https://bugzilla.wikimedia.org/show_bug.cgi?id=73177
--- Comment #5 from Bryan Davis <bda...@wikimedia.org> --- (In reply to Bryan Davis from comment #4) > (In reply to Tim Landscheidt from comment #3) > > Also, the current patch set's .travis.hhvm.ini has > > "hhvm.enable_obj_destruct_call = false", while above you wrote that it is > > set to true in MediaWiki-Vagrant and at the WMF cluster. > > I intended to set it to true but apparently got some additional testing > settings included. This points more strongly to disabling JIT as the thing > that fixes the test. I'll run a few more iterations on Travis to see if I > can narrow down what actually fixes the failure there. It turns out that disabling JIT is all that is needed to make the test past (<https://travis-ci.org/bd808/mediawiki/jobs/40476541>). I'm hoping this is just an artifact of something in the cli process and not a systemic problem in a running wiki. HHMV without JIT is not highly compelling speed improvement for the investment we have made in it. Does anyone have an example of a page/feature in a live MediaWiki instance that would be highly sensitive to this sort of scoped destructor behavior? I think we need to figure out if the problem is isolated to the unit test or more systemic. -- You are receiving this mail because: You are on the CC list for the bug. _______________________________________________ Wikibugs-l mailing list Wikibugs-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikibugs-l
