[10:51:58] Hello, I have a problem with my mediawiki installation after upgrade and move of server. Pages with a ' (single quote) in the title. I cannot access. However all other pages are fine. Chrome says: "ERR_TOO_MANY_REDIRECTS". Any ideas of where to look? [10:54:11] RogueFiber, could you add a comment to https://phabricator.wikimedia.org/T106793 about your browser etc? [10:56:26] Will do! [14:33:18] I have an old installation of mediawiki (v1.16). Upgrading isnt something that I am currently able to (easily) do. Have a problem that I have not been able to find the answer to and hoping that IRC can help relating to email notifications of watchlist items. Am finding that sometimes the email links show http://wiki/ whereas other times (and this is what I want as the default), it is http://wiki.judouk.internal/. Anyone know what cou [14:34:39] judouk: your message got cut off after "Anyone know what" [14:35:06] judouk: i would guess that you do not have $wgServer set, and it's being autodetected to the weird value. [14:35:37] judouk: but if your wiki is meant to be accessible through two domains, you probably can't fix that without upgrading, as only MW 1.18 introduced $wgCanonicalServer [14:39:55] MatmaRex: the last bit only said "Anyone know what could be causing this?" [14:40:32] MatmaRex: I'm wondering if the cause is down to how the editor opens the URL - if they use the short name vs fully qualified [14:40:41] and then when the page is saved, the URL is passed through [14:40:50] probably, yes [14:41:42] so, I wonder if hard-coding the fqdn would fix this [14:50:06] MatmaRex: And the problem was fixed by adding two more virtual hosts ! [14:50:09] thanks for your help [14:54:40] hi, is it possible to let {{DISPLAYTITLE:}} use wiki links? [14:55:05] sp that page title contains links to other pages? [16:06:50] Hallo all! I'm having trouble finding a wfmsg page and hoping someone can point me in the right direction. It's in the same vein as MediaWiki:Img-auth-accessdenied but says You are not allowed to execute the action you have requested [16:07:02] Search isn't having any luck and my googlefu isn't strong for this one [16:07:12] ?uselang=qqx [16:07:21] ..well played [16:07:30] /Special:AllMessages [16:08:52] Wait, no! I didn't forget qqx, it just doesn't work because it's an img_auth rejection [16:08:56] trying all messages [16:09:09] aha [16:10:07] "badaccess-group0": "You are not allowed to execute the action you have requested.", [16:10:10] Ulfr: ^ [16:10:41] Reedy: Once again, you are my savior [16:10:54] Now I get to wait a week while there's several meetings to determine what the hell they should say [16:13:44] note that badaccess-group0 is used in a ton of places, not just img_auth [16:14:31] (notably, it's used whenever you try to execute something that requires a permission you don't have, and no other groups have that permission either) [16:15:19] or actually, even if other groups do have the permission [16:15:25] the list of groups is a different message [16:15:50] "badaccess-groups": "The action you have requested is limited to users in {{PLURAL:$2|the group|one of the groups}}: $1.", [16:17:18] Well, if it's the village bicycles of error messages indicating one should pound sand, I feel as though it should be a little easier to locate [16:17:41] It's pretty easy [16:17:49] Reedy: in a few cases, it shows group0 even if there are groups; but those cases are pretty rare [16:17:52] If it's a core message, look in en.json [16:18:12] Reedy: Well I KNOW it's easy for you, there's a reason I come visit when I'm stumped :P [16:18:17] (userCan hook denying access, attempting to use the hideuser feature when blocking, some api things) [16:18:31] https://github.com/wikimedia/mediawiki/blob/master/languages/i18n/en.json [16:19:11] grumble. it's a message that you edit thru the interface, no reason it shouldn't be searchable when I search the mediawiki namespace [16:19:14] so, rare enough that it's probably ok changing the wording to something img_auth specific without needing to worry about coming across the message elsewhere [16:20:02] Well it's the only instance this error is utilized in this application, (I hope) so I'm just going to change it all willy nilly like [16:20:10] Ulfr: mediawiki messages are not stored in the database by default (only when customized) [16:20:21] since the search feature searches the database, it therefore won't find uncustomized messages [16:20:41] Skizzerz: I never doubted there wasn't a good reason the universe doesn't bend to my whim, but I reserve the right to feel justified in needing assistance [16:20:48] sure :) [16:20:51] :D [16:21:09] Special:AllMessages?limit=5000 [16:21:12] Ctrl + F [16:21:13] Profit [16:21:26] Reedy: Pfft. I did that when you sent me the special page [16:21:35] Guess what was missing [16:22:09] * Reedy invoices Ulfr [16:22:11] but that may have multiple pages, that means you might need to ctrl+f more than once! the horror! [16:22:31] Skizzerz: There are in total 9 separate messages that come up from that special page [16:22:53] * Ulfr might be American, but does his best to not be willfully ignorant [16:23:20] * Skizzerz goes back to stabbing flaggedrevs with a rusty spatula [16:23:28] hi, I am writing a special page for an extension, just a class extending the SpecialPage class. However, when inside a function (inside that class), I get an error called "Using $this when not in object context", which prevents me from getting an OutputPage object via $this->getOutput() [16:23:31] Is that normal? [16:23:46] static function? [16:23:55] you can't use $this in a static function [16:24:13] it's not a static function, I think? [16:24:20] (if you want to call another static function from your static function, use self::blah()) [16:24:29] Skizzerz: Okay, there's a lot more than 9, I forgot about a filter. But badaccess0 isn't on there still [16:24:32] throw the code on gist? [16:24:35] sure [16:25:01] also give line # of error [16:25:13] https://dpaste.de/aGsy [16:25:13] Ah, plumbing the mysteries of how to reference oneself in a variety of contexts. Some days I really loathe object oriented logic [16:25:16] line 113 is the error [16:25:48] ->setSubmitCallback( array( __CLASS__, 'onSubmitInput' )) [16:25:51] You're doing a static callback [16:25:55] Swap __CLASS__ for $this [16:26:11] your setSubmi... what Reedy said [16:26:23] oh bah [16:26:42] I always thought __CLASS__ was a nice way to avoid typing long class names, never thought it would cause issues [16:26:51] thanks, you're great <3 :) [16:26:58] SPF|Cloud: On the bright side, you didn't spend two hours troubleshooting why string concatenation in php doesn't work when you use + instead of . [16:27:06] It's basically doing __CLASS__::onSubmitInput() [16:27:31] Ulfr: haha. I spend much time investigating this error too, though [16:27:45] well, yeah, but that's like a legitimate headscratcher [16:27:56] not trying to php too soon after becoming one with the jquery gods [16:28:04] Ulfr: Many libraries do that [16:28:06] like wut [16:28:26] I blame PHP [16:28:33] it's far too loose when casting strings to numbers [16:28:48] should emit a warning if the string has garbage characters that were discarded when casting [16:29:24] I feel like it could be resolved easier if you had to actually assign a type to a variable [16:29:29] rather than being all loosey goosey [16:29:35] hack [16:29:39] or use a decent ide [16:29:46] or use a different langauge :) [16:29:48] PHP 7 has type hinting? [16:29:51] *language [16:30:17] Grumble. I'm working on upgrading to 7. It better serve me tea while it allegedly runs twice as fast [16:31:59] It wouldn't be nearly as bothersome if I didn't have to upgrade mysql at the same time. That's what I find terrifying [16:33:09] mysql isn't suually too bad [16:34:04] It's less that it's not usually too bad and more the architecture I inherited isn't precisely straightforward and it's a fairly high load system [16:34:34] When I putter with frontends, usually I can get away with a couple mistakes. If I splat the database ain't nobody happy [16:36:01] Like, it's hosted on AWS, and near as I can tell the backups that it runs are accomplished by elves, and the volume management is equally wizardlike because there's 8 EBS stores that go all power ranger like [16:39:30] Ulfr: always make a dump before upgrading anything. actually, make two: one for mysql (don't forget --default-charset=binary) and an XML dump of the content. [16:41:05] DanielK_WMDE: Oh, trust me. I'm going belt, suspenders, second belt, and anti-gravity underwear when I muster up the courage. The part that absolutely mystifies me is how the existing template makes backups happen. It seems like it takes images of all 8 drives and recombobulates them on demand xD [16:50:23] Ulfr: that does sound scary >_< [16:51:05] DanielK_WMDE: Like, someone wanted me to recover a specific table's data from a certain time. Then I discovered that the EBS stores are striped and paired. Then I had several drinks [17:10:48] Ulfr: sounds like a two pipe problme ;) [17:11:31] If you listen VERY closely you can hear the sound of that going over my head [17:14:57] Because I'm not going to lie, that data was lost foreeever [17:51:12] i once saw a page that allowed me to pre-create useraccounts and invite ppl via mail, but i cant find it anymore... anyone knows where that was? [18:32:57] Hi there. I'm using Mediawiki with $wgAuth = LDAP, but I need to create a local user with read-only access only. MultiAuthPlugin seems to solve another problem. Any tips? [18:35:33] * Ulfr grumbles [18:35:45] So that whole badaccess0 thing really doesn't want to have a link inserted, does it [18:39:39] silas: Be advised that as of 1.27 $wgAuth is being deprecated in favour of a more flexible authentication model [18:40:25] Yet another victim of the depreciation monster [18:41:52] For the uninitiated, I'm modifying a wfMessage page on my wiki, and it stubbornly refuses to let me encode a link. The message in question is MediaWiki:Badaccess-group0 and I have $wgRawHtml enabled [18:44:37] Well, looks like that escaped call in img_auth says I can't do that. Boo. [18:44:42] Thanks anyways! [19:22:54] so.. Windows Server 2012, Apache 2.4, PHP7.. anyone successfully install MediaWiki 1.26 on this platform using an Oracle 11g database? [19:23:29] It does beg a "why?" [19:24:38] You had me up until Oracle 11g... [19:24:40] Oracle support isn't good in