[08:26:51] ohmy .. I managed to mess up a wiki with Special:ReplaceText, I'll revert it now to the db backup that I took and verified before I started this. Say .. is it within the regex power there to search for text with either capital or small 'l' in the start with [L|l] and capture that and print out with $1, coz I messed it up when I tried my first theory [08:27:12] mi regex a little rusty [08:28:46] no probs if cannot be done, wiki got like 16 articles so I can decapitalize by hand in under 5 minutes, by searching for any upper-case that should not exist [08:37:26] Iamthehuman12: yes, it does seem to take into account [A-Z], but as to what exactly it will replace, it is always tough to say as the result highlighting is not very helpful and there is no dry run functionality [08:37:34] https://www.mediawiki.org/wiki/Extension:Replace_Text#Using_regular_expressions [08:37:48] due to how limited it is, I've stopped trying to use regexes for doing actual replacements [08:38:16] I filed a feature request for a preview https://phabricator.wikimedia.org/T293121 [08:39:38] [L|l] will match correctly when I tick the "use regex"-tickbox, but when I tried just a plain $1, I had a lot of $1 in my wiki, now reverted from verified db backup and running the "too lazy to automate, so I'll just manually fix my lazy solutions glitches and bugs by hand" :D :D [08:41:11] thanks buovjaga [08:43:13] Aklapper's response reminds me that someone should pour money and finger miles into making a service that will do 3 things: #1 Tell's whose bug what likely is and point to the right place to report #2 Ability to easily send small amounts of money to any FLOSS developer #3 ability to attach tiny messages to the tiny money orders [08:43:59] with this service FOSS would really start to rock. ok, scuzes mon bruit sur le chan [08:49:08] there is probably a way to alter the default ReplaceText temporary throttling, since I have 16 articles, I would rather wait 30 seconds than 8 minutes ;) [08:49:57] temporal -> temporary, nah the island anglos broke the rule again. They always have no logic in how they rip Latin [08:51:11] nah, I just messed up in English, but being totally random seeming in the borrows from Latin, the island anglos are [08:59:42] but tiny money will only get you a tiny amount of work :) [09:02:08] languages do not care about rules, they care about convenience and new additions fitting into the language model [09:03:23] Iamthehuman12: a language that can adapt a borrowed word to its logic is a living language; a language that cannot is doomed [09:05:03] Remilia: there are two main schools of langauge developement: #1 develop compatible language (e.g. with neighbours or world powers) and #2 develop incompatible language [09:05:25] linguistics has no concept of ‘compatible’ [09:06:21] I'm not talking liguistics, I'm talking practice [09:06:51] I think you are talking simplified tech-inspired bs [09:07:32] ahaha, you really think this is "tech-inspired" and bs? You must be new [09:08:25] the "incompability" I was talking about, I was referring to so-called "false friends" in vocabularies [09:08:29] I have about 24 years of study of linguistics and language evolution under my belt by now so I am not sure [09:09:30] but this is off topic in any case, my point is there is no need to put down other languages for not following conventions you prefer [09:09:46] Remilia: Ok, you're a pro and I'm just me. I do not even know the names of all the sciences and scientists who work on language, so I could learn from you [09:10:34] When French rips from Latin, it does it with logic and coherence, anglos don't [09:12:41] French is a Romance language, English is a Germanic, your point is completely moot [09:20:09] OK, back to ontopic [09:20:47] Replace text messed my wiki up in a weird way. Look at what it did to my sitenotice https://stop-synthetic-filth.org/w/index.php?title=MediaWiki:Sitenotice&diff=4207&oldid=2977 [09:21:12] I'm going to see if any of the other contain the right contents eitehr [09:24:04] something algorithmic went very wrong, it swapped the contents of at least two articles when it was just supposed to do a replacement of "[L|l]aws against synthetic filth" to "Laws against synthesis and other related crimes" [09:24:30] this is probably something I messed up, not MediaWiki [09:25:40] I did do a 'sudo apachectl graceful && sudo service memcached restart', but I think I may have missed dropping the database before reloading it from a mysqldump backup. Perhaps I should do this again and this time empty the DB, before filling it again [09:27:13] yes, I am going to assume, not doing a 'drop database' was the culprit [09:27:55] or does someone want to help giagnose this one. it swapped the contents of one article with the sitenotice and vice versa [09:28:04] s/gia/dia/ [09:29:04] no, let me just run the steps again, this time dropping the old db first and that way no version numbers mismatch in the DB will happen, coz this looks like that [09:30:45] database backups on mysql usually contains a drop table before creating each table and filling it with contents [09:33:33] ok, now I'm confused, the old db is still in the wiki, even if I just dropped and recreated it from backup and restarted apache and memcached. Wrong order of restarting? [09:35:18] 'sudo service mariadb restart && sudo apachectl graceful && sudo service memcached restart' no help [09:36:11] MediaWiki may be pointing to a different database (or server) from where you're restoring the database [09:36:23] only localhost, but lemme doublecheck [09:37:08] You can just drop the database and reload the wiki without restoring the database. That should give an error [09:37:56] Vulpix: Yes, the site says it is experiencing technical difficulties [09:38:04] no db connexx [09:38:18] anything I should before loading the .sql ? [09:39:02] Then maybe the backup already contains the "errors", which means the fuck-up was not done with your last use of ReplaceText [09:39:46] Vulpix: I think the only logical explanation must be that I accidentally ran the old dump command when I should not have done so [09:40:01] I'll check the timestamp [09:40:03] recent changes, and page histories, may give you more ligtht about when the changes