https://bugzilla.wikimedia.org/show_bug.cgi?id=72778
--- Comment #7 from Bryan Davis <[email protected]> --- (In reply to Tim Landscheidt from comment #6) > Interestingly, while this has silenced the Jenkins tests, the change has > introduced failures in the same spots with Travis CI's HHVM: The commit > before passed (cf. https://travis-ci.org/wikimedia/mediawiki/jobs/39621924), > this one (cf. https://travis-ci.org/wikimedia/mediawiki/jobs/39624363) and > the ones after fail now with: > > | [...] > > | 2) FormatJsonTest::testParseTryFixing with data set #3 ('[1],', false, > true, '[1]') > | Expected isGood() == true > | Failed asserting that false matches expected true. I wonder if the HHVM packages that the WMF is building have some Facebook default settings enabled that aren't enabled in the builds used by Travis CI? When I looked through the upstream source for HHVM it looked to me like the JSON loose parsing behavior should not have been enabled by default and instead would require a custom HHVM option (JSON_FB_LOOSE) to be passed to json_decode to enable the loose parse behavior. I'll look a bit deeper and see if I can figure this out. I think that if we are enabling this in our build it is probably by accident. If nothing else, I can probably craft a guard to add to the test that detects if the default behavior is loose or not. -- You are receiving this mail because: You are the assignee for the bug. You are on the CC list for the bug. _______________________________________________ Wikibugs-l mailing list [email protected] https://lists.wikimedia.org/mailman/listinfo/wikibugs-l
