currently tomee doesn't build locally Tests in error: SubjectServiceTest.start:56 » EJB Provider error. No provider definition found SubjectServiceTest.stop:67 NullPointer
Tests run: 3, Failures: 0, Errors: 2, Skipped: 0 Environment: mbp, OSX 10.9.4, java Orcl 1.7.0_51 LieGrue, strub On Thursday, 17 July 2014, 23:44, Romain Manni-Bucau <[email protected]> wrote: > > >Yeah, a lot of work done! > >Side node: the fact we need the javaagent is an openjpa issue of >subclassing (known but kind of regression for them without the javaagent or >build time enhancement :() >Le 17 juil. 2014 22:11, "Andy Gumbrecht" <[email protected]> a >écrit : > > >> To sum up what was wrong with the failed release on Sunday. >> >> There was a SNAPSHOT dependency added to the parent POM and missed on >> deploy - This was a breaking issue on the staging. >> >> The builds could not be run in parallel, as many tests were using the same >> conflicting ports. We have not had concurrent builders running on the same >> buildbot for some time, but we have just been 'lucky' that this was >> 'mostly' working. >> I think I have most, if not all, the tests that bind to ports changed so >> that the ports are random - There will surely be more issues, but >> concurrent builds have become significantly more stable. >> We have to be able to cope with this as we now run builds of TomEE 1.7.x >> on Java 6/7 and TomEE 2.0.x on Java 7/8/9EA - Most of these will fire at >> the same time on commits and merges. >> >> Another big issue that cost significant effort was that the javaagent had >> been commented out in the openejb-core.pom during testing on Java 7, and >> unfortunately forgotten. This broke the enhancement for several tests in a >> very subtle way so that it was not obvious what was going wrong, the issues >> were only apparent on Java 7 build. This was also done at time that the >> build was not green, so made tracking down additionally hard. Obviously >> this led us to believe that the Java 8 support we were about to announce >> was broken. >> >> As trunk had already had significant formatting and finals changes >> committed it was necessary to make the 1.7.x branch comparable for >> successful merges. This also led to several test failures that needed to be >> rectified. The upside is that many tests were tweaked and updated during >> this process. >> >> Java 7 builds on Windows platforms also revealed several 'slightly' flawed >> tests, again the upside is that some minor issues were addressed during >> analysis. >> >> We are now really close to getting all issues resolved - We will have >> green builds by the end of the day. >> >> Andy. >> >> >> On 17/07/2014 17:18, David Blevins wrote: >> >>> Wondering if we can get a complete list of issues holding up 1.7 on a >>> single thread. >>> >>> On the legal side, we still need: >>> >>> - Library diff >>> - Review/adjust LICENSE and NOTICE files >>> >>> The last one on the list frequently tanks releases. Doesn't affect >>> security releases, but with larger versions is just a bear. Let's make >>> sure we do that one before we put up another vote as voting drastically >>> decreases each time we reroll. >>> >>> >>> >>> >> -- >> Andy Gumbrecht >> >> http://www.tomitribe.com >> [email protected] >> https://twitter.com/AndyGeeDe >> >> TomEE treibt Tomitribe! | http://tomee.apache.org >> >> > >
