[01:24:49] anomie: https://logstash.wikimedia.org/app/kibana#/doc/logstash-*/logstash-2018.03.06/mediawiki?id=AWH44pdnp0cDAFqtXrJc&_g=() [01:31:31] James_F: https://etherpad.wikimedia.org/p/mw-php70-tests look good? anything else to mention? [01:38:27] tgr: Crazy rows from 2005 in the ipblocks table: select * from ipblocks where ipb_by = 0 and ipb_by_text = ''; [01:41:23] should they just be deleted? after 13 years those blocks don't seem too valuable [01:57:32] Either that or set ipb_by_text to some system user. Whatever code made those blocks in 2005, if it still exists, probably already created one. [02:32:22] TimStarling: Are we going to do the 1:1, or did we talk about everything in the team meeting? I feel like I'm good if you want the latter. [02:32:40] right, sorry [02:32:44] we will have a 1:1 [03:04:02] Krinkle: the globalcssjs + core patch look good to me, but it will still get loaded in global state, correct? [03:04:27] legoktm: nope, not unless we want to preserve that through other means. [03:05:12] addModules vs addModules() only difference is whether to use only=scripts which means 1) dedicated request, 2) global scope, 3) ignores any syles. [03:05:14] styles* [03:05:48] how are we currently ensuring global state for the 'user' module? [03:05:57] addModules() means it doesn't do any of those things. It will still get its own request usually, but only because it tends to be the only thing of the (source=metawiki, group=user) tuple. [03:06:36] legoktm: The 'user' module is essentially handcoded to wrap getScript() in the UserModule class with $.globalEval. [03:06:53] Essentially class UserModule subclasses WikiModule::getScript() and wraps in $.globalEval as the script. [03:07:00] Not generic in any way, but specific to that one module. [03:07:21] Which we can do for GlobalCssJs if we want that, but from what I understood, you don't want that. [03:09:28] right. [03:09:37] ok, I was missing that it happens in UserModule [03:10:05] people can test the lack of global state by just wrapping their global.js in function() { ... }(); right? [03:11:18] (I haven't tested the patch yet, I'll do that tomorrow) [03:14:43] legoktm: Right. Wrapping global.js in (function() { .. }()); will produce the same effect as merging these patches [03:17:31] ok, we can recommend that to people in the announcement so they can find breakage ahead of time [03:18:35] https://meta.wikimedia.org/w/index.php?search=intitle%3Aglobal.js+insource%3A%2F%28var%7Cfunction%29%2F&title=Special:Search&profile=advanced&fulltext=1&ns2=1&ns8=1&searchToken=47055n9ksl9w0iixq4t3imlj [03:19:06] 750 pages on Meta-Wiki that use var or function, which is what could be a global right now if there is no closure [03:19:18] so far all the ones I spot-checked either have a closure already, or only use it locally [03:19:29] it=the variable or function [09:06:18] <_joe_> is anyone around? I have a doubt regarding mediawiki-config and scope [09:08:00] <_joe_> specifically, here https://gerrit.wikimedia.org/r/#/c/416470/3/wmf-config/CommonSettings.php I'm not sure if I need to declare the variable as global in commonsettings.php [09:08:06] <_joe_> outside of the callback I mean [14:13:46] _joe_: I'd think not, since nothing else gets declared as global where it's defined outside callbacks. [14:15:32] <_joe_> anomie: I reconstructed the whole chain of includes, and CommonSettings.php gets called directly from LocalSettings.php, so I think it's ok not to declare the variable as global there [14:15:38] <_joe_> since you're in file scope AIUI