Build: phase out git clone option --no-single-branch (one of several approaches)
#10422
Conversation
| # TODO: Remove the 'not' | ||
| return not self.project.has_feature(Feature.GIT_CLONE_SINGLE_BRANCH) |
There was a problem hiding this comment.
noting the TODO here :)
|
Would be nice with a review of the general approach taken before updating tests. CC: @ericholscher @humitos - there's a wall of text here - #9736 (comment) - the executive point is in the "proposal" but I think the context is needed and I understood what to do by writing it down and reading it myself. |
|
Hmm... the tests passed without updating them. I'd want to add at least something here. |
There was a problem hiding this comment.
I haven't read this PR or the associated issues in depth yet, but I want to write down a comment here before we go too deep into this.
It's important to have the same behavior (git clone command) on PRs than on tags/branches. I want to avoid building successfully on PRs but failing on regular versions, or similar.
Also, to debug builds will be a lot harder if we have different behaviors here.
| # External builds aren't relevant, there is always an individual fetch operation. | ||
| # See self.update() | ||
| elif not self.version_type == EXTERNAL: | ||
| cmd.extend(["--no-single-branch"]) |
There was a problem hiding this comment.
I'd like to always have the same behavior for branch, tags and external versions. Otherwise, things are going to be pretty complicated to debug. Also, there will be things that work on PRs but fail on regular versions --which is the main feature of PRs builds.
There was a problem hiding this comment.
Let's go over this tomorrow then 👍 If you can set aside some time to read the initial analysis in #9736 (comment), that should give the pretext of why this was necessary.
benjaoming
changed the title
--no-single-branch--no-single-branch (one of several approaches)
|
Closing work on this approach in favor of #10430 |
Fixes: #9736