Christian, We talked about what has happen in Surefire with Herve, Stephen and Robert. We want to revert to [1] where CI was ok and never commit directly to master. We will open a branch with every change one after another and Herve will participate in code review. Herve told me that he is open for review. We do not need to be in hurry nor to rush nor to struggle. We can talk in ML or in pull-request in the git branch and after we got consensus we can proceed more further.
If you agree, I would send the VOTE to revert to [1] and I will fix integration tests in a branch which failed with your logs - I see the reason why it failed. And then we will start normal code review and we will enjoy it together. WDYT? Can we do it like this? [1] 66bc4c0839ba11af7a8915930f76abf3cd58ee53 On Tue, Jan 10, 2017 at 10:50 AM, Tibor Digana <[email protected]> wrote: > Sorry I could not get back to ML earlier. I got home early morning and > could not work in ASF. > I spent hours around my Nisan. It was so cold that my car could not start > up and the electronics in Nisan was totally crazy and could not lock the > car. We moved father's car from pumping station because it suddenly freeze > as well. So yesterday everything what could be broken was broken totally. > I need a day to get everything to normal and then I will participate in > ASF very soon. > > On Sun, Jan 8, 2017 at 9:31 AM, Christian Schulte <[email protected]> > wrote: > >> Am 01/08/17 um 06:48 schrieb Tibor Digana: >> > Christian, >> > !!! Stop committing to the project immediatelly !!! >> >> Issue making that IT fail is in this commit >> <https://git-wip-us.apache.org/repos/asf?p=maven-surefire. >> git;a=commitdiff;h=b50ea3acf0701e4a66ad8f55930cb41dd2bfc999>. >> >> In class 'ThreadedStreamConsumer', you changed >> >> thread.join() >> >> to >> >> if ( thread.isAlive() ) >> { >> thread.join( SECONDS.toMillis( 10L ) ); >> thread.interrupt(); >> } >> >> That call to interrupt seems to be an issue. You do need to join on that >> thread. It's important it does call consume with all remaining items. >> Interrupting appears incorrect to me. I changed that locally to >> >> if ( thread.isAlive() ) >> { >> thread.join(); >> } >> >> and the IT started to succeed again. See the attached patch. With this >> patch applied, all ITs are passing here locally now. >> >> >> >> >
