[00:00:02] When an account successfully passes authentication without a mediawiki account, an account is autocreated for them [00:01:06] * @param User $user User being created (has been added to the database now). [00:01:16] That sounds slightly suspicious in wording [00:02:02] So autoCreatedAccount is called here [00:02:03] https://github.com/wikimedia/mediawiki/blob/master/includes/auth/AuthManager.php#L1736 [00:02:11] From the autoCreateUser function [00:02:41] On line 1698, the user is added to the database [00:03:31] So by line 1736, which is where autoCreatedAccount is called, the user is indeed in the database [00:04:47] And autoCreateUser is called from continueAuthentication, on line 621, as long as the ID of the user that logged in was 0. [00:06:44] I've not done any mediawiki development before this extension. So this has been a lot of digging around and learning for me, but I must say that I'm impressed in this new Auth system. Looks far better than what (it seems to me) existed before. [00:07:30] Yeah, it's definitely a lot better than it was before the rewrite :) [00:10:10] Ok well I've added a bit more checking to the code that requires the setting... Hopefully this isn't something that visits me again in the future [00:10:42] So Reedy, are you aware of the process to 'publish' an extension? [00:10:52] Not really [00:10:57] Create a page on mediawiki.org [00:11:16] Ahh ok no worries :) [00:11:21] Thanks for your help with this [00:11:55] I'd be intrigued to see why it's not actually working from there [00:12:13] I'll just test it a bit more with some fresh accounts I think [00:12:19] It *should* work [00:12:37] And I'm sure my IT department (my best mate) won't mind me making a million new accounts in Active Directory [00:13:13] DeleteME [00:13:16] DeleteMeToo [00:13:21] WhyArentYouDeletingMe [00:13:53] * st_prg deletes Reedy [01:08:21] is it just me, or does Echo really not notify me of the moment when the recipient read the message that i sent to them via a username ping? [01:08:35] there is a markasread action, but it doesn't seem to generate a notification to the sender [01:31:59] Sveta: that would be a privacy violation [01:46:04] st_prg: one possibility is that you are using an outdated cached User object, and the CAS check fails [01:46:35] I think these days that triggers an exception but in older branches it just complains and fails silently so check your logs [01:47:10] tgr: I'm running the latest version :) [01:47:28] What is the CAS check? [01:50:20] MaxSem, really? email has read receipts [01:50:52] compare-and-set [01:51:30] the user record has a version counter, the save fails if +1 is not larger than the current version [01:51:57] or maybe we use timestamps in core [01:53:29] alternatively, something else has an outdated user object and does its own save and restores things? [01:53:57] I'd add some debug logging to User::saveSettings() / saveOptions() and see if that turns up anything [03:47:00] tgr: Thanks for that - I'll do that [07:26:04] hello [07:26:15] can i ask here [07:32:25] no :) [13:32:31] Hm. Upgraded to 1.30.1; Everything working 'cept for Contributions pages. [13:32:42] Getting "Fatal error: Class 'OOUI\Theme' not found in OutputPage.php on line 3986" [13:33:05] Can't see an obvious way to circumvent this OOUI thing that I assumed was optional. [13:35:03] Well, not, optional, but. A way to make it work, more relevantly. [13:37:21] flow says me "Insufficient permissions to execute this action." when i trying to delete a topic but this topic already was deleted [13:37:45] sgtlion: Missing composer stuff? [13:38:25] I guess that sounds plausible. Any way to check? Just reinstalling oojs-ui through composer? [13:38:29] also it don't activate delete button if i use ctr+c ctrl+v [13:38:40] MW should be declaring a dependancy on it [13:38:44] so composer update --no-dev [13:42:49] Hm, well, for some reason it wasn't, I guess. [13:43:01] require oojs/oojs-ui:* seemed to get all on that and sort it ou. [13:43:17] That's really not the way you should be fixing it [13:43:38] https://github.com/wikimedia/mediawiki/blob/REL1_30/composer.json#L28 [13:43:54] N-No? This is what I get for not reading enough documentation. [13:44:21] Oh, right. I remember messing with that line now. So it *was* all my fault. [13:44:37] That particular version was conflicting with something else. [13:44:49] https://www.mediawiki.org/wiki/Do_not_hack_MediaWiki_core [13:44:59] Dang. I am a bad person. [13:45:16] Also, if it's a bug in the released version, we like to know so we can try and fix it for the next one :) [13:47:51] Oh, right, the oojs wouldn't update because of a constraint. [13:47:55] I am not familiar enough with composer. [13:49:02] "The requested package oojs/oojs-ui (locked at v0.23.0, required as dev-0.23.0-no-at-ease) is satisfiable by oojs/oojs-ui[v0.23.0] but these conflict with your requirements or minimum-stability." was the issue I was circumventing with that. [13:49:33] It was not an error message I knew how to resolve correctly, so I just fiddled. [13:56:56] Why that error occurs, I ain't sure. It came up trying to upgrade SMW to 2.5.8. [14:47:33] I can't put my finger on it, but I feel there's deficiencies in the haskell syntax that make things harder than they need to be for me [14:47:38] Dodek: uh [14:47:41] or how does it work [16:13:06] Possibly in addition, yeah - but I think aggregation is the key requirement. [16:13:13] unbound may be dns pledged aswell [16:13:17] right, I see - as in, it will actually initialize it by calling the constructor, not just set all of its memory to 0 [16:32:25] davor, i think you'd have to modify it, however slightly. [16:32:30] chksum, yep. [16:37:23] eh looks like i'm going with fzf now [16:37:29] the hash join is slow, yes, but that's mainly because you're querying the whole of both tables [16:37:37] in the case of an s command, you can add a p flag on it [16:37:45] sigh [16:37:48] this again? [16:44:55] Skizzerz: sounds like you need to block spammers again? [16:45:12] yeah :( [16:45:26] I really hate closing off the channel to unregged users though [16:45:42] because there are plenty that come here legitimately and need support [16:46:23] true that, but then you have the problem where if the spammers continue you also drive away the legitimate users [16:46:24] :P [16:46:28] pros/cons :| [16:46:35] (the catch 22 of IRC / chat moderation) [16:48:01] I wonder if I can make a helper bot or something... messages unregged users that if they need help they should send a command like !helpme [16:48:08] if they do, bot gives them voice so they can talk [16:48:12] since the spammers won't be doing that [16:48:45] but that involves having time to code, which is something I am in very short supply of at the moment :( [16:49:19] Skizzerz: indeed. Just a thought though but for spammers is there a reason you didn't op Sigyn? Because if you +q $~a and +z you ops can still see messages, but if you also op Sigyn you can squish the psammers hitting many channels [16:49:27] but meh, not going to quesiton how the channel is run :) [16:49:30] *returns to lurking* [16:49:31] Sigyn was opped before [16:49:42] I'm guessing the split from a few days back caused her to disconnect [16:50:01] (what you described was exactly what we were doing) [17:58:16] why would you use (( instead of [? [17:58:20] THE_GFR|WORK: ensuring consistency, writing encrypted data, going through the journal...who knows? [17:58:26] well, in constructive logic, "not not X" does not imply "X" [18:02:02] meh [18:05:10] what's going on [18:05:32] Spam attack [18:54:24] How do you trim whitespace around a token using Megaparsec? [18:54:33] My what a large URL you have. [18:54:37] haha ok I clicked. There was one goatse and a dick pic [18:54:45] sub watt stuff