Future of PR merges #242
Replies: 3 comments 3 replies
-
Sounds reasonable to me. I don't quite understand the 3rd point though, what use cases do you imagine where we would have PRs which go directly to master, not going via dev? I'd imagine this only being some minor fixes, that can be done in one commit anyway? |
Beta Was this translation helpful? Give feedback.
-
I think this is a very good idea, especially because it will hopefully reinforce the practice of keeping branches focused on the feature they are implementing. |
Beta Was this translation helpful? Give feedback.
-
Thanks, @silvanheller for the suggestion. I personally thnk any step towards a formalised process is beneficial for this project. One minor note though, since there sometimes there are Bachelor's / Master's Theses involved, I suggest to explicitely continue the notation you used recently while merging:
|
Beta Was this translation helpful? Give feedback.
-
Currently, we merge PRs "as-is", with a full commit history. While this is not bad per se, it clutters the commit history with commits which can anyway not be reverted to (since ideally, a PR only has a single "feature").
I'd avocate for the following:
dev
are squashed into one commit (potentially also rebased), since features should probably anyway be developed on separate branches (except for branches which are already cleaned up)dev
tomaster
are not squashed - here, the commits actually represent real featuresmaster
except those fromdev
are also squashed.As an example, see https://github.com/polypheny/Polypheny-DB/commits/master
Thoughts? @sauterl @lucaro @Spiess
Beta Was this translation helpful? Give feedback.
All reactions