On 20 August 2013 00:02, Phil Steitz <phil.ste...@gmail.com> wrote: > On 8/19/13 3:37 PM, sebb wrote: >> On 19 August 2013 23:24, Phil Steitz <phil.ste...@gmail.com> wrote: >>> On 8/19/13 3:20 PM, Gilles wrote: >>>> On Mon, 19 Aug 2013 13:05:10 -0700, Phil Steitz wrote: >>>>> Anyone have a clue why this is happening? Some Jacoco failure. >>>>> Googling the symptom from the Continuum build results indicates that >>>>> this may be related to the version upgrade we did recently in >>>>> commons parent. But the build runs fine for me locally and it does >>>>> not look like other components are failing similarly. Any ideas? >>>> http://markmail.org/message/xhji2hetipa6ncku >>> I don't get it. Did we break something in the commons parent >>> upgrade? >> Probably the jacoco version changed and it now wants attention again. >> >>> Can we somehow just turn this thing off? >> Yes, but why not just attend to jacoco? >> >>> Why does "mvn site" work locally but somehow Continuum is failing? >> It fails locally if you run the full Continuum command which is: >> >> mvn deploy --batch-mode -Pjava-1.6 -Dgpg.skip -Prelease > > Good grief. I have no idea what all of that actually does.
deploy builds and tests and uploads it includes the verify phase which is where jacoco:check runs -Pjava-1.6 sets the JVM -Dgpg.skip skips creating gpg sigs -Prelease is our version of the apache-release profile that causes assemblies to be built etc. --batch-mode disallows keyboard input > I know > that "mvn clean site" and "mvn clean install" both work fine for > [math] and the former generates a correct-looking jacoco report. The deploy phase also runs verify which runs jacoco:check, which is a separate goal. The check configuration syntax was completely redesigned between 0.6.2 and 0.6.3 The change report says just this about it: "The coverage check has been reworked to allow checks on all counter values on all element types" That may be the truth, but it's certainly not the whole truth! > In > any case, from what I can glean from commons parent, it looks like > whatever got borked must have gotten borked there, so we will need a > parent release to actually "fix" this. Yes, at some point. I've already checked in the changes. Meanwhile reverting to the previous jacoco version works. > We have no jacoco config in [math] unless I am blind. There is no local config, but CP32 comments mention the jacoco profile and the marker file src/site/resources/profile.jacoco which is what causes Jacoco to be run. --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
