On 08/02/2017 07:06 PM, Dominik Ruf wrote:
Hi all,

while working on https://bitbucket.org/conservancy/kallithea/pull-requests/357 I noticed, that even though the PR fixes the broken celery functionality on linux, it still does not work on windows. I'm not even sure it can be fixed on windows.
And Celery 4 officially dropped windows support
http://docs.celeryproject.org/en/latest/whatsnew-4.0.html#removed-features

So this raises the question, do we want to
a) drop windows support as well
b) switch to something different (maybe twisted)
c) try to fix it on windows and stay on celery 3 (for ever!?)

What are your thoughts?

Two very different things:

I am not aware of any changes that could justify that it no longer should work on Windows. It is our code talking directly to the queue through Celery - I would not expect TurboGears to play any role in that.

The most likely culprit is the upgrade from Celery 2 to 3.

So ... if it is broken on the default branch, I guess that "just" is a bug. Perhaps some oversight in the upgrade, or an issue in Celery 3 (there are probably reasons they gave up supporting it).


Then, long term:

Celery might be needed for big setups, but it might not be a good idea to run those on Windows anyway. And those who do are probably companies who can afford to get in touch with Celery and get support or sponsor or contribute development. We will have to move on from Celery 3 at some point. It would be sad to lose Windows support in an upgrade to Celery 4, but that might be the best option.

But also: we only use a small part of Celery. It is probably feasible to keep that part working.

/Mads

_______________________________________________
kallithea-general mailing list
[email protected]
https://lists.sfconservancy.org/mailman/listinfo/kallithea-general

Reply via email to