[00:25:36] LuaKT: you should `ssh -vvv bastion.wmflabs.org echo foo` and pastebin the output [00:25:49] also, what's in your .ssh/config ? [00:28:08] http://pastebin.com/NVtXQKtK heres the output from running that in cygwin [00:28:15] I don't have a .ssh/config fil [00:36:10] (03PS1) 10MarkTraceur: Move config into a default file and WMF files [labs/tools/grrrit] - 10https://gerrit.wikimedia.org/r/98141 [00:36:35] Notably, I'll be fucking with grrrit-wm a little bit in the next hour or so [00:36:54] But not to worry, it's all for the greater good [00:38:52] marktraceur: feel free to self merge and stuff :) but only after testing! (in production!) [00:39:07] i mean, it's okay if the production instance fucks up, but git history! [00:40:50] Yup [00:41:58] LuaKT: 10:1 the problem is case significance in your shell account name. (Hint: it's case significant, and probably all lowercase) [00:43:16] * YuviPanda waves at Coren [00:43:28] not going to make you do anything :D Just a status update [00:43:47] Coren: Yeah I checked that earlier it's luakt lower case, but I get the "Permission Denied" before asked for a username when using jeremyb's command above [00:44:54] ... ssh doesn't ask for a username; it uses the username you have on your local box unless you specifiy one. [00:45:13] i.e.: ssh luakt@tools-login.wmflabs.org [00:45:42] I rewrite portgranter to work with redis and also use identd for auth [00:45:42] so that only things calling with a tool's service account submit routes for things under that tool's name [00:45:42] now documenting it, will submit a patch in a bit [00:45:42] Coren: I hope to get the entire thing done (and tested!) before Monday, so hopefully you can set aside a day next week to do the migration :) [00:45:59] Still permission denied [00:46:34] http://pastebin.com/4nGJRmcw [00:49:21] LuaKT: Your key isn't valid. How did you generate it? [00:49:38] In fact, the file that should contain your ssh key doesn't contain an ssh key at all. [00:50:03] It looks like the beginning of... a PEM certificate of some sort? [00:50:12] Or a PGP key. [00:51:10] Coren: ssh-keygen -t rsa [00:51:33] -----BEGIN RSA PRIVATE KEY----- first line of id_rsa [00:52:16] Yes, but it's not recognized as such, so I presumed it was something else that started with ----BEGIN :-) [00:52:35] debug3: Not a RSA1 key file /root/.ssh/id_rsa. [00:52:35] debug2: key_type_from_name: unknown key type '-----BEGIN' [00:53:09] Pro tip: don't ever use root to ssh somewhere else. For that matter, don't use root. :-) [00:53:32] Yeah it's a vm I was messing with [00:54:15] Try generating (and using) a DSA key; perhaps you have an odd bug. It should work, but at least this will help isolate the problem. [00:55:07] Coren: see http://toodadoo.wmflabs.org/lolrrit-wm/r/ - it is proxying dynamically to a simple python test server I'm running (temperoarily!) on tools-login [00:55:39] YuviPanda: Not gonna do this tonight, buddy. :-) [00:55:44] Coren: oh no [00:55:48] Coren: I'm not making you do it :D [00:55:50] Coren: just telling you [00:56:04] Coren: BECAUSE IT HAS BEEN A GHOST TOWN AND I HAVE NOT FOUND ANYBODY TO TELL IT TO! [00:56:06] :P [00:56:19] i'm just excited. nevermind the caps. [00:57:22] Coren: http://pastebin.com/6k6sWate Thanks for your time by the way [00:59:42] LuaKT: Well at least your client now recognizes your key, but you haven't added it to wikitech. :-) [01:00:10] Coren: I have added it, I presume it takes some time to update [01:00:48] It does, but it's normally fairly short. Let's wait a couple of minutes. [01:01:37] Coren: I'm off to sleep. Hopefully we can get this migrated next week :) [01:02:32] Still not seeing it. You see it if you look on your preference page on wikitech? Because right now I'm only seeing an RSA key. [01:03:57] Coren: I have 6 keys on there, you're only seeing 1 [01:04:00] ?