[00:42:51] doesn't show anything later than january, why is that [06:17:01] [1/2] https://issue-tracker.miraheze.org/T15205#318414 [06:17:01] [2/2] Big CSP PR waiting for feedback, especially on the set of default-enabled external services. [08:59:58] reviewed [10:52:05] is it possible to use weird gloop's fork of datamaps instead? https://github.com/utdrwiki/DataMaps [10:55:10] [1/2] I think that's the UT/DR fork by Kocka. [10:55:10] [2/2] I don't think we should use fork when it's only 4 commits ahead. We should merge these upstream if they are genuine improvements. Unfortunately no one might be around to review them. [11:20:24] Depends if gloop are going to be long term maintainers and take over [11:22:13] It's on gerrit so I'd rather it gets pushed up to Gerrit and @abaddriverlol or @paladox bribed if the maintainer doesn't exist anymore [11:23:25] @alex4401 might be maintainer and seemed around not that long ago [11:24:30] Ye he's still around so I say push to Gerrit and let @alex4401 merge [11:26:22] @whostacking Kocka is not here so you'll have to reach him through other means to see if he's willing to submit to gerrit. You can also submit patches on other people's behalf as long as you use the proper git commands, but it gets tricky if the maintainer asks you questions about the patch that you cannot answer. [11:34:27] I imported the repo to gerrit from our fork so the maintainer is Miraheze + Alex [11:34:59] I poked Kocka [11:35:03] We just don't have a gerrit group yet (should probably request that at some point) so nobody else has +2 in the repo [11:36:09] Probably ye [11:50:55] for clarification, this is kocka's fork and not related to weird gloop in any way [11:51:40] kocka's submitted his patches to gerrit alr as far as i'm aware tho [14:30:58] [1/2] I don't know if this is a script error or cache thing, or W/e... [14:30:58] [2/2] Funnily enough; if the modelviewer is appended on upload, the render does not work. [14:31:09] https://maille.wiki/File:HP2-1Scale.glb [14:31:20] a null edit fixes it. [14:48:42] [1/2] Probably a race condition when parsing the file description. On upload the file corresponding to `{{PAGENAME}}` does not yet exist, so `{{filepath:{{{file}}}}}` resolves to an empty string, and then `data-src` becomes an empty string. [14:48:42] [2/2] When null editing the page is parsed again and this time `{{filepath:{{{file}}}}}` can be found. [14:49:22] Ah; so it's a wiki thing, and needs the null edit. [15:04:15] Can now confirm that it’s resolved with server cache update. [15:28:23] [1/2] > MediaWiki\Extension\UrlShortener\Hooks::onWebRequestPathInfoRouter [15:28:23] [2/2] No way is urlshortener providing more value to MH than e.g. tabber. Is it just on by default or somthing? (necroing because I shamelessly search "bucket" in here every few weeks) [15:30:57] feels like an easy performance win to either turn that extension off in a bunch of places that don't actually use it, or delay the generating of the shortlink to an external request that is fired on some kind of "Get shortlink" button [15:36:49] `onWebRequestPathInfoRouter` is not about generating short links, it's a hook that's called on every request and I think URLShortener uses it to register `/r/` to be able to serve the short links [15:40:27] It calls SpecialPage::getTitleFor() in the hook which triggers the special page list creation and instantiates a bunch of services; ofc that's not great, but a lot of that code would also be run later in many requests so if we were to disable URLShortener some parts of the time spent on the hook wouldn't be saved but instead moved to other places in the profile [15:40:49] ah that makes sense [16:43:00] it is on by default and scrapers hit it all the time [16:43:38] also tabber's definitely a false positive, since it has to parse everything inside the tag [18:10:19] Seems weird to have it on by default but maybe it gets used [18:15:26] Can someone merge the simple PR in this ticket for wiki footer icons for me pretty please? https://issue-tracker.miraheze.org/T15870 [18:34:48] [19:34:25] I can see the usefulness for non ASCII page titles. Would be interesting to see how many short links are ever visited. I would expect it to be like 100 links per month [19:42:58] speaking of, was it ever decided whether to host the short urls on a central database or per wiki? i've been holding off reverting our links to short urls until a decision is made on this [20:21:02] I think we forgot we even needed to decide that tbh [20:21:09] Is there a ticket? [20:21:57] I'm not sure if a ticket was ever made? 😅 should I make one now? [20:22:13] Yes [20:22:23] If there isn't a ticket, I can deny it ever existed [20:22:37] Tickets we can then blame procrastination [20:23:12] would this be a configuration change request or a feature request? [20:24:04] Call it either [20:32:40] ok done: [21:06:28] @cosmicalpha ^ [21:28:56] if a wiki made before the replacetext cutoff date made a dummy edit to every page, would that make replacetext usable on the wiki [21:35:18] as for , is there some special stuff we have to do before disabling smw [22:44:27] I would guess so but I don’t know if replacetext accesses old revision content [22:44:50] but if it only accesses the latest and the latest is uncompressed then I would assume it works [22:51:21] [1/2] idk why it would do old revision content [22:51:22] [2/2] seems to indicate latest revision only