Skip to content

insight: hst-imaging-mass-samplers 2026-10-19 clean-up record (cancel, flat layout, wave-1 attempt 1 aside) - #37

Merged
Jammy2211 merged 4 commits into
mainfrom
insight/hst-cleanup-20261019
Oct 19, 2026
Merged

Jammy2211 merged 4 commits into
mainfrom
insight/hst-cleanup-20261019

Conversation

@Jammy2211

Copy link
Copy Markdown
Contributor

Summary

This PR records the human-directed clean-up of campaign hst-imaging-mass-samplers on 2026-10-19. It is a focused record, not a full sweep, so last_checkin and last_checkin_report stay at the 2026-10-18 sweep (CHECKIN.md step 6 stamps those only for a completed sweep).

It is stacked on #36 (feature/wave1-sliced), which carries the 2026-10-17 and 2026-10-18 records this one builds on.

  • checkins/2026-10-19T171637Z.md follows the durable template. The human's decisions are quoted verbatim. It covers:

    • the scheduler read at 15:49:05Z;
    • 412539 COMPLETED;
    • 412541 and 412565 CANCELLED about 15:51Z, with stage completion at cancel;
    • autolens_inference#36 merged as c3ca4de, and the flatten_cleanslate.py --apply output on RAL and on the laptop (1974 files, identical on both);
    • where the wave-1 attempt-1 material was set aside.

    The other campaigns were not re-swept and their rows are unchanged.

  • campaigns.yaml changes only this campaign's row and its two task rows, all moved to needs-decision. Nothing is queued, and every next step needs human authorisation. Paths now name the flat layout:

    • output/autolens/imaging/hst/slam/<variant>/<config>/seed_<n>/<stage>/
    • output/autolens/prepared/<config>/seed_<n>/<variant>/
    • results/slam/imaging/hst/<variant>/<config>/
  • Task contracts: the wave-1 manifest path now uses the flat layout, and the base task gains a dated note on the layout change and the cancellation.

  • Keeping prepared/ is recorded as an agent proposal awaiting the human's confirmation. It is not recorded as a decision.

No compute was submitted. The only scheduler action was the human-directed scancel.

Checks

  • board --offline: OK
  • ruff check .: OK
  • ruff format --check .: OK
  • pytest: 204 passed
  • check --offline: FAIL, on a pre-existing lens: finding. Several results/searches/point_source/multi_start_adam* rows fail with "dependency revisions must be text". The same FAIL appears on the unchanged feature/wave1-sliced checkout, so it is not caused by this diff. The campaigns, dashboard and state checks pass.

🤖 Generated with Claude Code

Jammy2211 and others added 2 commits October 19, 2026 18:19
…5 cancelled (human), flat layout applied, wave-1 attempt 1 set aside

Focused record (not a full sweep; last_checkin unchanged). 412539 COMPLETED;
412541 and 412565 scancelled ~15:51Z on the human's decision with partial
stages left on disk; autolens_inference#36 merged (c3ca4de) and
flatten_cleanslate.py --apply run on RAL and the laptop mirror (1974 files
identical); wave-1 attempt-1 experiments/experiment_sources/failed rows moved
to the aside folders. Keeping prepared/ recorded as a proposal pending human
confirmation. No compute submitted.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…ate and 412539 elapsed (Cortex PR #73 e3ab356)

From sacct queried by the session architect 2026-10-19 ~17:25Z.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Base automatically changed from feature/wave1-sliced to main October 19, 2026 19:15
Jammy2211 and others added 2 commits October 19, 2026 20:16
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CPVGVBGnnMB6cL4H2gio2g
@Jammy2211
Jammy2211 merged commit 99d103c into main Oct 19, 2026
2 checks passed
@Jammy2211
Jammy2211 deleted the insight/hst-cleanup-20261019 branch October 19, 2026 19:27
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.

1 participant