* [01:04:20] Yep. [01:04:40] Possibly, the second one is broken in such a way that the copying mechanism stops processing them. Odd. [01:04:48] Try to replace them all with the new DSA key? [01:05:54] Coren: Ok done [01:06:18] Not seeing a change yet; let's wait a few minutes. [01:07:02] What's your Wikitech username? [01:07:43] LuaKT [01:09:13] What I find "Interesting" is that, as far as I can tell, your keys have not changed since Nov 25 [01:11:24] Coren: Thats the first key I added before getting approved. Going to try this 'Restore all default settings' [01:11:44] Oh, didn't reset keys [01:11:47] I don't think that affects keys. [01:11:52] Yeah, that. [01:12:28] andrewbogott_afk: Ping for when you're around; it looks like wikitech is no longer dumping key changes to /public -- or at least not for LuaKT [01:12:53] Coren: Thanks [01:12:56] I'm going to look into what's going on there in the meantime. That really should be automated. [01:16:32] Actually, I now see that it's been ~2 days since any key has been touched; that seems unlikely. [01:20:37] Coren: fyi, almost certain the validity of RSA key thing is red herring [01:21:10] jeremyb: Regardless of the kind of fish, the issue with the keys not updating is certainly a problem. :-) [01:21:21] yes, certainly [01:26:17] LuaKT: key="$HOME/.ssh/id_rsa"; ssh-keygen -lf "$key.pub"; ssh-keygen -yf "$key" > "$key.newpub"; ssh-keygen -lf "$key.newpub" [01:27:36] re red herring: it's just misleading log msgs because it first tries to load it as RSA1 and when that fails tells you a novel about why and then tries RSA2. [01:28:05] but something's also wrong because of """debug2: key: /root/.ssh/id_rsa ((nil))""" [01:31:52] jeremyb: I generated a new RSA key then ran that command http://pastebin.com/arQmdqgX [01:32:04] jeremyb: debug2: key: /root/.ssh/id_rsa (0x7fd938b2f4e0) instead of nil [01:35:50] LuaKT: so we lost the old key i guess [01:35:59] LuaKT: new one starts with approximately the same first line? [01:37:53] jeremyb: both the old and the new start with -----BEGIN RSA PRIVATE KEY----- then Proc-Type: 4,ENCRYPTED on second line [01:38:02] I'll run that command on the old one [01:39:14] LuaKT: if you can move old one back into place then maybe we can get you logged in before the other problem is fixed :-] [01:40:06] Coren: if you want mysql pointers (re replication) i have a few queries you may want to run. but leaving for sean is fine too [01:40:21] jeremyb: Ah, the "old" one isn't the one I added on the 25th that Coren can see, Sorry for the confusion [01:40:39] jeremyb: The one from the 25th I don't have anymore [01:40:58] c9:78:33:b2:bd:33:a1:c6:be:d0:f2:13:ce:7b:46:da [01:41:08] that's the one i see in the keys dir [01:42:14] jeremyb: I see why replication has stopped, but the problem is that the replication system is fairly new and not documented so without Sean's intervention it'd be rough to restart it. [01:43:05] jeremyb: Yeah that doesn't match the "old" one [01:43:10] Coren: yeah, restarting is a different question. i just meant to get "what is the latest error msg" [01:43:31] LuaKT: and in ldap: [01:43:31] 1024 2f:7c:a8:b2:fa:d5:3e:27:cb:3b:ba:a2:48:80:c1:1b root@luakt (DSA) [01:43:31] 2048 02:08:65:cc:7f:6e:47:d4:73:32:09:05:4e:c5:de:c3 root@luakt (RSA) [01:43:48] so that's 3 different keys [01:44:21] * jeremyb waits to see what coren cooks up with key syncing [01:44:31] jeremyb: Binlog replay failing because it's trying to delete a row that's not there in externallinks [01:44:44] jeremyb: yep I have those 2 keys [01:45:25] Coren: lovely. sean was adding a primary key to that table. idk what the progress is (or whether s5 is done yet or not) [01:45:49] Coren: (that primary key will help with syncing it up if the table gets somehow out of sync with master) [01:46:11] in fact it's about time for me to inquire about the status of that change again :) [01:46:14] jeremyb: It will. I'm thinking that may be the problem; el_id might be in prod but not in the sanitarium yet. [01:46:26] hrmmmm, interesting [01:46:49] So syncing of externallinks failed; thus replaying binlogs