MediaWiki [19:24:54] And by "not good", we mean horrible [19:25:27] I'd suspect, at minimum, the database schema would be out of date [19:25:41] I'm not sure if sql server would be much/any better [19:26:00] I know this. but since they spent huge dollars on Oracle, they feel that there's no reason not to use it, even though it's not supported. "The DBA's aren't familiar with MySQL" and would be annoyed. [19:26:25] Well, if it's not supported, you can't really use it? [19:27:00] oracle's used for all sorts of other functions in our company, not just this. [19:27:26] Unless you've a PHP developer on staff, it's probably not worth the effort [19:27:39] That's what I told them. I'm just checking everywhere just to make sure I can give them a fully researched "I told you so" [19:27:58] The person who used to previously support Oracle isn't about much now [19:28:37] I don't even know if he's using it now [19:29:58] interestingly enough, I'm kind of suprised that they gave me an 11g database to point at, since they're supposed to be using 12c. but hell, we just had to re-install some 6i forms in a special way to not import the v5 baggage that comes with them. and some apps still require java 1.6.. so you know the kind of pain I'm dealing with here. [19:30:31] Our oracle support is so bad, its probably even money if mediawiki still works with oracle [19:30:51] Jorden: What sort of load is the wiki going to be having? [19:31:10] As if it's not huge, mysql will require almost no management [19:31:58] not super-huge I think. It's meant to be part of a collaboration initiative on an intranet [19:32:20] And if it is huge, our oracle support is probably not performant [19:32:25] lolol [19:32:42] lots of procedural/process documents for many manufacturing plants. [19:32:49] MySql backend is the only one that gets tested for performance [19:33:12] Sqlite, postgress, and Mysql are the backends that work, but mysql is the only one that's "fast" [19:35:16] I'm just a server infrastructure guy, but got tagged to be the one to set this up for them. They have an 'already decided on mediawiki with oracle' mindset, so I'm just triple-checking the info that I can find to make them do it with mysql. [19:35:37] I don't think the oracle layer has been maintained for years [19:35:52] microsoft sql server should work, at least with 1.27 [19:36:03] but I recommend mysql [19:36:40] freakolowsky hasn't touched it since Feb 2014 [19:36:41] https://github.com/wikimedia/mediawiki/commits/master/maintenance/oracle [19:38:23] was freakolowsky the main Oracle guy on the project? [19:38:30] Yup [19:38:38] Main/only [19:39:19] ok, that link is my ammo to tell them to get off the oracle bandwagon! Thanks much! [19:39:51] which reminds me that I still need to finish up that RFC about moving the db abstraction drivers out of core (a la skins/extensions) [19:40:28] just bundle the well-supported ones with mw, and users can download the less-supported/experimental ones on their own with knowledge that it may not work right [19:40:58] downside is that it makes schema change patches a pain to do (since you'd have to submit to like 5 different repos) [19:42:01] Later(tm) [19:43:02] thanks! [19:47:32] Skizzerz: You say that as if people actually write schema change patches for things that aren't mysql/sqlite :P [19:47:44] lol [23:09:58] hi, someone here? I get this error while running runJobs: PHP Fatal error: Unsupported operand types in /var/www/mediawiki/extensions/CirrusSearch/includes/Job/ElasticaWrite.php on line 46 [23:10:29] is this some bug? [23:11:12] PHP Fatal error: Unsupported operand types in /var/www/mediawiki/extensions/CirrusSearch/includes/Job/ElasticaWrite.php on line 46 [23:11:20] sorry for dupes [23:16:35] MaxSem: ? [23:16:44] MaxSem: just joined, so i dont have backscroll :P [23:16:52] auvajs, ask ebernhardson [23:16:54] What's the line of code? [23:16:55] parent::__construct( $title, $params + array( [23:16:55] 'createdAt' => time(), [23:16:55] 'errorCount' => 0, [23:16:55] 'retryCount' => 0, [23:16:56] 'cluster' => null, [23:16:56] ) ); [23:16:56] ? [23:17:08] PHP Fatal error: Unsupported operand types in /var/www/mediawiki/extensions/CirrusSearch/includes/Job/ElasticaWrite.php on line 46 [23:17:24] if thats the line of code, it means $params isn't an array [23:17:30] as for why ... [23:18:28] both invocations of that pass an array explicitly, so something in job queue is broken? [23:19:01] yea i just checked as well and all invocations create arrays when they call it, so it's certainly an array. best guess would be some sort of job queue brokenness [23:25:48] whatever, I just rerun runJobs, but since I have some 100k jobs, I must do it in a loop with --maxtime, otherwise I would never finish the queue [23:27:13] if you end up losing search updates, you can use https://wikitech.wikimedia.org/wiki/Search#Recovering_from_an_Elasticsearch_outage.2Finterruption_in_updates to try again later [23:28:33] ebernhardson: I was uploading a lot of new content into the wiki, 40k articles, the updates are probably not yet ready [23:30:10] as I said, I'll try running runJobs script in a loop and go to bed now, let's see if it's ready when I wake up [23:30:15] gn :) [23:35:22] good luck :)