tycho
tycho copied to clipboard
Delete corrupted jars and download them again
I don't get which issue is fixed by reading the commit title. Can you please enrich the message with an explanation of why this change is necessary?
If you look at these PR: https://github.com/eclipse-tycho/tycho/pull/1476 there are two "failing" test because a jar can't be read (I assume it is corrupted and now cached somewhere), so the goal is in such a case delete the file and trigger download again. I already fixed something similar somewhere else and seems here is another such case where it can happen.
Please add such information in the commit message.
Test Results
332 files 332 suites 2h 20m 6s :stopwatch: 295 tests 289 :heavy_check_mark: 4 :zzz: 0 :x: 2 :fire: 590 runs 579 :heavy_check_mark: 9 :zzz: 0 :x: 2 :fire:
For more details on these errors, see this check.
Results for commit 89d558cf.
:recycle: This comment has been updated with latest results.
Thanks
Well lets see if this actually helps, first attempt failed, so I added more debug output for now :-D
Okay debug output is:
[WARNING] First attempt failed, deleting file C:\Users\runneradmin\.m2\repository\p2\osgi\bundle\org.eclipse.equinox.launcher.gtk.linux.x86\1.1.551.v20171108-1834\org.eclipse.equinox.launcher.gtk.linux.x86-1.1.551.v20171108-1834.jar :: org.eclipse.tycho.core.osgitools.OsgiManifestParserException: Exception parsing OSGi MANIFEST C:\Users\runneradmin\.m2\repository\p2\osgi\bundle\org.eclipse.equinox.launcher.gtk.linux.x86\1.1.551.v20171108-1834\org.eclipse.equinox.launcher.gtk.linux.x86-1.1.551.v20171108-1834.jar: zip file is empty
[WARNING] FIle is not null but do not exits! C:\Users\runneradmin\.m2\repository\p2\osgi\bundle\org.eclipse.equinox.launcher.gtk.linux.x86\1.1.551.v20171108-1834\org.eclipse.equinox.launcher.gtk.linux.x86-1.1.551.v20171108-1834.jar
So it seems for some reason Tycho thinks that the file is already downloaded and just returns a file pointer to a non existing file ...
It seems somehow related to the windows execution only, all attempts to verify an re-download seem to result in the same problem (beside that I was able to fail the checksum validation on Jenkins...)