Conversation

Notices

1. Embed this notice
   Jan Lehnardt :couchdb: (janl@narrativ.es)'s status on Friday, 13-Dec-2024 02:30:11 JST Jan Lehnardt :couchdb:

   in reply to

   … but it will definitely make integrating all PRs take forever because they touch so much structure.

   The alternative is everyone committing to main/trunk and forbidding force pushes. The trade-off is that everybody has to deal with everybody else’s changes, but it will be a net benefit because those integration decisions happen closer to the changes and and committers likely working time overlapping for out of band coordination.

   • Embed this notice
     Jan Lehnardt :couchdb: (janl@narrativ.es)'s status on Friday, 13-Dec-2024 02:30:14 JST Jan Lehnardt :couchdb:

     Had a git workflow epiphany this morning because it went (mildly) badly for the second or third time and I’m finally seeing the pattern:

     Very early on in a project, when folder structure, naming conventions and little test-spike-code are all being tried out, a PR based workflow is worse because the nature of the changes will likely make the PRs long running, because lots has to be nailed down before it is “good”…