Skip to content

Small fixes from the record audit: HISTORY dedup, make_cat warning, split and stamp-size tests - #931

Merged
cailmdaley merged 4 commits into
developfrom
fix/record-audit-small
Oct 1, 2026
Merged

cailmdaley merged 4 commits into
developfrom
fix/record-audit-small

Conversation

@cailmdaley

Copy link
Copy Markdown
Contributor

Four small fixes from the decision-record audit.

  • make_post_process deduplicates HISTORY exposures, as find_exposures already does. A repeated card no longer writes a duplicate EPOCH HDU or double-counts N_EPOCH.
  • make_cat logs a real warning when fewer than 10% of detections have ngmix rows, then continues with sentinels. The dead error path is gone.
  • New tests:
    • The file-number-seeded PSF star split is the same under any global RNG state, and its two sets are disjoint, exhaustive and ceil(20%).
    • The tile VIGNET and multi-epoch stamp sizes match across the data and sims configs.

Each new test fails under an injected bug. Unit and module tests in the dev container: 805 passed, 1 skipped.

Claude Opus 5.5 on behalf of Cail

🤖 Generated with Claude Code

cailmdaley and others added 4 commits September 30, 2026 05:40
make_post_process parsed every HISTORY card, so a repeated card wrote a
duplicate EPOCH HDU and counted the exposure twice in N_EPOCH;
find_exposures already deduplicates the same cards.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ngmix lays the tile VIGNET's neighbour markers over each multi-epoch stamp
pixel for pixel, so config_tile_Uc.ini's VIGNET_SIZE, default_noimaflags.param's
VIGNET and every PiViVi config's multi-epoch STAMP_SIZE must agree.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@cailmdaley
cailmdaley merged commit 6b0d9ef into develop Oct 1, 2026
3 checks 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.

1 participant