[07:33:23] I have no idea lol [07:33:38] I spent awhile on this today. I have no idea what could fix it... [07:33:58] Deleting and undeleting page? [07:34:13] one thing I didn't try... [07:47:59] oh maybe I did fix it lol https://issue-tracker.miraheze.org/T13014#264919 I think that was one of the wikis I was trying on but a ran a bunch of stuff I have no idea what actually fixed it lol [07:48:21] I didn't appear fixed immediately so maybe was delayed lol [07:49:57] I ran a lot of stuff with eval trying to force an update also but no idea what fixed it or if it was even something I did lol [17:48:45] Chat i have a question [17:50:10] What [17:53:47] [1/5] Basically, in my rewrite of AWB to be cross platform, i need to figure out how to make the profiles save to the machine. AWB does this by saving them to the windows registry. [17:53:48] [2/5] That is an option for mac and linux, however, on booting up, in order to get all of the profiles, some funky stuff has to be done to search the keychain (which is also where other passwords are stored etc) to retrieve all of the saved profiles. I think this is potentially abit erronous and I dont really wanna do it like that so im thinking the options are either [17:53:48] [3/5] - do like above, just save all the profiles to the keychain/registry etc and search the registry for all the profiles at boot [17:53:48] [4/5] - have the app create a sqlite database in the directory that the app was downloaded to and save all of the accounts there. Encrypt the password with some key that is saved to the keychain etc, this way, we ever have to search the keychain etc for all the profiles, we can just simply retrieve the encryption key to decrypt the password before its sent to the api for login [17:53:49] [5/5] Which do you think is better/which would you be more comfortable with? [18:03:07] @originalauthority the 1st one is built in to the OS and it's Apple so it's technically preferable to me [18:03:25] But I don't really see an issue with 2 [18:03:32] Assuming you don't like use md5 [18:06:05] Hmmm, I just thought people would think its a bit weird if the app was searching through their keychain (granted, it would only be searching through the "name" field for "AWB" so I guess its technically not an issue) [18:08:56] I think it's fine [18:09:18] I'll have to have a look anyway to see what the performance difference is [18:10:11] Theoretically should be quick since I'm interoping with the native binaries to access keychain [19:00:07] If the issue seems solved, would it be possible to refresh links on other wikis soon? [19:10:30] If we haven’t already we’re probably gonna get them running globally soon [19:10:35] I’m not familiar with this issue [19:20:47] A global refresh probably not as that’d take over a year but definitely we’d keep honoring its run if requested on affected wikis [19:21:08] Wait seriously [19:21:31] Took over a week to run on allthetropeswiki [19:24:43] Jesus [19:25:12] I mean, can we run on a subset of wikis that won’t take too long, or go up the list small to big? [19:25:19] Or does it automatically run at some point [19:31:12] why is this a thing, again? [19:31:22] the need to manually refresh [19:37:12] Mediawiki fucking sucks [19:37:20] More specific I don’t recall [19:37:43] I don’t know if we know actually [19:38:06] but we didn't have that problem before [19:38:20] This is the page links special pages not working big [19:39:20] pre miser mode they would update in real time, after it felt like a day fir update, like on fandom [19:39:59] I think we’ve always been in miser mode [19:40:04] Certainly before this broke [19:40:14] We had a cron job that run it I think? [19:40:19] @agentisai understands better [19:40:33] when I joined in 2021 miser mode wasn't there [19:43:51] Okay maybe I’m just confused af [19:44:02] I think you’re conflating two concepts. The MediaWiki links table has always existed and sucked. The refreshing of certain special pages which would query the links table is completely different and independent of the table itself. [19:44:33] so miser mode isn’t the cause of the links table failing to update [19:44:40] all miser mode does is cache the latest result [19:45:00] thanks for explaining [19:46:16] So the table is what broke, and the updating is done by the cron job? [19:47:13] something was causing the table to not be updated [19:47:30] the table was updated with new entries every time a page is updated [19:47:43] not sure if it also updated when you purged the cache [19:48:34] Links will only be updated if there's a change on a reparse [19:49:12] The links tables are all being normalised slowly [19:49:27] I wouldn’t be surprised if it was an extension conflict [19:49:29] I wonder if any of them migrations have broke something tbh [19:49:29] So unless you edit every link that don’t update or have the script run the table is just permanently out of sync? [19:49:37] Possibly [19:49:50] refreshLinks should work [19:50:08] Could an extension not support the new links schemas and be doing something weird [19:50:24] DPL3, for example, didn’t work until a few days ago [19:50:38] so it could be [19:50:47] https://discord.com/channels/407504499280707585/1006789349498699827/1338228119714271312 [19:51:18] normally, the table should never fall out of sync [19:51:25] but it does happen [19:51:34] Is there anything in common with the wikis that have reported problems [19:51:39] so it’s likely most wikis won’t ever need refreshLinks ran [19:51:53] Does refreshLinks on them fix it and does it give it any error? [19:52:57] it wouldn’t change a thing before but also wouldn’t spew any errors [19:53:14] refreshLinks not fixing it doesn't make sense [19:53:29] Because that rebuilds the entire bloody tables [19:54:02] @agentisai is the actual schema correct? [19:54:10] It is [19:54:31] @agentisai did any READ / WRITE variables change around when it started [19:54:44] I wonder if someone added the variables but the code ain't working [19:54:51] cause that's happened before [19:55:02] The config has been there but the code to support it hasn't [19:55:55] https://www.mediawiki.org/wiki/Manual:$wgPageLinksSchemaMigrationStage says the default is SCHEMA_COMPAT_NEW in 1.43 [19:55:59] So it should be fine [19:56:52] Do we have it set [19:58:14] Set to new [20:00:38] The timing matches up with https://github.com/miraheze/mw-config/commit/75d8c04dd071a275e6d40477d50ec164fc0349ef [20:00:49] Which makes sense because it wasn't the default until 1.43 [20:01:05] But I have no idea why running refreshLinks in 1.43 wouldn't fix it @agentisai [20:01:47] @agentisai thoughts? [22:00:55] https://cdn.discordapp.com/attachments/1006789349498699827/1338268418259943537/image.png?ex=67aa7716&is=67a92596&hm=31906d2ff420ba4653f777d4c813ee89058412636b40208cf02b0d926ebd66c1& [22:01:29] Cloudflare was checking whether I was a robot, then this screen before it went to the actual wiki page 😮 [22:04:31] meaningful if it is constant, otherwise probably a blip [22:09:54] Hmm [22:10:22] Probably something took a bit longer than expected [22:10:46] Thanks for the report though @raidarr [22:10:50] @rodejong [22:11:04] @rodejong do you often get asked if you're a robot? [22:25:21] Never. Just twice today [22:25:38] Slightly weird [22:25:52] If it becomes a problem, dm your ip and I'll take a look [22:26:10] It was all automatically, it just redirected from page to page [22:26:17] Ye it should [22:26:38] But if it gets super grumpy then it can affect stuff more [22:26:51] It depends why you're being challenged [22:27:10] Most challenges should only happen every few hours at most [22:27:21] Unless something bad is going on [22:27:47] hmmm. Well as I said, Today was the first time. [22:27:56] So I'll keep an eye on it [23:29:24] wtf is https://github.com/wikimedia/mediawiki-skins-Schulenburg