Dug in a little bit. It seems like svn can return changed_paths in different order between calls. I've added sorting to make order deterministic. I hope that will fix the issue (can't reproduce locally). It's in branch svn_paged_diff_fix, please take a look.
On Mon, Apr 6, 2015 at 9:00 PM, Dave Brondsema <[email protected]> wrote: > It seems to give random results, it failed with different data different > times > and it passed twice. > > I set up our Jenkins builds quite a while ago to print out lots of > environment > info, so we can know what it runs with. You can see that near the top of > the > output e.g. at https://builds.apache.org/job/Allura/799/console > > It looks like it failed with 1.6.17 and then passed once with that > version. And > then the most recent pass had version 1.6.23 (perhaps the different jenkins > slaves have different versions) > > Thanks > > On 4/6/15 5:44 AM, Igor Bondarenko wrote: > > I think it's svn version or platform specific too. Is it failing every > time > > on jenkins? Which svn version used there? I'll try to reproduce it and > dig > > in. > > > > On Tue, Mar 31, 2015 at 12:24 AM, Dave Brondsema <[email protected]> > wrote: > > > >> I tried fixing this by adding a sort: > >> > >> > https://forge-allura.apache.org/p/allura/git/ci/758b51bc8c25a1951745ae36d0104d185f560170/ > >> > >> But it didn't work, in fact the test failed on more than sorting after > that > >> "fix", which is weird. Igor maybe you have a better idea? > >> > >> I'm thinking it is platform or SVN-version specific since it fails here > >> but not > >> on other environments. > >> > > > > > > -- > Dave Brondsema : [email protected] > http://www.brondsema.net : personal > http://www.splike.com : programming > <>< >
