From an outside review of the web UI. The finding is the complaint; the suggestion below is one way out, not a decision.
A run row on the builds page leads with a ten-character sha
(internal/web/templates/builds.html:44):
32e337bd1c main · 2026-09-20 06:00 UTC success instances ×2
Nothing on the page says what the commit is. Recognising a build means
opening the commit link or already knowing the sha. #240 split runs that
were queued at different times into separate rows, which fixes the collapse
but not the naming: two rows for the same sha still read identically apart
from the timestamp.
Reported:
- Hashes lead the list without commit subjects.
- Repeated attempts of the same job within a group are not distinguishable.
One option: lead with the commit subject and the branch, demote the sha to
secondary metadata, and mark repeated attempts of a job (an attempt number,
or the queue time per chip). The subject is not on BuildOut today, so this
needs the run's commit resolved for the page.
closed by cmc in commit 92c816c58e: builds: lead a run with its commit subject, and page the list
2026-09-21 21:18 UTC