Skip to content

prompt: file the three jax-grad-local-vs-ci-assertions follow-ups - #263

Merged
Jammy2211 merged 1 commit into
mainfrom
claude/jax-grad-assertions-ci-hontn3
Aug 22, 2026
Merged

prompt: file the three jax-grad-local-vs-ci-assertions follow-ups#263
Jammy2211 merged 1 commit into
mainfrom
claude/jax-grad-assertions-ci-hontn3

Conversation

@Jammy2211

Copy link
Copy Markdown
Collaborator

Three draft prompts for the follow-ups left open by complete/2026/08/jax-grad-local-vs-ci-assertions.md (PyAutoLabs/PyAutoArray#471), which closed the imaging manifestation of the small-datasets regime collision and deliberately scoped out the rest.

Prompt Tracks Difficulty Priority
draft/bug/pyautonerves/small_datasets_regime_stamp_at_writer_funnel.md PyAutoLabs/PyAutoNerves#153 large high
draft/bug/autoarray/small_datasets_rmtree_of_committed_data.md PyAutoLabs/PyAutoArray#470 medium medium
draft/maintenance/ci/smoke_install_stale_jax_pin.md (not previously filed) low medium

The first is high because the interferometer case it closes fails silently — a capped run writes identical NAXIS with different values, so there is no shape mismatch and no assertion to trip. That is worse than the loud imaging failure that started the investigation, and a shape-based check structurally cannot see it.

The third was never filed anywhere: smoke_install.sh:9's jax<0.7 downgrades jax to 0.6.2 and conflicts with autonerves' own jax<0.11.0,>=0.7.0; the install lands on the intended 0.10.2 only because the next line's [optional] extras pull it back up. CI is correct as a side effect of line ordering.

draft/bug/pyautonerves/ is a new target folder — PyAutoNerves had no prompts under draft/ before. Named to match the existing lowercase convention (pyautohands, pyautoheart, pyautobrain).

Each prompt records where its declared difficulty disagrees with the sizing faculty's derived score, with the reasoning, rather than overriding it silently — the faculty derives too-large/large/medium against declared large/medium/low. Two of those I kept, one I conceded on reflection.

lifecycle check: OK; dashboard regenerated (146 prompts).

🤖 Generated with Claude Code

https://claude.ai/code/session_01VEHLT33XpVcRt5YCJGLRMJ


Generated by Claude Code

Follow-ups from complete/2026/08/jax-grad-local-vs-ci-assertions.md, which fixed
the imaging manifestation of the small-datasets regime collision and left the
rest deliberately open.

- draft/bug/pyautonerves/small_datasets_regime_stamp_at_writer_funnel.md
  (PyAutoNerves#153) — the silent interferometer case; a shape-based check cannot
  see it. Priority high for that reason.
- draft/bug/autoarray/small_datasets_rmtree_of_committed_data.md
  (PyAutoArray#470) — rmtree of committed, gitignore-allowlisted data.
- draft/maintenance/ci/smoke_install_stale_jax_pin.md — not previously filed
  anywhere; CI resolves the intended jax only as a side effect of line ordering.

Each records where its declared difficulty disagrees with the sizing faculty's
derived score, rather than overriding it silently.

Co-Authored-By: Claude <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VEHLT33XpVcRt5YCJGLRMJ
@Jammy2211
Jammy2211 merged commit 8a91d10 into main Aug 22, 2026
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants