As I've said before I believe not pre-installing celery and k8s executors should be the ultimate goal, so I totally agree with this, but we need to do it in a way that minimizes impact. It's hard to catch every angle possible with these sorts of things (i.e. Daniel's point of folks installing required deps outside of extras, I could see this being done pretty commonly).
What do folks think about doing this in stages? I don't see a reason why we need to make all the changes at once. We could release the executors in providers and have them come pre-installled for some number of releases. This would allow the majority of folks to slowly migrate to newer versions of both core and providers which would then minimize the blast radius when we eventually turn off the pre-installation of those providers. ________________________________ From: Daniel Standish <[email protected]> Sent: Friday, July 21, 2023 10:07:44 AM To: [email protected] Subject: RE: [EXTERNAL][DISCUSS] Should we pre-install celery/k8s providers? CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you can confirm the sender and know the content is safe. * *provider* extras or provider optional features