against it goes boom. [01:47:14] Fixing this without hosing the rest of the shard is beyond my confidence level. [01:48:02] yeah, some future update in the log could rely on externallinks being accurate [01:48:07] during replay [01:48:21] so you can't just skip it [04:07:09] Coren: Can you help with my lighttpd config? [05:30:21] * a930913 pets Ryan_Lane. [05:44:33] a930913: tried /lastlog but i can't quite find what you're looking for. what exactly are you trying to do? how do you want lighttpd to behave? [05:44:46] you should state that before you start poking people :) [06:19:32] good morning [06:19:43] the replication of Wikidata is still broken ... day 3 [08:24:06] I need help [08:24:06] Hi Ahmad_Sammour, just ask! There is no need to ask if you can ask [08:24:53] i want to put a limit to number of interwiki in SQL code. [08:25:52] e.g. x article has 5 or more intrwiki [08:30:46] Ahmad_Sammour: Use the limit clause [08:31:01] e.g. SELECT something FROM table WHERE condition=1 LIMIT 5; [08:31:41] limit of interwikis ? [08:32:01] * SigmaWP shrugs, I don't know what your query is [08:32:50] "limit" will make it so that you will receive at most 5 results from the query [08:33:29] ('SELECT CONCAT(":",page_title,"]]||", count(*)) FROM langlinks JOIN categorylinks ON ll_from = cl_from AND cl_to="'+PageTitle+'"JOIN page on ll_from = page_id WHERE page_namespace = 0 AND page_is_redirect = 0 AND NOT EXISTS (SELECT * FROM langlinks as t WHERE t.ll_lang="ar" and t.ll_from = langlinks.ll_from) GROUP BY ll_from ORDER BY count(*) DESC,page_title;') [08:33:47] this is the query. [08:34:29] I'm not good enough to know what that means [08:34:29] Sorry :( [08:35:05] It's OK. [12:01:04] jeremyb: Trying to add an access control header to the page. [14:10:40] Coren: dewiki_p.externallinks still seems to be missing some million entries (used to be around 10 million) [14:13:29] and rc is still 2 days behind and not increasing, if i see correctly. [14:36:21] giftpflanze: Slave status shows that the replica is caught up to to the santiarium. [14:38:06] Coren: pardon? [14:39:34] giftpflanze: Mostly thinking aloud. I'm trying to see what may be up. [14:44:16] Coren, what special page on wikitech will allow me to create a tools project. [14:44:47] Cyberpower678: That's though through the 'manage project' interface. [14:44:57] There is a 'add service group' link there. [14:45:25] Coren, can you provide a link. You guys seem to be moving it around. :/ [14:45:42] Cyberpower678: It has never moved since Labs existed. [14:46:32] It used to be located in Special:NovaProject. [14:46:41] Now it's not. [14:47:18] Coren: got anytime for some ssh keys debugging on labs ? :D [14:47:55] Ah, yes, the /destination/ did change. https://wikitech.wikimedia.org/wiki/Special:NovaServiceGroup [14:48:23] hashar: kinda-sorta. What's up? [14:48:58] the context is setting up Jenkins slaves in labs hosted on the same machine as the jenkins master [14:49:22] to do that, I created a jenkins-slave user in the 'integration' labs project. Generated ssh key pair in that user home directory and added the public key in wikitech [14:49:25] Coren, I seem to be the goto person for tool takeovers. :p [14:49:36] ldaplist -l passwd jenkins-slave : confirms the public key is there [14:50:00] but, when I connect locallhost ( sudo su - jenkins-slave … then … ssh localhost I get rejected) [14:50:39] Coren, how long does it take for the service group to be ready for use? [14:52:06] Coren: long story short, I suspect PAM prevents the authentication :/ /etc/security/access.conf is non obvious though [14:52:57] Hm. Is the key actually there? [14:53:47] It's not. [14:54:02] That looks suspiciously like yesterday's bug with a user. [14:54:20] It looks like whatever moves keys from LDAP to /public/keys is no longer working. [14:54:40] * hashar looks there [14:55:01] ahh [14:55:10] /public/keys/jenkins-slave does not exist :] [14:55:56] I can't find any documentation on how