Open jayantbh opened 1 month ago
This makes a lot of sense for improving the accuracy of DORA metrics in Middleware. Since CFR is based solely on reverted PRs, it misses out on a significant number of failures or issues that might not result in an actual PR revert but still require resolution.
Why do we need this ?
Currently CFR is based on reverted PRs. While it's a bit of a challenge to associate it with much else based purely on PRs, perhaps some filtering/calculating can be done based on PR titles, branches, labels, etc.
For example, maybe I could include in any PRs that start like "Fix #1234 ABCD" to mean that it's a fix for PR 1234, so the "failure" could be associated with that PR, and the resultant resolution time could be calculated as the time between PR 1234 being merged, and "Fix #1234 ABCD" being merged.
Thoughts?