[12:33:28] Any idea if I can get the list of words in the page title, in order to create a filter which prevents page creation if PROBLEMATIC_WORD is in that list? [13:00:06] acagastya: page_title [13:00:21] https://www.mediawiki.org/w/index.php?title=Extension:AbuseFilter/Rules_format#Variables_from_AbuseFilter [13:02:27] It gives the words in the title? [13:07:17] It gives the title as a single string, with spaces instead of underscores [13:09:32] Maybe a regex " PROB_WORD " might be a potential solution. Though the way this LTA is working, I don't know if even that is eough. [18:29:48] Skizzerz: I created https://www.mediawiki.org/wiki/Skin:Pandora based on our chats. Might be something of interest to you [18:30:05] thanks I'll check it out [18:30:37] interesting [18:32:16] Jdlrobson: any reason you're using LightnCandy directly rather than TemplateParser shipped with core? [18:34:22] (well, reasons besides "it's a prototype/quick project" :P ) [18:35:24] I should prototype out some of the stuff I'm thinking of [18:46:30] great name for a wiki-editable skin [18:47:42] yeah, could use a bit more love though code-wise [18:48:26] (CodeEditor integration for the .mustache page, locking it down to interface-admins, maybe making it support Pandora.less in addition to Pandora.css, ...) [18:49:00] y'all sound like you're reinventing Wordpress themes ;) [18:49:24] is that a *bad* thing? [18:49:37] haha, no [18:49:39] well, maybe [18:49:40] > maybe making it support Pandora.less [18:49:40] I requested support for .less pages and was shot down [18:49:41] :D [18:49:59] https://phabricator.wikimedia.org/T272774 [18:50:26] DannyS712: it's something that could be supported at the skin level even if it isn't in core [18:50:43] and seeing as this particular skin is literally "go ahead and edit the raw HTML of the skin on the wiki" [18:50:49] supporting less isn't too much of an extra step :P [19:01:09] DannyS712: https://phabricator.wikimedia.org/T56864#630614 I asked for it 7 years ago! [19:07:53] Hi, i'm trying to install mediawiki but i'm indecisive if i would need a better than sqlite database [19:08:17] what's the use case where i would need mariadb/postgres [19:10:51] Elodin: always [19:10:55] :) [19:11:09] sqlite support exists for unit testing, but its usefulness in a production wiki is approximately zero [19:11:34] if you're the literally only one using/editing the wiki, then sqlite is sufficient [19:12:07] alright [19:12:14] thanks for the input [19:13:21] basically sqlite can't support more than one person doing a write at a time, and while someone is doing a write, nobody else can read from it. And this locking is across the entire db instead of just a table/row [19:13:50] so it's largely non-functional for any internet facing site where you may have 2 people doing something on the site simultaneously [19:14:47] ah, i understand. Thanks Skizzerz [19:44:32] > Jdlrobson: any reason you're using LightnCandy directly rather than TemplateParser shipped with core? [19:44:33] Because TemplateParser requires a file on the file system. Would need a bit more abstraction. [19:49:56] hi, asking here because #wikidata is slow -- I'm trying to use CirrusSearch with Wikidata via their docker image setup and it seems that nothing at all is being indexed in elasticsearch. any idea what steps I can take to debug? [19:50:09] https://www.irccloud.com/pastebin/Fy3fjzhg/ [19:59:11] elasticsearch seems to be up and running and cirrus scripts seem able to access it just fine, no errors I can find, and yet when I force it index everything it claims it's indexed it all and then the index is still empty(???) [20:02:40] ningu: I suggest asking your question on https://www.mediawiki.org/wiki/Extension_talk:CirrusSearch (CirrusSearch esp with wikidata can be very difficult to setup) [20:02:51] thanks [20:12:49] it looks like it was the job queue [20:59:37] I've set up the job queue to use redis but it looks like it still isn't automatically running jobs [20:59:47] or maybe it's the number per request that's an issue, still? [21:00:15] yeah, it looks like the job queue is working in redis but still not auto-running [21:09:00] well, this is annoying but I am slowing making progress [21:13:05] you'll still need something like cron or a systemd service to automatically run jobs [21:13:19] that is independent of where the jobs themselves are stored [21:13:45] yeah I see that now [21:14:03] problem is this is in docker and that gets all weird -- at least, I haven't done it before :) [21:18:00] but, at least i know now what I need to do [21:28:29] amazing coincidence that happened: a file named Skin_.... uploaded to a mediawiki instance and placed under /a/ads/ leads to users' advert blockers triggering [21:28:49] since easylist has "/ad/skin_" in it [21:30:27] er. /a/ad/