that process works, nor where it takes place. [14:56:25] will look at Openstackmanager extension [14:57:00] I don't think that's written to from the extension; I'm pretty sure there's a cron job somewhere to do that. [14:57:12] The extension, IIRC, just puts stuff in LDAP. [14:58:59] Coren, I'm no linux expert. How do I create a folder redirect? [14:59:39] In apache, you mean? [15:00:51] I'm creating replacement tools involving the use of the cgi-bin, but it was located in the public_html folder. Labs' cgi-bin is located outside of the public_html. [15:02:03] ... just move the contents to the actual cgi-bin. No redirection needed. [15:02:33] Fine. :| [15:02:34] Cyberpower678, the ~/cgi-bin folder is handled correctly by the web server, see https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/Help#Web_services [15:03:47] ireas: It's not handled "incorrectly", it's handled as a CGI directory (which is the intent). If you want things that aren't CGI, then one shouldn't put them in cgi-bin. :_0 [15:04:29] Coren, that’s what I wanted to say ^^ the web server interprets ~/cgi-bin correctly as the cgi-bin directory and "mounts" it to //cgi-bin [15:04:51] Ah. Well, that's clearly correct. :-) [15:05:02] ;) [15:14:31] Coren: I tried out by putting the key somewhere under /etc/ssh/userkeys/ , not much luck [15:14:32] :( [15:14:45] * hashar feels unproductive [15:22:47] hashar: Yeah, I've been trying to find documentation on how the keys move from LDAP to the shared directory with no luck. Perhaps we'll be lucky and either andrewbogott_afk or Ryan_lane will stop by during their shopping. [15:26:16] !newweb [15:26:16] https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/Help/NewWeb [15:26:20] Cyberpower678: ^^ [15:27:22] Coren: will fill a bug and call it an end on my side :D [15:27:47] hashar: That also works. :-) [15:30:18] ahh [15:30:21] uid mismatch probably [15:30:49] jenkins-slave LDAP uid is 4098 [15:30:53] but id gives me 1001 [15:53:01] well opened https://bugzilla.wikimedia.org/show_bug.cgi?id=57751 [16:16:21] addshore, hihi [16:16:28] hihi [16:17:52] * YuviPanda waves at addshore [16:17:55] addshore, I'm working on making scottywong's tools available on labs. [16:18:02] I hit a hurdle. [16:18:09] which tools is that? :P [16:18:24] * addshore waves at YuviPanda :) (care to give me a minor puppet review? :)) [16:18:30] sure [16:18:32] link? [16:18:52] YuviPanda: https://gerrit.wikimedia.org/r/#/c/96552/ [16:19:12] dont need to +2 it or anything, just a sanity check as this is the first puppety thing i am writing :) [16:19:25] I can't +2, remember? :P [16:19:25] reading [16:19:35] hence I want to know if therre is anything I could do better or anything I have missed totally or might want to think about :) [16:19:41] empty init? [16:19:56] addshore, You've got a pm. [16:20:06] addshore: omg the entire thing uses tabs! [16:20:08] NOOOOOO! [16:20:09] :P [16:20:17] 4spaces plz [16:21:49] ahh yes, aude already told me that :P [16:21:55] Need to fix in a followup ;p [16:25:28] addshore: simple comments [16:25:41] [= [16:25:46] why 4 spaces? :P [16:25:48] why not 2? :P [16:26:04] addshore: consistency with python [16:26:04] :P [16:27:53] pah [16:27:55] thats silly :P [16:28:53] addshore: not sillier than tabs [16:29:09] a n d e x c e s s i v e s p a c i n g e v e r y w h e r e [16:30:29] :d [16:33:57] !log deployment-prep Added a ssh public key for jenkins-deploy user. Should only be accepted from 10.4.0.58 which is deployment-bastion.pmtpa.wmflabs [16:34:49] addshore: all new puppet code is 4spaces, btw. only ancient code has tabs [16:35:18] addshore: and I suggest putting a license header / copyright header on all files, mostly because, uh, puppet repo has... no licensing information :D [16:39:01] pash [16:39:06] ammending now ;p [16:42:09] taaaaaaaaaaaabs! [16:43:16] YuviPanda: question :P [16:43:23] addshore: sure? [16:43:33] * aude says no :) [16:43:38] require => [ Package['npm'], File['/home/wdbuilder']. Git::Clone[''clone_wikidatabuilder''] ], ? [16:43:47] yeah that works [16:43:48] does that work? :P [16:43:49] [= [16:43:57] although I think it is Git::clone [16:44:00] rather thatn Git::Clone [16:44:04] okay [= [16:44:08] that... has always been a little confusing to me :D [16:44:29] addshore: it is Git::Clolne [16:44:31] Clone [16:44:32] with a capital C [16:44:36] not small [16:44:51] addshore: put them in separate lines tho :) [16:44:59] heh, okay :P [16:46:04] YuviPanda: my ide complains about the lowercase c in clone :P [16:46:13] what IDE? [16:46:20] PHPStorm's puppet support? [16:46:23] phpstorm plugin :/ [16:46:25] xD [16:46:55] :C [16:46:56] err [16:46:56] :D [16:47:00] /me is sticking to Emacs for now [16:47:03] :P [16:48:20] /me considers sleeping early today [16:48:35] /me gives YuviPanda coffeee [16:48:36] YuviPanda: which editor are you using under emacs? [16:48:43] hashar: vim :D [16:48:48] +2 [16:49:15] hashar: I've Emacs with Evil mode, so I can get access to Emacs modes and stuff but don't have to suffer its stupid keybindings :D [16:49:35] hashar: also modal editing FTW! Emacs can pry my text objects from my cold... dead... hands [16:51:16] YuviPanda: ty, addressed your comments [= [16:51:20] addshore: :D [16:51:46] hashar: can git::userconfig be used to push to repos? ;p [16:51:52] I see your name in the example ;p [16:52:12] just a wrapper to generate a ~/.gitconfig file [16:52:29] get a puppet hash of section name => key => value [16:52:31] and expand it as: [16:52:35] ahh okay :) Still cleaner than what I am using ;p [16:52:39] [section name] [16:52:39] key = value [16:53:02] addshore, MariaDB [enwiki_p]> SELECT * FROM namespaces; [16:53:02] ERROR 1356 (HY000): View 'enwiki_p.namespaces' references invalid table(s) or column(s) or function(s) or definer/invoker of view lack rights to use them [16:53:24] YuviPanda: does your new labs proxy support redirecting sub directories to different instances ? [16:53:51] hashar: you can either have subdirectory redirect or domain redirect. not both... yet. [16:53:53] YuviPanda: I could use domain.labs to be put routed to an instance but domain.labs/directory/ to go to another ip:port [16:54:13] hashar: that feature isn't available on the proxy deployed from wikitech yet :( [16:54:27] hashar: but I built it for toollabs last week, so I can add it to the general proxy in a week or two [16:54:32] well I could do a subdirectory redirect of / :-D [16:54:43] YuviPanda, MariaDB [enwiki_p]> SELECT * FROM namespaces; [16:54:43] ERROR 1356 (HY000): View 'enwiki_p.namespaces' references invalid table(s) or column(s) or function(s) or definer/invoker of view lack rights to use them [16:54:45] you can have otherdomain.wmflabs.org :P [16:54:49] go to a different one [16:54:54] but not ideal [16:55:25] hashar: but this feature (URL based routing) should turn up sometime soon. On my radar for suer [16:55:26] *sure [16:55:31] nicee :-] [16:55:42] take your time. [16:55:54] hashar: my current labs focus is to move toollabs to the dynamic proxy :) [16:56:10] so we can finally get rid of goddamn CGI (or even fastcgi) [16:56:20] Helloooo? [16:56:22] and have websocket access, and also write webapps in whatever language we want [16:56:31] * YuviPanda flails around like an excited kid [16:56:36] and that feature will most probably let us get rid of most public IP assigned to labs [16:56:45] hashar: yup! [16:57:00] hashar: I expect a lot of the public IPs to be taken back before the eqiad migration [16:57:13] will also make the migration easier [16:57:19] petan, ping [16:57:25] * addshore has already given 1 back ;p [16:57:43] nice :D [16:57:51] Cyberpower678: pong [16:57:59] petan, MariaDB [enwiki_p]> SELECT * FROM namespaces; [16:57:59] ERROR 1356 (HY000): View 'enwiki_p.namespaces' references invalid table(s) or column(s) or function(s) or definer/invoker of view lack rights to use them [16:58:03] Why is that? [16:58:09] I have no idea [16:58:13] * aude hoards