Contributors mailing list archives


Re: Proposal for new workflow, incorporating "Optimistic Merging"

by Jairo Llopis <> - 08/06/2016 08:48:00
Also I'd like to highlight another reason why I'm against this optimistic merging: OCA is no company, it's a community.

This means we have no plans of what should OCA modules do. They just do what the contributor needs them to do, and what reviewers ask them that it should do, and keep growing by themselves covering more and more needs each time.

Odoo SA, for instance, have company plans, have a product vision, and then they have some strict guidelines about what should go to what version, and they have code freezes and "stable" (haha..) releases. Many times I send them a patch they consider unstable and they ask me to push it to master. But they are a company, not a community.

So OCA's goals, as I see them, are:
1. Share effort.
2. Provide tools to have...
3. ... solid code.

So if you just want OCA to merge something expecting somebody will look at it later and fix your own failures:
1. You're lazy.
2. You are truly optimistic. That's not going to happen...
3. ... unless OCA pays somebody to fix others' garbage code, which returns to point 2.

For us, the "betas" here are the PRs themselves. We often merge a PR in production and undo that when it gets in core OCA. With git it's quite simple, and you have this fixed version of what you need with stability everywhere else.

Right now, OCA has 1197 Modules. That's a burden to maintain. It's no problem because effort is shared, and mostly people maintain what they need, so if something is not needed anymore, it will probably not get to recent versions. But now sum 1197 modules + garbage code to clean. Too much for OCA.

And besides, one more thing: if you want to help new contributors to fix their garbage instead of teaching them to fix it by themselves (which I think should be the road to follow), then you can do it now by opening a PR against their branch, which I have done several times. Easy for you, easy for them.

So, in short, I see optimistic merging as a bad idea because:
1. Adds zero benefits above current workflow.
2. Adds maintenance burden.

Jairo Llopis