You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@yetus.apache.org by Sean Busbey <bu...@apache.org> on 2016/08/17 19:18:55 UTC
github plugin might be misbehaving?
Before I file a JIRA wanted to see if anyone else agreed.
take a look at YETUS-309:
https://issues.apache.org/jira/browse/YETUS-309
Its two different patches failed in precommit. I reran the last in debug
mode:
https://builds.apache.org/view/PreCommit%20Builds/job/PreCommit-YETUS-Build/471/console
The patch is clearly from format-patch, but the yetus output shows:
[Wed Aug 17 19:14:08 UTC 2016 DEBUG]: locate patch
Processing: YETUS-309
[Wed Aug 17 19:14:08 UTC 2016 DEBUG]: github: YETUS-309 is not a pull request #
[Wed Aug 17 19:14:08 UTC 2016 DEBUG]: jira_locate_patch: trying
https://issues.apache.org/jira/browse/YETUS-309
[Wed Aug 17 19:14:08 UTC 2016 DEBUG]: jira_http_fetch:
https://issues.apache.org/jira/browse/YETUS-309
YETUS-309 appears to be a Github PR. Switching Modes.
[Wed Aug 17 19:14:10 UTC 2016 DEBUG]: github: test-patch is not a pull request #
[Wed Aug 17 19:14:10 UTC 2016 DEBUG]: generic_locate_patch: failed to
download the patch.
ERROR: Unsure how to process YETUS-309.
The only github related thing I see on the JIRA is a link in a comment
from Dima pointing to one of our source lines. My guess is that we're
claiming github-pr based on just that link.
plausible?
Re: github plugin might be misbehaving?
Posted by Allen Wittenauer <aw...@effectivemachines.com>.
> On Aug 17, 2016, at 12:18 PM, Sean Busbey <bu...@apache.org> wrote:
>
> The patch is clearly from format-patch, but the yetus output shows:
That doesn't matter because ...
>
> YETUS-309 appears to be a Github PR. Switching Modes.
... as soon as Yetus picks up something that it thinks is a github PR, it switches to github mode and there is no mechanism for it to go back.
> [Wed Aug 17 19:14:10 UTC 2016 DEBUG]: github: test-patch is not a pull request #
> [Wed Aug 17 19:14:10 UTC 2016 DEBUG]: generic_locate_patch: failed to
> download the patch.
> ERROR: Unsure how to process YETUS-309.
>
>
> The only github related thing I see on the JIRA is a link in a comment
> from Dima pointing to one of our source lines. My guess is that we're
> claiming github-pr based on just that link.
>
>
> plausible?
Completely. But it's really two separate bugs:
* the JIRA plugin's github bridge that tries to detect if it is a JIRA w/a link to github probably needs to get it's regex strengthened a bit.
* if the github plugin detects it isn't really a github PR it should really have a way to send it back to JIRA if the bridge collapses.
Relatedly, it's been requested that the bridge should detect which one is "newer" and use that, but that's *really* hard to do efficiently without making a ton of queries all over the place.