-contract.py:1009-1012); re-running gst-inspect-1.0 <plugin> on failure and appending its stderr would have shown the cause here.

docs-currency

The merge brought main's 499-line docs/skills/package-contract.md plus this PR's 6-line paragraph (docs/skills/package-contract.md:126-131) to 505 lines; check-skill-frontmatter hard-fails at 500 (contract job 127543307436: error: docs/skills/package-contract.md is 505 lines (hard max 500)). Compress the paragraph:

The on-image RPM verifier runs `gst-inspect-1.0` on each contracted decoder element with a
fresh temporary registry, so cached metadata cannot hide a broken load; missing tools or
elements, load failures and timeouts fail the build. Off-image `--check` skips these probes.

(Line 124's "probe in PR #919 failed" sentence should also be reworded once the probe is fixed.)

No findings from: security, intent-alignment, style.

Confidence: 2/5 (needs attention) — 1 high finding; 1 medium finding

— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88

@Danathar

Danathar commented Nov 5, 2026

Copy link
Copy Markdown
Contributor

@hanthor, a question for you before anything is closed here.

Where this stands:

  • The author's agent has been offline since 10-26.
  • The hive refused to adopt the PR (fork-rehome-refused).
  • Your review asked for the decoder check to be made non-fatal or dropped.

What the check found is a real bug, though. On the Utah image, the H.264 software decoder doesn't load even though its package is installed. From this PR's image build (run 40600999027):

Media decoder 'avdec_h264' (gstreamer1-plugin-libav) failed to load: No such element or plugin 'avdec_h264'

The AV1 decoder (dav1ddec) loaded fine. docs/skills/package-contract.md on main records the same failure. #383, the issue that covered GStreamer H.264, was closed as completed on 10-28, so nothing open tracks this right now.

Proposal:

  1. Close this PR. A non-fatal check would only turn a real failure into a warning nobody reads.
  2. Track "H.264 decoder doesn't load on Utah" in a new, narrower issue, or reopen Media codec parity: no Intel HW decode, no GStreamer H.264, thumbnailers missing #383 if you'd rather keep it there.
  3. Once H.264 works, the hive can bring the check back as a blocking test.

Closing this PR also frees #12, which the hive currently treats as covered by this PR. Does that work for you?

hivecommons-hive Bot pushed a commit to hivecommons/hive that referenced this pull request Nov 5, 2026
…d branch (#16820)

projectbluefin/utah#919 (fork-rehome-refused on its head) and
projectbluefin/testsuite#1037 (a fork changing .github/workflows/e2e.yml;
the fixer's pushes were refused three times) were decided "hive's fixer: …
then it merges" and read "the hive is fixing it" for days.

- delegatedecision: Input.AdoptedPushRefused; a connected author gets the
  fix list back (return-to-contributor), otherwise recommend-only
  adopted-push-refused keeps the park with the person who asked.
- dashboard: adoptedPushFacts (fork without maintainer edits, fork with a
  workflow-file diff, fork-rehome-refused or workflow-path escalation on the
  head, capability blocker); gathered for the delegate; ownership keeps an
  answered review on such an adopted head your turn (adopted_push_refused).
- glossary (Go + SPA mirror), advisor wording, docs, changelog.


Claude-Session: https://claude.ai/code/session_01BQYnPGFJx81c8JXR3o1sg1

Signed-off-by: Danathar <Danathar@users.noreply.github.com>
Co-authored-by: Danathar <Danathar@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
hivecommons-hive Bot added a commit to hivecommons/hive that referenced this pull request Nov 5, 2026
…16846)

settleInPlaceAdoption yielded on any new head without App push
evidence, so a maintainer rebasing or fixing an offline contributor's
adopted PR handed it back to the absent author, who was then nudged,
re-adopted and stalled again (projectbluefin/utah#919, Danathar's
2026-10-27 push; projectbluefin/utah-packages#804, #812, #822,
hanthor's 2026-10-31 pushes).

Before yielding, classify the GitHub-linked pusher (committer, or the
author when the committer is web-flow or unlinked): the PR author, an
unlinked identity, or a read-only third party still yields; an App bot
or a person with write/maintain/admin (authorRepoPermission) is a
takeover and the adoption follows the new head with no yield note and
no nudge. An unreadable permission keeps the adoption for the sweep.

Signed-off-by: James Reilly <jreilly1821@gmail.com>
Co-authored-by: hivecommons-hive[bot] <280983584+hivecommons-hive[bot]@users.noreply.github.com>
@hanthor

hanthor commented Nov 6, 2026

Copy link
Copy Markdown
Member

@Danathar agreed. A non-fatal probe would hide a real failure, and the author's agent has been offline since 10-26. I'm tracking the decoder in #1116. That issue asks for this load probe to come back as a blocking check once the supported H.264 decoder loads. Thank you @arteeh: this check found the bug. Closing.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

author/clanker A contributor's own hive authored this PR under their login dco All commits are DCO signed-off role/trusted Author is trusted for unattended merge (trusted-author role or registry/claim tier in require_tiers)

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants