Following discussions during the Vancouver OpenStack summit about the level of technical debt of patches in packages, the prototyped tools were created to explore these ideas.
Each package has a score which is the sum of the scores from each patch using the following rules:
- Not submitted upstream -> +200
- Abandoned upstream -> +150
- Not merged upstream -> +20
- Merged upstream -> +10
- Cherry-picked upstream -> +10
- Do not pass Jenkins -> +50
- Got a +2 -> -10
- Lowest vote is -2 -> +100
- Lowest vote is -1 -> +50
- Lowest vote is +1 -> -5
- Difference to upstream patchset is null -> 0
- Difference to upstream patchset is small -> +10
- Difference to upstream patchset is big -> +100
For not gerrit managed project (mainly outside of OpenStack), the tool tries to follow upstream submission using cherry-pick comments so it's less precise.
Actions that can be taken to lower the scores:
- be sure to have the Change-Id or cherry picked comment present.
- make the patchsets pass Jenkins.
- make the patchsets merged.
- use the current patchset when the patchset evolved in the review.
extract.sh does the following in sequence:
- extract files from the src.rpm using rpm2cpio
- for each patch lookup the
Change-Id:or thecherry picked from commitstring. - if patch is cherry-picked or in review, extract the patch from the git of the project using the id.
- compute an interdiff between the patches from the package and from the git repo.
- use diffstat to store statistics about the interdiff.
- if the patch is in review, extract the json info from gerrit.
score.py computes a score from the review json and the interdiff
stats and generate HTML output on stderr.
You can use run.sh to download packages and then process them
through extract.sh and score.py.
We are not using PR from github but the review system from Software Factory at http://softwarefactory-project.io/. So feel free to submit reviews.