But now this also did not work. It is just terribly annoying, I can hardly change a line of code in my Tycho-Dev workspace. I just attached a debugger and my first impression is that the DependencyResolutionContext passed to the ProjectRegistryManager. But I will investigate further. ProjectRegistryManager line a new facade is created but I can't tell from the code if this is intentional. One think that might help here: If I understand the code right, it triggers some kind of "project changed", maybe it would be better to just note all projects that require such a change event in a list, complete the cycle and then trigger one big "project s change" event afterwards?
Now that I used my Tycho-Dev workspace from an Eclipse instance that is launched in debug mode from another Eclipse having the m2e projects in its workspace the endless loop is gone. Even when I use the workspace from the standalone Eclipse again. When I just hat the debugger attached I observed that constantly pom-files are added to the DependencyResolutionContext context in the very end line of ProjectRegistryManager.
Since the calling method runs in a loop while context is not empty this would explain the endlessness. From the relevant code snippet above I concluded that the pom files are only added to the context if hasDiff oldRequirements, requirements is true.
When I looked into the elements of both sets I noticed that they contain the same artifacts but with different versions. Some were Tycho 2. This also coincides with my observation that the endless loop first occurred for me when I fetched and rebased the commits for the Tycho 2.
So the modules in my workspace went from 2. Maybe m2e does not update the project-data correctly and holds some old version strings. But this is the first time I look into the sources of m2e so this might be the wrong path.
I don't think I'll be available to work on this issue soon no more school for my kid here, so reduced productivity and another kid to get born sometimes this month making me off from work for one month ; but it's definitely something I'm interested in seeing fixed. What I have missed the most so far are steps to reproduce from a simpler project. If someone can identify some steps, that would be help in making more concrete progress. The current depedency resolution in m2e is not immune to build cycles as far as I know.
As a consequence, if Tycho has some build cycle it can make things fail. I don't think Tycho has a build cycle but Tycho x. The problem is that the error comes and goes and when its gone there is normally no demand to broke it again But I can of course understand that this is not a top-issue as long as it is not reproducible.
What would be of major interest for me would be: How good is this area covered by unit test? So if I start hacking on his, how big is the chance to break something unintentionally and uncovered? A reproducible case would indeed be a goo thing. I deleted and re-provisioned my tycho-workspace a few days ago because of this problem and run again into the endless loop I think it was after going back and forth in the git history between and after the 2.
So I can try to find steps to reliably reproduce this issue. Regarding cycles in the build path: Sometimes I get problem markers indicating cycles in the build path. The last time the vanished by cleaning and building the entire workspace so maybe this was based on outdated metadata that cause the same problem here.
Below there is an extract from the log, while for some times there are some items different after a while the following entries are repeated over and over again:. I then cancel the job and if I update the 'org. So this seems to happen if.
I now get an error marker: Description Resource Path Location Type The project cannot be built until its prerequisite tycho-core is built. Cleaning and building all projects is recommended sisu-equinox-launching Unknown Java Problem. This also happens for other projects, interestingly from time to time the maven builder kicks in and begin to build something without any action that would trigger this, e. Now the error marker on org. SO it seems there is something stuck or timmed-waiting in m2e.
We recently experienced such a loop as well. The last method hasConfigurationChanged compares the previous configuration with the current configuration, which include the concrete timestamps for the manifest. The timestamp's precision is 1s.
So whenver the file gets generated which triggers a change event leading to the above chain and the processing takes more than one second or occurs around the switch to the next second , hasConfigurationChanged returns true and triggers another build. Jira Service Desk. Interests Feedback Forum. Atlassian Cloud Migrations. Team Playbook. Atlassian logo. Training and Certification icon. Top groups groups-icon. It's not the same without you Join the community to find out what other Atlassian users are discussing, debating and creating.
Sign up for free Log in. How to debug stuck maven dependency download in Bitbucket Pipelines. Piercarlo Slavazza Mar 29, The dependency is downloaded via scp. When Maven tries to download it via wagon ssh , it remains stuck. Any hint about debugging it? Answer Watch. Like Be the first to like this. Answer accepted. And indeed upgrading the wagon extension to 3.
I cant remenber to be honest its been almost 2 years but i think so — Paulo Botelho. Add a comment. Active Oldest Votes. I remembered that I was using a VPN connection. Turning off VPN solved my issue. Wlad Wlad 2, 2 2 gold badges 20 20 silver badges 35 35 bronze badges. Sign up or log in Sign up using Google. Sign up using Facebook. Sign up using Email and Password. Post as a guest Name. Email Required, but never shown. Learn more about Collectives on Stack Overflow.
The Overflow Blog.
0コメント