my ips [16:58:24] well, actually I do know what does it mean, but I don't know why it happens [16:58:38] petan, what does it mean? [16:58:38] either the view is invalid (the schema changed) or permissions changed? [16:58:57] * YuviPanda passes IP control laws and takes away aude's IPs [16:59:08] it means that the definition of view references some columns or tables that either don't exist or you can't read them [16:59:21] * addshore sets up a man in the middle attack and steals the ips that were going from aude to YuviPanda :O [16:59:23] Well, can that be fixed? [16:59:25] :) [16:59:30] like create view blah as select * from bleh; [16:59:45] if you didn't have access to bleh you would see this when you did select * from blah; [17:00:05] Cyberpower678: yes it can be fixed [17:00:19] petan, can you fix it? [17:00:21] no [17:00:28] I have no powers [17:00:33] Because Coren|AFK is away. [17:00:42] you need Coren or anyone with dba permissions [17:01:15] Cyberpower678: When was the last time you did a successful query on this table? [17:01:33] Never. [17:02:00] Cyberpower678: Ok Then, Never saw it before either [17:03:37] Cyberpower678: Doesn't exist :P Ghost table. [17:04:05] hedonil, oh yes it does. [17:04:16] show tables; will tell me that [17:04:44] Cyberpower678: It's no public view nowhere I've been so far [17:04:59] namespaces would be specific to labs [17:05:01] Cyberpower678: i think the namespaces tbale is labs specific, I dont see it in the mediawiki schema [17:05:04] e.g. toolserver had something like it [17:05:07] * aude thinks [17:05:09] Cyberpower678: file a bug or poke Coren|AFK ;p [17:05:16] has info about namespaces on all the wikis? [17:05:22] * aude clueless [17:05:58] aude may be correct. Toolerver does and I'm trying to access it as I migrate a tool from toolserver. [17:06:06] Cyberpower678: it doesnt appear to exist on all wiki dbs on labs [17:06:14] ie. i find it on enwiki but not on ptwiki or wikidata [17:06:15] sounds vaguely familiar [17:06:39] huh, it's only on enwiki? [17:06:47] Queried that info some time ago for all wikis, think it was in l10n_cache [17:06:48] well, its not on ptwiki :P [17:06:55] not on metawiki [17:07:04] or eswiki :P [17:07:40] https://bugzilla.wikimedia.org/show_bug.cgi?id=49167 [17:07:49] wontfix? [17:08:03] from magnus Thanks, I'll use the API instead. [17:08:16] use https://en.wikipedia.org/wiki/Special:ApiSandbox#action=query&meta=siteinfo&format=json&siprop=namespaces%7Cnamespacealiases ;p [17:08:58] * YuviPanda considers setting up a way to use ansible on labs [17:09:17] addshore cheers :) [17:09:24] YuviPanda: yay [17:09:27] i like anisble :D [17:09:37] though im starting to come around to puppet a bit :P [17:09:40] addshore: I havne't fully checked it out yet, but... python :D [17:09:46] haha xD [17:09:52] addshore: what I really want to do is per-project easy-to-use puppet/ansible repositories [17:09:59] * aude likes puppet [17:10:00] addshore: so we don't put every goddamn thing into one repo :P [17:10:06] * YuviPanda likes puppet too [17:10:08] YuviPanda: see https://github.com/Orain/ansible-playbook [17:10:11] [= [17:10:22] addshore: indeed, have already checked it out :) [17:10:25] :D [17:10:26] * YuviPanda hangs out on #orain [17:11:09] * YuviPanda tries to not get distracted [17:11:13] proxy for toollabs first :D [17:13:02] Cyberpower678: If you just want some info try SELECT * FROM xxwiki_p.l10n_cache where lc_key='namespaceNames'; [17:13:56] Cyberpower678: If that's what you need. [17:15:18] Cyberpower678: blob in lc_value contains serialized array of namespaces [17:21:36] hedonil, python can't unserialize. [17:21:42] Sadly. [17:23:06] Cyberpower678: php can. [17:23:59] hedonil, I'm not in the mood to completely overhaul and rewrite Scottywong's tools [17:24:14] But yes, I like using PHP. I hate Python. [17:26:07] Cyberpower678: make a php script which takes a single param (serilized data) and returns in in some other format python can