[00:00:45] <RoanKattouw>	 Krinkle: I agree the PageTriage code is interesting in the most British sense of the word
[00:00:50] <RoanKattouw>	 I don't exactly endorse that code
[00:01:03] <Krinkle>	 RoanKattouw: TextExtracs extension :)
[00:01:07] <RoanKattouw>	 Also yeah gotta love how it calls strip_tags() twice
[00:01:13] <RoanKattouw>	 Oh right, that could be used here
[00:01:21] <RoanKattouw>	 IIRC I found questionable code there too
[00:01:40] <Krinkle>	 Which isn't all that much better, I mean, it's a complex problem to solve without a semantic way to abstract this for users, but at least it's a centralised approach.
[00:01:44] <Krinkle>	 Yeah.
[00:02:39] <RoanKattouw>	 https://github.com/wikimedia/mediawiki-extensions-TextExtracts/blob/master/includes/ExtractFormatter.php#L57
[00:03:03] <RoanKattouw>	 Please appreciate how that calls html_entity_decode() without any ENT_ flags
[00:03:09] <RoanKattouw>	 That is almost always a mistake
[00:03:54] <RoanKattouw>	 Then it also goes and does its own whitespace normalization, although Sanitizer::normalizeWhitespace() probably handles double newlines differently than what that wants to do
[00:04:08] <Krinkle>	 Google tries their best too - http://i.imgur.com/nT26O1s.jpg
[00:04:23] <Krinkle>	 Had to help it in this case (from a few weeks ago)
[00:05:29] <Krinkle>	 Oh, Jan 2016, not 2017. Have had this screenshot in my dump folder for a while now. Keeps reminding me..
[00:08:06] <RoanKattouw>	 Krinkle: I think I'm done converting stuff to stripAllTags now: https://gerrit.wikimedia.org/r/#/q/topic:stripAllTags+(status:open+OR+status:merged)
[00:36:52] <Krinkle>	 RoanKattouw: thx
[00:37:26] <Krinkle>	 paladox: I haven't forgotten, but there's quite a long backlog before I get back around that one, also due to it being an enhancement, the higher priority backlog may grow before I get down to it, so it may take a while.
[00:37:38] <paladox>	 ok
[00:38:58] <Krinkle>	 paladox: I can spare a few minutes for review, but last I checked it needed help writing some of the code as well. I've described the needed changes as best I can, but I understand this is a new area for you. It might be too high a reach to finish that one at this point.
[00:39:15] <paladox>	 ok
[00:39:16] <Krinkle>	 I can help you find easier ones, but I think you've found a number of other things to work on already, is that right?
[00:39:24] <paladox>	 yep
[00:40:36] <paladox>	 I've been busy with helping try deploying gerrit with scap and other things. Though i think that my patch will benefit wmf as it will allow svg images instead of retina png.
[00:42:14] <Krinkle>	 RoanKattouw: Reviewing and, in lieu of tests, doing some eval stuff. 
[00:42:40] <Krinkle>	 Finding that it is still fairly trivial (by design) to produce html in the output, I assume you know, and that all uses do escape it properly and/or treat as plain text, right?
[00:42:56] <Krinkle>	 e.g. &amp;amp; will produce &amp; and &lt;script&gt; will produce <script>
[00:42:58] <Krinkle>	 as it should.
[00:43:40] <Krinkle>	 order of strip and decode matters too
[00:44:10] <Krinkle>	 but yeah, I think they're all the same, strip first, and then decode, thus allowing such text to leave, which is of course valid text.
[00:58:34] <RoanKattouw>	 Yes, and then make sure to treat it as text, not HTML