[03:28:31] i wonder if we can use [tinyproxy](https://tinyproxy.github.io/) for it? it has a path-based reverse proxy mode [03:28:41] or we can bastardise nginx [03:39:45] That sounds redundant, what about NGINX isn’t already bastardized [03:40:09] we haven't used it as a reverse proxy for external services yet [03:43:33] cc @rhinosf1 lol, i don't think scribunto would upstream a base64 encoder when they hear about my use case [04:00:45] It seems that CSP is broken for wikis that have a custom domain but not a 301 redirect for every page? Some js requests would go to the custom domain, which is not in the CSP, so they will get blocked and fail to load. [T13627](https://issue-tracker.miraheze.org/T13627) [06:19:59] It's not broken, the old Miraheze.org subdomain isn't designed to be fully functional [06:20:24] We allow access to it so things aren't completely down if your domain breaks [06:20:30] But we don't guarantee it'll work [06:55:39] Thanks for the explanation. In that case, it might be good to nudge anyone who requests a custom domain setup to do a 301 redirect as well (or simply do both of them as the custom domain setup process) since this seems to be a saner default. But, as you said, if the domain expires/breaks, there would be no way of visiting the wiki. [07:58:16] That's exactly why we don't [07:58:29] So there's at least some level of access if the custom domain breaks [07:59:12] Could possibly add a tickbox on RequestSSL as long as it's clear [07:59:15] @cosmicalpha [18:50:51] Perhaps it'll be nice to document this in the [custom domains page on meta](https://meta.miraheze.org/wiki/Custom_domains) either as an optional 3rd step or as a part of the FAQ? Neither this page nor the RequsetSSL form say anything about redirects besides a link to Phorge, so people won't know why they would want/not want to have a redirect set up. [20:35:23] * The Wikimedia /page/data-parsoid REST API endpoint is no longer in use and will be deprecated. It is scheduled to be turned off on June 7, 2025. - We don't use this right? [20:37:11] [1/2] Ooh... Can we get this?: [20:37:11] [2/2] ```The "Get shortened URL" link on the sidebar now includes a QR code. Wikimedia site users can now use it by scanning or downloading it to quickly share and access shared content from Wikimedia sites, conveniently.``` [20:39:48] The task is still open so nobody has implemented this yet [20:40:36] oh wait it has apparently been implemented [20:40:49] https://meta.wikimedia.org/wiki/Tech/News/2025/20 [20:42:21] this has not been backported to 1.43 [20:42:27] it is only available in the 1.44 version of the extension [20:58:33] Features are not routinely back ported [20:58:58] We don’t officially offer that service. The redirect issue is more of a caching issue. You’ll keep getting the original Miraheze domain to work and not redirect for a few hours, maybe days. Some people are impatient and think that it not working immediately means it won’t ever work which isn’t the case. [20:59:49] You don't need to care about any Parsoid rest endpoints being turned off [21:00:19] Mediawiki has been accessing Parsoid directly for us for years [21:01:57] Wikimedia is scare mongering with all the parsoid stuff [21:01:59] Boooo [21:03:58] On the QR code stuff, we can enable it already in the sidebar [21:04:16] It's just becoming integrated into the short url option rather than its own separate one [21:04:36] No idea why we never did but it's existed since 1.41 @abaddriverlol @rodejong [21:05:04] It's useful to Wikimedia but not us, Ro likes to blindly copy stuff from Tech News [21:05:22] Recurring theme from miraheze [21:05:56] Amazingly we're ahead of the curve on this on [21:08:20] While the answer in this case is partially cause we're about 18 months behind, in most cases, the answer is yes in about 6-12 months. File a Phab task and we can tag it for the next release if it's super exciting. We will forget if you ask on discord. [21:09:02] No I like to read the updates on the tech front. I just lack the (complete) understanding, and I am not always sure whether we use something, or should use something, etc. 