On Dec 20, 2013 7:49 AM, "Allan McRae" <[email protected]> wrote: > > On 20/12/13 07:58, Dave Reisner wrote: > > On Thu, Dec 19, 2013 at 04:24:00PM -0500, Jeremy Heiner wrote: > >> On Thu, Dec 19, 2013 at 3:51 PM, Dave Reisner <[email protected]> wrote: > >>> How did you encounter this? 'make distcheck' passes for me on master > >>> currently, and this does a VPATH build which includes 'make check'. > >> > >> Here are the steps I just used to recreate the problem: > >> > >> cd <top_srcdir> > >> git checkout master > >> git clean -Xdi > >> git status --ignored > >> ./autogen.sh > >> mkdir ../clean > >> cd ../clean > >> ../<git>/configure > >> make > >> make check > >> > >> All the tests except sync200 run, then make fails with: > >> > >> make[4]: Nothing to be done for 'test/pacman/tests/sync200.log'. > >> fatal: making test-suite.log: failed to create test/pacman/tests/sync200.trs > >> fatal: making test-suite.log: failed to create test/pacman/tests/sync200.log > >> Makefile:1035: recipe for target 'test-suite.log' failed > > > > Thanks, I can reproduce the failure with this recipe. > > > >> I'm guessing your working dir has a copy of sync200.py which isn't > >> getting cleaned before distcheck does its thing? > >> Jeremy > >> > > > > Something like that. > > > > Personally, I'd rather we get rid of the wildcard matching and just add > > an explicit list of tests along with a build rule for %.py.in to %.py. > > I'd suspect that this won't be the last time being cavalier about what > > the buildsys includes in the dist tarball and what tests are run bites > > us in the ass. > > > > OK - that sounds a proper fix. Someone do that!
I can send a patch in a bit > > Allan > >
