pirl-unc / mhcgnomes / 33469415726
90%

DEFAULT BRANCH: main
Ran
Jobs 1
Files 41
Run time 1min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
coverage: 89.273%. Remained the same
33469415726

push

github

web-flow 
Make serotype regeneration idempotent, and fix the documented release flow (#157)

* Make serotype regeneration idempotent, and fix the documented release flow

I filed #156 with the hazard backwards. I claimed regenerating
serotypes_generated.yaml drops 15 rows, having compared
build_serotype_mappings output -- the pure dictionary transform, 135
rows -- against the file without ever running the script. generate_yaml
explicitly carries forward serotypes the dictionary has no row for, and
running it produces 170, not 135.

The real hazard is narrower and points the other way: regenerating
re-adds A*2418 to A3, undoing a curation fix. The dictionary gives
A*24:18 the dual type "A24(9)/A3" and comments "short A24 with most A3
and A9 reactive; NN: A24", so A24 is the assignment; curation had
removed it from A3 and a regeneration would put it back unnoticed.

A CURATED_EXCLUSIONS set in the generator carries the pair and the
reason, and regenerating now reproduces the runtime table exactly. Four
tests pin it, including one asserting every exclusion is a pair the
dictionary actually asserts, so a stale entry cannot outlive its reason.

This also settles part of #153: Cw15 and Cw17 are in the carried-forward
set rather than derived from the dictionary, so they are not evidence
about what it says, and a test now names them.

#152: AGENTS.md documented `deploy.sh <version>`, which deploy.py has no
positional argument for -- and since deploy.sh runs lint and tests
first, following it burns the full suite before failing on argparse.
Corrected to the flow that exists. Implementing the argument instead
would make deploy.sh mutate the working tree, which its own
ensure_clean_tree check exists to prevent.

0 of 11,558 corpus names change; the runtime serotype table is
unchanged.

Closes #156
Closes #152

Claude-Session: https://claude.ai/code/session_012s4siLj2Vm33eNMGjNSazH

* Track the HLA dictionary so regeneration is reproducible

The tests pinning regeneration ... (continued)

4186 of 4689 relevant lines covered (89.27%)

0.89 hits per line

Jobs
ID Job ID Ran Files Coverage
1 33469415726.1 41
89.27
 GitHub Action Run
Source Files on build 33469415726