Open xmfcx opened 1 week ago
However, my question is: why are we generating architecture-specific tags initially and then merging them later? Wouldn't it be cleaner and more efficient to push a single tag directly, without separating them at the start? This could help in keeping the package repository more organized.
I don't know how to do it properly because the CI runners are different from machines and architectures so that I think the manifest could be also different. I hope @oguzkaganozt knows more than me.
This is not my emergent issue. I'm waiting for your contribution not only by discussion but by PR. Thanks.
I will investigate try to find out more straight-forward solution to this.
Checklist
Description
Right now we have different tags for
arm64
andamd64
images.This is redundant since docker supports multiple cpu arch's in same tag.
Example:
Purpose
Simplify the docker image tags by making use of modern docker features.
Possible approaches
https://www.docker.com/blog/multi-arch-build-and-images-the-simple-way/
Definition of done
Image tags don't have
arm64
oramd64
keywords in them.