Fixed ambiguous ls-remote with refs/heads/...#135
Fixed ambiguous ls-remote with refs/heads/...#135alexanderlink wants to merge 3 commits intojenkinsci:masterfrom
Conversation
This is the smallest possible change to solve the issue of CI builds being triggered again and again. ls-remote will always be definite when "refs/heads/<branchName>" is used as branchSpec. All other issues tried to solve with Pull Request jenkinsci#129 are still open. But using refs/heads/ would already solve the CI issue.
|
plugins » git-client-plugin #470 ABORTED |
|
Thank you for a pull request! Please check this document for how the Jenkins project handles pull requests |
|
plugins » git-client-plugin #471 SUCCESS |
|
plugins » git-client-plugin #472 SUCCESS |
...and refs/tags/ (to be discussed - but does not harm).
|
plugins » git-client-plugin #473 SUCCESS |
|
Hi Mark, I added some integration tests to the git plugin verifying the changes in this (git-client plugin) pull request. Kind regards, |
|
Any updates on this? I'm waiting for this to be fixed really badly... |
|
@MarkEWaite any update? We're also waiting for this fix. Sooner or later we will get trouble (I already thought about releasing a patch internally). |
|
I just released 1.9.2 and set the next version to be 1.10.0, preparing for this change and one other change. I spent several hours experimenting with the change last weekend but did not complete enough testing to be confident to release it. I hope to release 1.10.0 with this fix and 1-2 others within the next 1-2 weeks. |
|
This change has been merged into git-client-plugin prior to the release of 1.10.0. |
Branch Spec Fix: jenkinsci/git-client-plugin#135 This commit relates to: jenkinsci#232
Branch Spec Fix: jenkinsci/git-client-plugin#135 This commit relates to: jenkinsci#232
Pull Request 135 was merged. I tested and provide feedback in this commit jenkinsci/git-client-plugin#135 jenkinsci#232
Pull Request 135 was merged. I tested and provide feedback in this commit jenkinsci/git-client-plugin#135 jenkinsci#232
|
I added feedback regarding #135 and open issues: I closed #129 because it became too complex, but maybe we should continue to try solving the remaining issues. |
|
I'm not clear that we can do much more without breaking compatibility with existing behavior. The existing behavior is flawed (ignoring everything to the left of the rightmost "/" character in the "Branches to build" value unless refs/heads is used), but I think there are too many use cases which will break if we improve the behavior for those old cases. I think it would be very nice to include help text in the "Branches to build" field of the git plugin which describes the use of refs/heads as a way to unambiguously specify a branch to build. |
...to latest changes in git-client plugin. jenkinsci/git-client-plugin#135
|
Hi Mark, I adjusted the help: jenkinsci/git-plugin#244 Thanks and regards, |
...to latest changes in git-client plugin. jenkinsci/git-client-plugin#135
This is the smallest possible change to solve the
issue of CI builds being triggered again and again.
ls-remote will always be definite when "refs/heads/"
is used as branchSpec (except if branch names start with "refs/heads/" which no one should do - but could).
I wrote down my findings and thoughts here:
https://github.com/alexanderlink/git-plugin/tree/branchSpecDiscussion
All other issues tried to solve with Pull Request
#129
are still open. But using refs/heads/ would already solve
the CI issue.