marfrit/libva-multiplanar
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

iter7+: drop candidate D (upstreaming) — philosophical, not technical

Browse Source 
Operator's stance, recorded verbatim in
memory/project_no_upstreaming_philosophical.md: the AI-slop-buster
review climate in 2026 open-source maintainership makes submission
cost > benefit when personal requirements are met. Multiple iterations
of substantive work (sonnet pre/post-commit reviews, formal pixel
verification, regression test harnesses, clean commit history) don't
necessarily survive first contact with reviewers who treat
AI-assisted = automatic slop regardless of substance.

Track D was carried iter3+4+5+6 as a possible culminating iteration.
Dropped 2026-05-06.

Distinct from the prior feedback_no_upstream.md rule (which was
procedural — "no PR without explicit instruction"); the new memory
entry records the underlying philosophical reason. Procedurally,
the new instruction is "don't ever, regardless."

Reopen criterion documented in the memory entry: operator may change
mind if the climate softens, or if a trusted maintainer signals
"send it." Until then, none.

Remaining iter8+ candidate: E (performance binding cell) only.
Plus iter7 carries (low priority): STREAMON-on-context-recreate,
pool-size parameterization, fault-inject build for slot-leak.
Plus separate top-level campaigns: fourier-fresnel, panvk-bifrost.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
Claude (noether)
2026-05-06 11:46:33 +00:00
parent ec769a9687
commit 2707725fea
1 changed files with 3 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files
phase0_findings_iter7.md
+3 -9
View File
@@ -52,17 +52,11 @@ Currently when REINIT or DQBUF fails mid-cycle, the slot stays busy=true until `
**Effort**: 2-3 hours including writing test, fixturing bbb's first I-frame as raw H.264 NAL.
### D. Bootlin / Mozilla upstreaming prep (carried from iter3 candidate G + iter4 + iter5 + iter6)
### D. Bootlin / Mozilla upstreaming prep (DROPPED 2026-05-06)
> File the firefox-fourier patch with Mozilla Bugzilla (bug 1833354 / 1965646 reference for V4L2 stateless analogue). File libva-v4l2-request fork's iter1-iter6 patch series with bootlin's libva-v4l2-request maintainer (Paul Kocialkowski) as a coherent series.
Carried iter3+4+5+6 as candidate. **Dropped 2026-05-06** on operator's philosophical-stance grounds: the AI-slop-buster review climate in 2026 open-source maintainership makes the social cost of submission exceed the benefit when personal requirements are already met. See `memory/project_no_upstreaming_philosophical.md` for the operator-verbatim rationale.
**Plan (Mozilla)**: Bugzilla account, write up "V4L2 stateless decoders blocked by RDD+Utility sandbox" with reproducer, attach combined 160-line patch.
**Plan (bootlin)**: structure the iter1-iter6 commits as a clean patch series. Possibly squash some of iter5's instrumentation-removal commits with their original patch landings. Address upstream concerns about per-call diagnostics that survived as `request_log` lines.
**Risk**: socially-mediated. Maintainer may push back on architectural decisions; per `feedback_no_upstream.md` no PR/MR happens without explicit operator instruction.
**Effort**: bug filing 2-4 hours; patch-series prep 4-8 hours.
This is a **stance change** vs the campaign's prior `feedback_no_upstream.md` rule. Procedurally that rule said "no PR/MR without operator instruction"; the new memory entry records that the instruction is now "don't ever, regardless." The campaign deliverables stay on `git.reauktion.de` for personal use + reference.
### E. Performance binding cell (carried from iter1+2+3+4+5+6)