read? :P [17:26:12] then call it from the py script? xD [17:26:21] Cyberpower678 <3 [17:26:29] Cyberpower678: Well, It's pretty static. One time job - get it some lines of php-code, put it then on tools-db table - finally tell us, so we can use it too:P [17:26:35] I thought I am only person out here who doesn't like python [17:27:23] petan, no. I even went as far to rewrite SnotBot to PHP from Python. [17:27:34] cool [17:28:12] SnotBot is surely very happy [17:28:12] Python has left me unimpressed, and the "oh so awesome" Pywikipedia framework is total crap. Peachy does so much better. [17:28:24] xD [17:28:41] Or addshore's framework. [17:28:59] mwahahahaa [17:29:02] I don't really know what makes it a "framework" it looks to me like some bunch of complete bots that you can just launch yourself using your own bot account [17:29:11] * hedonil applauses to Cyberpower678's thoughts on python [17:29:22] :P [17:29:35] we seem to have a few people here that havnt really caught the python bug :P [17:29:37] me included :P [17:29:40] * YuviPanda uninstalls php from toollabs [17:29:54] * petan uninstalls Panda from Yuvi [17:30:04] To be honest, from what I experienced while I rewriting SnotBot, the new script is 25% shorter. [17:30:37] and 26084600% faster and 78% less memory expensive [17:30:52] and 99% easier to read [17:30:53] Pywikipedia has virtually 0 error handling. [17:31:05] python can handle errors? o.O [17:31:23] It doesn't re-attempt edits on a failed attempt. [17:31:45] It hit a 503 response once. [17:31:50] What happened? [17:31:55] Python output: [17:32:04] Server response: 503 [17:32:22] Uncaught exception on first half of program. Skipping. [17:32:32] it doesn't matter if exception description is useful in python. only what matters is if exception is cute [17:32:51] the indentation of produced error text matters. not the text. [17:33:29] it's funny how it can determine where the "half of program" is :D [17:33:45] I wondering how much cpu and memory that computation needs [17:34:14] Peachy handles the errors, re-attempts, and kills the bot after multiple failures, to prevent damage to the wiki. [17:34:41] It also doesn't explode when it receives a 503 response. [17:34:59] And it can simply do more [17:35:04] implementing a code that prevents damage to wiki could damage the look & feel of python code and thus make the source code less beautiful. you can't do this in python [17:35:31] petan, lol [17:35:59] * Cyberpower678 wonders how people like this programming language [17:37:16] petan, my philosophy, is better stable than beautiful. [17:37:27] Afterall, the end result is what counts. [17:37:31] that should be philosophy of every coder [17:37:33] :P [17:37:59] Well apparently, python programmers have that issue. [17:37:59] but who cares about cpu or memory these days, every computer has a lot of that [17:38:10] I do. [17:39:18] when I installed ubuntu on my mini-laptop (2gb of ram) just gnome 3 (written in java I think) with lot of python stuff all over (installed by default in ubuntu) were using more than 1.5gb of ram, I couldn't even launch internet browser [17:39:33] then I replaced it with lxde (written in c) and ram usage dropped from 1.5g to 82mb [17:39:33] I may have an i7, 8 core at 3.7 GHz, 16GB RAM, and 800GB SSD, but the less RAM and processor I use for one thing, the more I have for something else [17:41:30] aha I take it back, gnome 3 is partially written in c as well, but it's rather a huge interpreter of js [19:00:59] why is select rev_page from revision where rev_user_text='Liangent-bot' limit 1; so slow? [19:01:12] or is Special:Contributions using another query? [19:01:57] you know https://wikitech.wikimedia.org/wiki/Nova_Resource:Tools/Help#Tables_for_revision_or_logging_queries_involving_user_names_and_IDs ? [19:03:41] giftpflanze: nice thanks. I knew this and wrote it in my db wrapper (to always use revision_userindex), then forgot it. then now I'm writing a raw sql query [19:37:24]