happened [09:40:52] nope, the file is as it should be. And I used it already once today, but after the latter go at the ReplaceText the wiki just went into incoherence [09:44:46] I spent 4 days trying to revert this kind of "ghost numbers in the DB" back in the day when I attempted to enable the VE, but the rust.php routing problem in the example httpd configs was not yet spotted [09:47:30] I do not know why "MediaWiki:Sitenotice" suddenly contains the full text of article "Organizations and events against synthetic human-like fakes". I reverted already once from this backup and nothing like this happened that time around [09:49:52] * Iamthehuman12 wanders off to wonder and brew some fairtrade organic arabica [09:53:26] should I drop database and reboot the server and see if the ghosts are gone. because restarting MariaDB, Apache2 and memcached does not remove the old ghosts in the db [09:57:20] I'll stop the job runner that is running some Java thing for the Extension:Translate to work, drop and reload database and restart MariaDB, Apache2 and memcached, maybe that Java-thing is holding some old data [09:58:49] Restoring a database without properly clearing MediaWiki cache can have nasty consequences apparently [09:58:54] Ok, that wasn't the culprit either, because it wasn't running, the 'sudo -u www-data php maintenance/runJobs.php --wait &' had died [09:59:25] Vulpix: I only know to restart memcached, what other cache-clearing would you suggest? [10:00:05] If you're using memcached, that should do it [10:00:14] it does not do it in this case [10:00:39] This usually means the problem is already in the database (backup) [10:00:42] I'm going to drop the db and restart the VPS and load the stuff in, cross fingers and hit load [10:03:15] Vulpix: I restored once from this backup without any problem after a failed ReplaceText today, but after running a ReplaceText of replacing "[L|l]aws against synthetic filth" with "Laws against synthesis and related crimes" I have a bunch of redlinks to "Aws against synthesis and related crimes" and article contents in other articles without any markings in their history about it happening [10:04:27] this situation is when the db should be same old one from 4 days ago, that I already restored once without any problems [10:05:48] I'ma drop db, restart server and load db and see if everything is back to normal or not [10:18:25] yes, success, ghosts are gone, wiki working fine. thanks for the MediaWiki [10:19:14] but I am afraid it will break in the same way if I run the same ReplaceText, should I? [10:22:03] I will run the ReplaceText again. If it works ok, this was probably just not restarting memcached at every right point in time ... like anytime anything changes in a way the program does not anticipate or expect [10:26:43] MediaWiki stores on memcached internal data like revision IDs or content IDs (internal ids where the actual text is stored). Restoring from db without clearing the "object cache" can wreck havoc if the internal cache has "the last revision/text of page X is on the id Z" from before restoring the backup, some edits are done, and now id Z points to a totally unrelated page/content [10:27:44] As long as you restart memcached right after restoring the backup, everything should be fine [10:29:54] thanks Vulpix ... now the replace seems to be going ok ... my mistake was probably not restarting memcached immediately after the db restore [10:31:02] Vulpix: I tried searching for it... you wouldn't know where is the variable for the temporal throttling of the ReplaceText, I would like it to run with less sleep in between running [10:32:17] I don't know much about ReplaceText. Maybe it waits until all jobs are processed? If that's the case, try running runJobs.php to process the job queue [10:33:08] That would make sense, since it would be pointless to do multiple replacements without waiting to the replace jobs, since they're not guaranteed to run in order [10:34:47] ok, will do [17:41:17] hi all, how can I edit another user's account (I'm admin) [17:41:44] specifically, I setup an user account but I need to update their email address / reset the password [17:47:48] digimer: You have https://www.mediawiki.org/wiki/Manual:ResetUserEmail.php for (re)set email (and password), and https://www.mediawiki.org/wiki/Manual:ChangePassword.php to only reset password [17:49:47] thanks! [17:50:41] I forgot to mention this requires shell access to run those maintenance scripts. I don't know if you have shell access [17:51:47] hello any tips on how to harden MediaWiki. Plan on exposing it to the internet. [17:56:44] Peppi: make sure things like config aren't accessible [17:56:58] It's designed to be usable by anyone publicly [17:57:48] it's gonna be spammed though unless you have some AntiSpam stuff / captcha active [17:58:12] so make sure to pick one of the captcha options [17:58:51] I'm wanting to use it for help documentation so don't want randoms commenting. But ok with the users doing it. [17:59:26] I'm using siteground as a host and they have an automatic install thing, so I hope it's secure that way, but need to set it up properly not sure how. [17:59:59] I think you can do "$wgGroupPermissions['*']['createaccount'] = false;" to disable self-registering. You would then have to create users for them. [18:00:08] mutante, is there a way to make it so only registered users can post and is there a way that I can confirm registered users? [18:00:30] mutante, ahh ok [18:00:46] wouldn't mind if they can create an account and then I'll give them permission to post [18:00:55] Peppi: by coincidence this is an article by your hoster: https://www.siteground.com/kb/how_can_i_disable_users_registration_on_my_mediawiki_website/ [18:03:39] I see there is https://www.mediawiki.org/wiki/Extension:ConfirmAccount [18:03:44] mutante, yes but I want them to be able to create accounts themselves, I just want to be able to manually give them permission to post. [18:03:58] mutante, ahh ok I see your second link [18:04:01] Peppi: seems like for that you need an extension like that [18:04:04] I'll take a look [21:04:48] Peppi: take away edit permissions from the user group and create a second group that does give permissions [21:23:21] p858snake, ok thanks