[02:51:54] Hello, how I transate MediaWiki messages (sidebar, as example), I tried to use the example on https://www.mediawiki.org/wiki/Help:Extension:Translate/Unstructured_element_translation/es#Configuraci%C3%B3n but leaves me a blank page [06:53:21] we have a wiki that's getting Main slot of revision not found in database. See T212428. after importing content but slots table has an entry that matches up to main in slot_roles and there's no replica so no replag. [06:53:22] T212428: includes/Revision/RevisionStore.php: Main slot of revision (number) not found in database! - https://phabricator.wikimedia.org/T212428 [15:06:44] would anyone have a little time to help with the above mentioned by RhinosF1 ? it would be useful :) [16:06:19] hello, is it possible to input filenames to parser tag via template parameters? I am trying it but it does not work trying this: [16:06:19] {{{1}}} Looks like template parameters do not work inside parser tags. Any ideas how supply filenames for gallery from external source other than Lua module? [16:07:16] I need to apply this to wikipedia [16:23:20] Kozuch: you'll have to use the {{#tag:}} parser function. See https://www.mediawiki.org/wiki/Help:Magic_words#Miscellaneous [16:28:21] Vulpix, thanks for hint. I just realized I actually can create a "gallery" from plain floating images which works so far and will also be responsive for width [18:48:13] Does anyone know what phan means by "SecurityCheck-DoubleEscaped HTMLForm label key escapes its input" or how to address it? [18:48:42] * hexmode would ask on a better channel if he were pointed to it [18:59:33] hexmode: I think it means you're escaping some input, and then passing it into an API which escapes it as well, causing the input to be double-escaped [18:59:52] solution is to not escape it on your end before passing it into HTMLForm [19:00:04] That's what I would think, too, but I don't see that happening. [19:00:31] Skizzerz: 1s and I'll show you where this is happening [19:00:34] the initial escaping could be happening further upstream perhaps, or phan is just wrong :) [19:06:36] phan output: https://pastebin.com/6pKTv2jW [19:07:27] for this change: https://gerrit.wikimedia.org/r/c/mediawiki/extensions/LDAPAuthentication2/+/677285/9 [19:09:07] hexmode: specify a thing for the wfMessage calls for label [19:09:34] I think by default Message::__toString() will html-escape the text. Which then gets html-escaped again by HTMLForm [19:10:15] you may wish to verify this yourself however before taking my word for it on what the default behavior of wfMessage() is without specifying e.g. wfMessage()->raw(), wfMessage()->text(), etc. [19:11:01] e.g. at https://gerrit.wikimedia.org/r/plugins/gitiles/mediawiki/extensions/LDAPAuthentication2/+/51f044874559f417435b4ab43f27f683115594c5/src/ExtraLoginFields.php#22 [19:12:47] k let me see if that does it [19:16:53] you're right! Thank you very much! [20:06:22] Message::__toString() is equivalent to ->escaped() by default [22:25:51] legoktm: dancy: Regarding rebuildLocalisationCache --offline, was the chain to disable DB access itself (apart from mesageblobstore purge skipping) preventative in nature, or was that the first pass and later we found it was only triggered by the purge action? [22:26:17] It looks like the tasks started with Gadgets extension accessing the database etc so I suspect perhaps the former. [22:26:28] (which is part of the purge operation) [22:27:04] I assumed they ran into it while getting the images to build the l10n cache and just repurposed the Gadgets task since it was the same underlying issue [22:27:08] when the db is unavailable, the script will fatal either way whether disabled or not, so that part should make no difference, but I'm wondering what else changed in terms of runtime behaviour as part of this - if anything [22:28:37] if there are other, truly optional, default things that the maintenance script does unrelated to localisation that we wanted to skip, then I'd like to know what those are, and perhaps we can just skip those by default instead of with this option. [22:29:16] I've proposed renaming the option back in https://gerrit.wikimedia.org/r/c/mediawiki/core/+/677375 to avoid misuse. [22:33:30] personally I think there's nothing wrong with having this script usually be DB_NONE unless you're using LCStoreDB [22:33:33] and make that the default [22:35:29] legoktm: that would break it for anyone who uses some kind of DB involvement for any part of their setup or hooks or later etc. [22:36:03] I don't think we should drop support for that right away [22:36:06] or possibly at all [22:36:48] but maybe we an approach it from a different perspective. rather than the --offline option doing anything different, it would just make the DB access a fatal explicit error. Bascially what it is already (except for the skip purge part). [22:37:05] but does that mean DB_NONE as a concept is never usable? [22:37:06] That would mean it skipis the servie intiliaisation, which I'm guessing is the other part that we ran into (\cc dancy ) [22:37:40] so its opt-in and up to the user to make it work without a db, possibly by setting -skip-message-purge whicih for WMF will suffice [22:37:43] but there may be other things [22:37:57] so it's more a promise from the user to us, than from us to the user. [22:38:08] yeah, I don't expect DB_NONE to generally be something that works with MW. [22:38:57] not as something to enable on random scripts that is [22:39:18] but things like minify.php and other standalone utils that use MW as a library, sure. but anything that constructs singltons or servie wiring, no.