🤷🏻‍♂️ [21:09:29] Yes, you blindly copy it here whether it's relevant or not [21:10:00] I've seen you copy updates on things being removed before months before they'd be useful to us [21:10:07] Not blindly, as I cannot always judge whether it is relevant for us or not [21:10:13] That's blindly [21:10:23] Without the knowledge of whether it's relevant is blindly [21:10:35] I read, think "Ooh... What would that mean for us" [21:10:39] Tech/news is almost never going to be immediately relevant to us [21:10:42] And then write [21:11:14] Better asking too many times, than not is my motto [21:11:20] If you see an exciting feature mentioned, feel free to flag it for us on Phorge and we can add it to the upgrade task [21:11:35] Otherwise, it's almost certainly either irrelevant or months too early [21:12:08] We don't need to be asked about things like endpoints going away and whether we use them [21:12:15] I will only create a Phorge, if I know it is relevant, and I personally want it [21:12:28] That's why I ask here first [21:12:41] If it's an exciting feature, file a feature request [21:12:55] Like the QR code stuff [21:13:03] Otherwise, just ignore it [21:13:25] It's nearly always going to be in the category of not relevant or months into the future [21:13:49] It's a list of things hitting Wikimedia sites +/- 1-2 weeks of now [21:14:08] It's not going to affect us for months in the best case as I've said before [21:14:19] This is send out tonight! 😛 [21:15:12] Yes [21:15:53] So given we're at the start of the release cycle, going to be like 9 months before it matters to us @rodejong [21:16:45] This week is wmf.1 [21:17:18] We won't even start testing it until the final wmf cut which is wmf.25 [21:17:28] And on production 2-3 months after that [21:17:39] Towards the end of the year [21:18:44] So anything in coming this week will be about December before it reaches us even if it's relevant [21:19:09] If it's in the last week, it should already be on beta but prod won't see it for likely a few months [21:19:10] And likely it will have been abandoned by wmf25 lol [21:19:42] It wouldn't be a wmf "owned" feature if it wasn't half developed and without a maintainer [21:19:49] It really pisses me off when they introduce something and then remove it without deprecation in the next release [21:19:53] Like bro come on [21:20:10] The wmf doesn't do code stewardship [21:20:36] Now it makes sense why Fandom ran such a heavily forked version of 1.19 lol [21:20:40] Literally [21:20:51] Code stewardship reviews have been on hold for 18 months [21:20:54] And going [21:21:36] @originalauthority [21:21:43] What a mess [21:24:06] My favourite thing about WMF is when they write code which is marked as temporary and is still there 10 releases later [21:24:27] 10? [21:25:17] I think there's temporary solutions older than me @originalauthority [21:25:45] Ig if it isnt broken dont touch it [21:26:11] [1/2] > In 2002, Lee Daniel Crocker rewrote the code again, calling the new software "Phase III". Because the site was experiencing frequent difficulties, Lee thought there "wasn't much time to sit down and properly architect and develop a solution", so he "just reorganized the existing architecture for better performance and hacked all the code". Profiling features were added to track do [21:26:11] [2/2] wn slow functions. [21:26:16] It ain't changed since [21:26:37] Yeah, i feel all they're doing recently is reorganise stuff and parsoid [21:26:41] There's never been a properly architectured solution [21:26:44] Other than that nothing much is going on [21:26:56] (Except vue tbf, but that fucking sucks, thanks ResourceLoader) [21:27:05] And how's that going? [21:27:14] You seen the list of stuff that don't work with Parsoid ? [21:27:34] I'll be retired before that list reaches an acceptable number [21:27:45] Yeah they gone fucked up with that [21:28:59] It's only been in development in some form for over a decade @originalauthority [21:29:53] Yeah; i feel the whole problem would have been avoided had they actually thought about architecture from the beginning and used something more respectable like markdown or something [21:29:55] I was in primary school when Parsoid started development in its first form [21:30:17] Also what a dumb fucking idea to initially implement it in nodejs? [21:30:28] I know [21:31:30] This is not an impressive fact [21:32:40] Anyway good night @originalauthority [21:33:46] Gn 👋 [21:44:41] Huh