-
Notifications
You must be signed in to change notification settings - Fork 189
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
Significant target-resolution runtime regression #670
Comments
@laeubi I cannot assign labels here, likely because I'm not a committer for Tycho. |
Furthermore |
The @akurtakov @mickaelistria given that we remove pack200 in tycho 3.0 I'd like to change this so we simply never use pack200 in tycho 2.7. anymore even if it would be supported by the running JVM... any concerns? |
Please ignore pack.gz files. They are already ignored if running on Java13+ IIRC |
Signed-off-by: Christoph Läubrich <laeubi@laeubi-soft.de>
I was able to reduce this to one download per build, but this is a really nasty problem... now that we are using the remote properties the following happens:
|
Signed-off-by: Christoph Läubrich <laeubi@laeubi-soft.de>
Signed-off-by: Christoph Läubrich <laeubi@laeubi-soft.de>
Signed-off-by: Christoph Läubrich <laeubi@laeubi-soft.de>
Signed-off-by: Christoph Läubrich <laeubi@laeubi-soft.de>
Signed-off-by: Christoph Läubrich <laeubi@laeubi-soft.de>
Currently there are too many test that rely on pack200 :-\ |
All tests are fixed in my pack200 patch against master. Maybe it's best to leave its removal for 3.0 |
Yes removal for 3.0 and as this issue must also be done in 2.7 I decided to keep everything as-is and still support pack200 at the level of 2.7, hopefully after the release we then can remove this stuff completely.... |
Signed-off-by: Christoph Läubrich <laeubi@laeubi-soft.de>
Signed-off-by: Christoph Läubrich <laeubi@laeubi-soft.de>
Signed-off-by: Christoph Läubrich <laeubi@laeubi-soft.de>
Signed-off-by: Christoph Läubrich <laeubi@laeubi-soft.de>
Sounds nasty. But one download per build (the state of PR #673 this noon) already reduced the resolution time to ~4min in my case compared to more than 30min before. But it is still much longer than the 30sec of Tycho 2.6.0. |
I hope I have catches all cases now wit the PR and everything goes fine. If yes, only one download should happen once to get in sync and after that everything should be cached again... |
Great job @laeubi! |
My local build of https://github.com/eclipse-platform/eclipse.platform.releng (mvn clean verify -Pbuild-individual-bundles) keeps redownloading stuff . SWT fragments are the first visible part noticed on every build. |
Ho do I change the tycho version to be used here? Can you try to reproduce the issue with #680 |
Signed-off-by: Christoph Läubrich <laeubi@laeubi-soft.de>
Just pass -Dtycho.version=3.0.0-SNAPSHOT |
Signed-off-by: Christoph Läubrich <laeubi@laeubi-soft.de>
Signed-off-by: Christoph Läubrich <laeubi@laeubi-soft.de>
Signed-off-by: Christoph Läubrich <laeubi@laeubi-soft.de>
Signed-off-by: Christoph Läubrich <laeubi@laeubi-soft.de>
As reported in #651 (comment) the initial resolution of the target platform is magnitudes slower with the current Tycho
2.7.0-SNAPSHOT
compared to2.6.0
, in my case ~30min compared to ~30sec before.I have the impression that each dependency of each plug-in being built is fully fetched from a mirror again, even if it was already fetched for another plug-in which has the same dependency in the same Maven build. Therefore the runtime becomes linear to the number of references to dependencies and not only linear to the cardinality of the set of dependencies (which would be relatable).
Furthermore the dependencies should not be re-downloaded on each build which happens often (or maybe always, I cannot say that clearly from the print-out).
Please find attached a relativly small reproducer that contains three identical plug-ins that have the same dependencies:
tycho-2-7.zip
The text was updated successfully, but these errors were encountered: