Handle failed experiments better
Open
- Dominant language
- Python
- Stars
- 1.2k
- Forks
- 302
- PR merge metrics
- No merged PRs in 30d
Description
Currently if every build in an experiment fails, the measurer gets started.
Because the measurer is so spammy, it's very hard to look through the logs and see what happened.
But really the measurer should quit if there is nothing at all to measure.
Contributor guide
Assessment
This issue has not been assessed yet.