-
Notifications
You must be signed in to change notification settings - Fork 2.7k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Make kotlin load parent first (revert PR which changes that) #29869
Make kotlin load parent first (revert PR which changes that) #29869
Conversation
Thanks for your pull request! The title of your pull request does not follow our editorial rules. Could you have a look?
|
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
I've checked the mac failure, and it's a machine failure (although I have no explanation for why it's happening, since there is enough disk space). |
The windows test failure is a bit more worrying, since it's in the kotlin tests. I'll trigger a new build to see if it's something intermittent, since these changes should be relatively innocuous. |
11979e9
to
f5ed20e
Compare
I did re-run it yesterday, so it failed twice already. |
Oh, thanks for re-running, @aloubyansky. Bah. I guess the two options are:
I've just pasted in from https://raw.githubusercontent.com/stuartwdouglas/quarkus/422a492c32ab1046f7a872ff4434ebac2923e06e/extensions/kotlin/runtime/pom.xml (the revision before Stuart's changes), and it's identical to what I made manually, so I don't think it's that. I'll try reverting the classloader changes and pushing to this PR to see what happens on Windows, since I can't try it locally. We like those changes, but maybe they expose some other bug. |
This comment has been minimized.
This comment has been minimized.
... and the build that had test failures the first time ... died before finishing. I've looked through the logs and I can't see any evidence it got to |
This comment has been minimized.
This comment has been minimized.
So from the test archives the remote side looks like:
And from the test logs:
Basically it is sending the first request, and then the remote side shuts down. |
I think this may caused by #29785 that automatically propagated parent first dependencies. It looks like somehow the remote and local side end up with different ideas about what is parent first. |
I wonder why it only affects Windows? The options now are:
Either way, we have confirmation that the |
9337dbf
to
94aba09
Compare
Yes, it shouldn't be specific to Windows then. I don't have Windows installed but I'll have a look. |
@holly-cummins can you try making |
Done, @stuartwdouglas. We'll find out in five hours if it helped :/ ... well, ten hours if the windows build gets cancelled before completing and we need to respin. (We should probably increase the timeout window on those.) |
This comment has been minimized.
This comment has been minimized.
Double-good news:
Mac M1 is playing up again with a podman initialization failure (sigh), but I think we can ignore that. So it looks like your suggestion did the trick, @stuartwdouglas. |
Did |
Ah, @aloubyansky, the build-speed-up I noticed was just because we only need to test the kotlin projects, rather than testing everything, as we do if we change a part of core quarkus like the classloader. |
I've just spotted this hasn't been merged. @aloubyansky or @stuartwdouglas, should one of you do the honours? |
78f9806
to
1e4a3e5
Compare
(I'll squash and rebase.) |
1e4a3e5
to
ea5625d
Compare
I think the changes in #29697 may save a lot of headaches, but they also have the potential to break users of the kotlin extension. We should hold off on making them to a major version boundary.
Discussion here (in the earlier messages): https://quarkusio.zulipchat.com/#narrow/stream/187038-dev/topic/Review.20of.20.22make.20this.20Kotlin-y.20library.20work.20in.20dev.20mode.22/near/315122157
cc @stuartwdouglas @aloubyansky