CI: the README tables check is a job of its own ("README tables", main only)

After a list merges, main's tables and counts lag until they are regenerated. As a failing unit test that
turned "Lists, tools and docs" red, which is also the signal the merge bot waits on between merges, so a
wave would stop at its first list. The check now runs as its own job, which the bot can leave out while
its counts step regenerates the tables after the wave. The strict unit tests that compared the committed
tables and the alt-text count with the folders go: the job's gen_tables.py --check covers both.
This commit is contained in:
Gathin23 committed 2026-10-24 16:09:38 +05:30
1 parent e4979b0ea6
commit 2ee1d306d9
4 files changed
+19 -19

No files matched your search

+2 -1
View File
@@ -65,7 +65,8 @@ it, and every one of them gets a home. Pick the smallest home that fits (CONTRIB
examples/README, the brand line and the counts are generated from those files: a list PR never needs README rows,
and rows an older PR added are dropped on merge (`tools/gen_tables.py --adopt` turns them into its `tv.json`).
- After merging lists, the maintainer runs `python3 tools/gen_tables.py` on main and commits what it rewrites
(with `--font Fredoka.ttf` it also redraws the card when a count changed); CI on main fails until then.
(with `--font Fredoka.ttf` it also redraws the card when a count changed); CI's "README tables" job on main
stays red until then, and only that job.
- `pm hide`/`pm block` needs the matching unhide/unblock undo in the header.
## Scope and taste