[00:01:44] ty .. i was initially tempted to disable that and use SimpleStack .. but once I looked at the code and the html5 spec, it wasn't too bad. it is dependent on a previous patch that does some renaming .. https://gerrit.wikimedia.org/r/#/c/399759/1 [00:03:52] TimStarling, ParserMigration has a fun "bug" / "feature" .. looks like it tidies each part with the mw-parser-output div whereas in default pageview, tidying is without that div .. causes https://www.mediawiki.org/w/index.php?title=Topic:U43efyqnynpfik9z&topic_showPostId=u4erl7wiklfu0ack#flow-post-u4erl7wiklfu0ack [00:03:55] might be nice to have a test case, if you're feeling generous [00:04:18] it would go in tests/local/tree-construction [00:04:29] TimStarling, I wanted to, but I didn't investigate the test set up .. ah, ok .. i'll take a look there. [00:05:07] you know there is a standard test suite from html5lib [00:05:36] reg the parsermigration bug, i didn't investigate the actual code .. i just verified it by feeding tidy @ http://infohound.net/tidy/ some html and verifying my theory .. i think it pans out. [00:05:45] we imported that, and in tests/local/ there are tests that are in the same format as the ones from html5lib, except they are local to remex instead of imported [00:05:56] ok [00:08:06] the tests are run by jenkins so you can just upload them and let jenkins verify them if you don't want to set it up yourself [00:08:18] but it is just composer update && composer test [00:23:55] ok .. will take a look. [00:25:53] grr ... i have forgotten how I reproduced that crasher locally ... i thought i ran bin/test.php and fed it input in /tmp/Australia.html .. but, i can no longer crash it on that input. [00:30:32] ah .. i was feeding it the "wikitext" as html instead of feeding it HTML generated from the parser. [00:36:12] TimStarling, https://gerrit.wikimedia.org/r/#/c/401412/ [00:40:03] so, how do we get this new version rolled out into production .. does it need a new version and a bump in composer.json in the m/w repo? [00:42:00] yes [00:43:36] tagging the new version causes packagist to create the new package, via a github hook [00:44:16] so you push the tag to gerrit, it is mirrored to github, then packagist picks it up [00:44:30] then you can update composer.json in MW [00:51:14] i tagged it locally ... i assume i have to direct push it to gerrit .. git push origin 1.0.2 HEAD:refs/heads/1.0.2 will do it? [00:52:21] I think I used git push --tags [00:52:33] " All refs under refs/tags are pushed, in addition to refspecs explicitly listed on the command line. " [00:53:05] but whatever works [00:53:06] ok, done. [00:53:14] git push --tags worked. [00:53:50] it's in packagist already https://packagist.org/packages/wikimedia/remex-html [00:55:22] you also need to bump the version in mediawiki/vendor, and then have the core patch Depends-On the vendor one [00:55:28] https://gerrit.wikimedia.org/r/#/c/401414/ [00:55:32] ah, ok. [00:56:05] i need to run now to help with dinner .. can one of you fix it up? :) [00:56:39] there is no remexhtml in vendor/composer.json [00:56:46] sure [00:58:05] TimStarling, i assumed fixing the parser migration issue is a simple thing ... to unwrap the mw-parser-output div but i cannot see an obvious way to do it .. that one is not a rush.. but if it is a quick fix, would be useful to roll that out as well. [00:58:18] alright, off now. ttyl. [14:29:09] subbu: https://gerrit.wikimedia.org/r/#/c/399907/, then add 'unwrap' => true to $poOptions in MigrationEditPage.php. The other option would be to use $parserOptions->setWrapOutputClass( false ), in which case you'd probably want to do like https://gerrit.wikimedia.org/r/#/c/399913/1/includes/ApiQueryExtracts.php so it doesn't throw deprecation warnings if/when that patch is eventually merged. [14:45:45] anomie, thanks .. will take a look later today. [14:45:51] and happy new year. [18:16:39] anomie, I'll wait for that core patch to be merged and deployed and set the unwrap => true after. [21:20:09] Warning: assert(): Assertion failed in /srv/mediawiki/php-1.31.0-wmf.15/extensions/TitleBlacklist/TitleBlacklistPreAuthenticationProvider.php on line 37 [21:59:02] no_justification: Looks like no one who was involved with I83f9374b2f noticed that AuthenticationRequest::getRequestByClass() can return null. [21:59:44] * no_justification nods [22:16:42] Also, I thought we don't use assert() in production? [22:17:10] Ah yes, the person who +2'd that change was a total idiot [22:18:28] I hear that legoktm is a real jerk [22:19:45] anomie: thanks for all the CR btw :) [22:20:04] I'll fix the TitleBlacklist issue tonight if no one beats me to it