Contributors mailing list archives

Browse archives


Re: 14.0 branches

Acsone SA/NV, Stéphane Bidoul
- 09/10/2020 12:30:14
Hi Pedro,

I'm just trying to simplify and resolve the issue we have with the current approach.
Maybe this will create other issues we don't know yet, we'll see.
It's trivial to come back to the previous behaviour anyway if needed.

Regarding the case you mention, it should not be a problem because the wheel is built immediately to the OCA wheelhouse by /ocabot merge (it's only the publishing to PyPI that occurs nightly but travis gives priority to the wheelhouse).

@Simone, I think it's too early to say the other way is deprecated. Let's give it some time.


On Fri, Oct 9, 2020 at 12:21 PM Pedro M. Baeza (Tecnativa) <> wrote:
I'm not sure about this change massively without asking. I know you always want PIP everywhere, but this can have a lot of unpredicted side effects. The first I'm thinking of, if I don't know bad the mechanism, it's the following:

- 2 PRs with module A and a module B depending on A.
- We merge module A.
- Rebasing nor restarting module B Travis will work until the wheel is built that night.

Is that true or am I wrong? I know this can be workarounded through test-requirements.txt, but it means 2 changes, while in the other mode, it was not needed.


Post to: