[00:07:48] anomie: No; it's still the same proxy though we plan to have Yuvi's in place in eqiad. [00:08:32] When's the eqiad move happening? [00:27:56] anomie: you may have seen mhoover here. he's [AIUI] been contracted primarily to make the eqiad move happen [00:53:10] Cyberpower678: fyi, I mentioned your bots at https://en.wikipedia.org/wiki/User_talk:%CE%A3#Lowercase_sigmabot_removing_PC1_template_on_pages_that_still_have_it [00:54:54] SigmaWP, don't burden your bot with having to look for PC protected pages. [00:55:16] That causes problems, as we've just seen [00:55:20] The API doesn't even support retrieving such information. [00:55:43] https://en.wikipedia.org/w/api.php?action=query&prop=info%7Cflagged%7Crevisions&rvprop=flagged%7Cids&inprop=protection&titles=Qnet&format=jsonfm does [00:55:54] Did, at least, a while ago [00:56:21] https://en.wikipedia.org/w/api.php?action=query&prop=info|flagged|revisions&rvprop=flagged|ids&inprop=protection&titles=Adam%20Silver&format=jsonfm is a better link [00:57:10] SigmaWP, so you're going to scan everypage on Wikipedia looking for those flags? [00:57:30] I'm not sure where I said I planned on that [00:57:39] But no [00:58:05] The link you provided me only provides PC protection of 1 page. Don't you need a list of pages that are PC protected? [00:59:08] That's where https://en.wikipedia.org/w/api.php?action=query&list=logevents&letype=stable&leprop=title|timestamp&format=jsonfm comes in [01:00:35] So that tells you when they're protected. How about detecting when they are no longer protected? [01:03:01] Cyberpower678: You're asking me to essentially write a function out of my ass, for what purpose? [01:03:44] Because I think Cyberbot's approach is much simpler. [01:03:57] I'd prefer to leave Cyberbot on. [01:04:31] I also don't like sending something through approval to shut it down later. [01:04:53] Cyberpower678: I think it's clear that from the discussion I linked that by dividing the protection templates between us, we will either have redundancies in code or errors that need to be solved [01:05:55] There lies a contradiction in "remove all the protection templates, but don't remove the pending changes one because I'm leaving that to cyberbot" [01:06:12] SigmaWP, it is. Cyberbot is functioning perfectly, no need to change that, your bot however is removing the wrong templates in some case. [01:07:00] SigmaWP, simply scan for the presence of pp-pc1 and pp-pc2 and remove all but those. [01:07:10] By adding code to ensure that it does not remove the wrong templates, I effective add code that does exactly what Cyberbot does sans the editing [01:07:33] Not really. [01:07:52] Again, just tell SigmaBot to leave those 2 templates alone. [01:08:02] You don't even need to check for protection. [01:08:42] I added code to ensure it doesn't remove the wrong ones. [01:36:17] SigmaWP: not gonna get mixed up with this [01:37:21] Where did that come from [05:54:29] hello [05:54:55] May one use code from a Wikimedia wiki on Tool Labs? [05:55:17] Apparently, only OSI-approved licenses are allowed. [05:55:38] But CC BY-SA and GFDL are not OSI approved for code. [05:57:43] If not, do I have to rewrite it from scratch? Is that even allowed if I've seen the original source code? [06:01:17] I'm going to go now, but I'll check the log later for a reply. [06:04:03] hah, good point [06:04:09] that should be clarified ;) [06:04:32] (or wiki code should be dual-licensed under an OSI-approved license [06:04:35] +) [06:30:28] Sorry for the joining/quitting, but I need to go soon. [06:30:45] greg-g: is it currently allowed? [06:31:02] or will it be in the future? [10:38:19] i don't get how to demonize/configure redis properly or do i have nothing to do?? [11:13:46] hello I would like to run the server currently running on mediawiki2latex.mooo.com on labs I need at least 8Gbyte of ram can I do that, and if yes how. I already got an ssh account on tools-login.wmflabs.org [15:45:36] greg-g: hi [15:46:33] Is it allowed to run code under CC BY-SA on tool labs? It's only a few lines. If I rewrite it from scratch and license it under GPL< then can it be used? [15:47:05] it's just an SQL query [15:49:42] i doubt sql queries can be copyrighted [15:49:57] not much possibility for rewriting that [15:50:02] It also has a loop for each wiki [15:50:12] seems trivial [15:50:16] okay, thanks [15:50:21] but i'm no lawyer ;) [15:50:27] of course they can be copyrighted - it's a creative work. [15:50:41] but it's hard to prove someone did not come up with the same result on their own [15:51:14] Tool Labs requires ``OSI approved'' licenses. Of course, CC BY-SA is not OSI approved, but I believe it is in the spirit of free software/content. [15:51:24] Maybe I should ask the author to dual license it. [15:51:56] valhallasw: comes approximately to the same point [15:52:15] it is 5 lines of code [15:52:45] the gnu people told me once that trivial code does not need to be licensed [15:52:50] erm, more like 10. But the SQL query is in a string that takes up a few lines. [15:52:56] inside a bash script [15:54:47] Is trivial code actually creative and thus can't be cc-by-sa licensed? [15:55:30] i think that it is also public domain [15:57:28] giftpflanze: https://wiki.toolserver.org/view/MySQL_queries#List_of_interwiki_links_from_all_wikis this is where I got it from, but I had to make a few changes. Is it trivial? [15:58:44] I just don't want to break the Labs rules. [15:58:52] it's not creative imho but straightforward. i would do the same thing basically [15:59:08] maybe ask an labs admin [15:59:15] to make sure [15:59:22] What is the best way/place to install crons in the Tools project? Toolserver had a special host for it (submit.toolserver.org) [15:59:36] Jelte: what about jsub? [15:59:40] Jelte: do it on tools-login [16:00:21] ok, thanks [16:00:48] giftpflanze: http://bots.wmflabs.org/~wm-bot/logs/%23wikimedia-labs/20131222.txt greg-g (employed by WMF) already commented on it, but his response was ambiguous [16:01:00] oh, thx [16:01:19] what position is he? [16:02:18]

hah, good point that should be clarified ;) (or wiki code should be dual-licensed under an OSI-approved license) [16:04:06] that reminds me, i have to rescue data from tswiki [16:09:32] Coren: is it planned to transfer any parts of tswiki to wikitech? [16:10:12] giftpflanze: Not by us, but It's A Wiki™ and anyone is welcome to salvage any useful part of the tswiki and house it on wikitech. [16:10:27] ok thx [16:10:58] compatible licenses [16:17:39] * valhallasw checks the pywikipedia archive [16:19:56] nothing to move pages with history, unfortunately [16:20:41] no special:export? [16:21:37] Sure, but then you need someone to special:import it ;-) [16:21:58] that's the lesser problem [16:22:34] Coren, for JIRA move requests, I suggest to make wmf.bugconverter@gmail.com the default assignee during the conversion -- that slightly reduces bugspam for whoever will become the default assignee. [16:23:33] and make a summary e-mail for them [16:24:32] Hm? They can just check the bugs in the category on bugzilla, right? [16:28:09] i have no idea :) [16:45:22] PiRSquared: giftpflanze I have no say in this, just an interested person, fyi :) [16:45:53] greg-g: then who does? [16:46:02] Coren? [16:47:39] greg-g: ok :) [16:47:44] I'd send the mailing list an email, if I were you. [16:47:57] or you don't bother [16:48:21] if you were me [16:48:37] well, I personally think it's annoying that there is code licensed under CC licenses (as a former CC employee, btw), as those licenses aren't for code [16:49:10] hence, where I do have some say, I'm pushing for Platform Engineering to work on the central code repo (gadgets/etc), which might be the place to deal with that [16:57:28] The code is already on tool labs. Should I remove it until I get a response on the list? [16:59:03] you can probably wait [16:59:15] IANAL, I don't represent the WMF, etc etc ;) [17:01:01] i am not sure i like our lawyers [17:01:14] but on the other hand i don't really know them [17:02:22] luis villa is awesome [17:03:17] he was the bug wrangler at GNOME for a while, then decided to go to law school, worked at Mozilla and then the private firm that represented Google against Oracle (re java in android). Now with us. He's solid, and I say that without my WMF hat on. [17:05:05] wmf hat, hehe [17:57:07] giftpflanze: yeah, just making it clear I'm not WMF-biased about luis, known him for a long time [17:58:15] i would be upset if you would be biased just because you work there [17:58:17] ;) [20:34:23] Hello Admins! I have checked locale on tool labs grid engine. There we have 'ANSI_X3.4-1968' as preferred encoding... would it make sense to change it to 'UTF-8'...?? [20:35:19] The default locale is not set at all... [20:36:11] DrTrigon: not set is then "C" [20:37:27] (sorry don't get get it...) [20:38:50] DrTrigon: I hate repeating myself. The problem is in your code -- you are opening a file, and you should pass the filename to open as bytestring instead of a unicode string. [20:39:17] I have explained before the behavior you have in the console is basically luck, and not something you can rely on. [20:39:42] The correct way of solving this issue is setting LOCALE=C in your shell so python will not coerce non-ascii unicode strings to bytestrings using utf-8 [20:40:29] As I said before could you PLEASE help me and explain the error in the code to me? Just sending me another link about that stuff does not really help... [20:40:45] you are opening a file, and you should pass the filename to open as bytestring instead of a unicode string. [20:41:03] the link I sent explains the difference between a bytestring and a unicode/text string fairly clearly [20:41:18] let say I have a variable "abc" containg the unicode string... what to do? [20:41:39] that depends on what you want to do with it.... [20:41:50] use it as filename [20:43:03] DrTrigon: .encode it [20:43:12] with a sensible encoding [20:43:24] in the case of linux, probably utf-8 [20:43:38] give me a concrete code example - what you tell me is basic knowledge whic i also have... [20:44:05] (I also hate to repeat myself... ;) [20:44:17] abc.encode('utf-8') [20:47:15] windows is actually somewhat strange in this aspect - you can use both a unicode string (which works because NTFS does native unicode) or a bytestring (which will be interpreted using the local codepage) [20:50:09] I do not (and hopefull will never) use windows - so one issue less... ;) [20:50:52] I will check your suggestion - it sounds reasonable, but seams strange cause I am quite sure that I checked and tested this particular script for exactly that what you are mentioning... I will add it to line 720: https://git.wikimedia.org/blob/pywikibot%2Fcompat/341ed66bd3ec1ac604aa06598b9f58144891e401/sum_disc.py#L720 [20:52:20] if you want warnings on coercions, use the unicode nazi (not my terminology...): https://pypi.python.org/pypi/unicode-nazi [20:52:23] http://stackoverflow.com/questions/12557447/how-can-you-make-python-2-x-warn-when-coercing-strings-to-unicode [20:52:52] good hint thanks I will check that also! [20:53:02] DrTrigon: but _datfilename is being read fro [20:53:18] frm the config file, why note carry it one-to-one [20:53:54] yeah, you probably want to change _datfilename to always be bytestring [20:54:09] i.e. on line 527 instead [21:02:22] valhallasw: what would be the smart/elegant/clever way to do this? just to put the ".encode('utf-8')" there...? [21:03:02] that should work, yes. [21:03:27] is the config file read as UTF-8? [21:04:21] good question... which config file are you referring to? The file name gets composed out of several strings... [21:04:47] I assume them to be unicode utf-8... [21:05:06] DrTrigon: 'unicode utf-8' doesn't mean anything. [21:05:26] it's either a unicode string, or it's an utf-8 encoded bytestring [21:05:38] 525-527 are strange [21:05:46] bytestring that contains utf-8 encoded text, to be more precise. [21:06:13] _user.name() is certainly asking of utf trouble :) [21:06:27] self.site.family.name - either unicode or bytestring, but probably ascii and thus will be coerced anyway [21:06:36] same for self.site.lang [21:06:45] self._user.name() - unicode string [21:08:02] your filesystem might have trouble with various special characters in the username [21:08:59] unix filesystems do not understand unicode -- they understand bytes, and they do not care what encoding they are in. [21:09:44] frankly, if we really need per-user files there, I'd use a numeric ID [21:09:48] "_user.name()" comes wrom wiki and is utf-8 as mentioned on the console this can be used (somehow converted whatever ;) as filename...so it should also work from grid engine. to me the question is just to find a "good" solution... [21:10:02] DrTrigon: it's NOT UTF-8 [21:10:26] what is it then? [21:10:34] what come from wiki? [21:10:34] it's a unicode string [21:10:50] wiki is unicode? [21:10:59] Python [21:11:10] it's irrelevant what happens between the wiki and your code -- _user.name() returns a unicode string [21:11:44] not its not irrelevant - thats a bug - it's supposed to be the same as wiki text... [21:11:48] so what is wiki text? [21:11:53] unicode or utf-8? [21:12:15] DrTrigon: have you read the link I sent? [21:12:40] http://www.joelonsoftware.com/articles/Unicode.html [21:12:54] valhallasw: as I said read at least 10 diffrent links about unicode utf locale and all the stuff... [21:13:09] I am not an expert as you are and yeas I do alwasy mix up the terms... [21:13:52] but thats the reason why I asking you here - if you do not want to help me - fine... [21:13:55] that's not what I asked. [21:14:01] I asked whether you read *that specific page* [21:14:20] will not help I will still mix up all the stuff... [21:14:23] whotf started adding 0o in the compat/ code [21:14:39] because *that specific page* explains the difference between unicode strings, bytestrings and what encodings have to do with it. [21:14:53] the only way *not* to mix up the stuff is by understanding what you're doing [21:15:07] saper: git blame? :p [21:16:05] not thats not the point - you still can swap and mix things... but you know what - thats not what i wanted to talk about... [21:16:24] DrTrigon: I have told you exactly what you need to change [21:16:33] so have a nice evening - I will have to solve the problem formysel I see... thanks anyway! [21:16:36] bye! [21:17:59] >>> print type(wikipedia.Page(wikipedia.getSite('de','wikipedia'), u"Wikipedia:Hauptseite").get()) [21:18:02] [21:18:02] ah, gone [21:18:20] yeah, that's also unicode [21:18:33] saper: where did you find a 0o? I can't find it with grep [21:18:36] valhallasw: ah, I was accidentaly in pywikibot (the "new" one) instead of TheOneTrue "compat" [21:18:46] aaah [21:18:51] yes, core is 2.6+ :-) [21:19:10] not interested in the new line of development really [21:19:58] if compat does what you want, feel free to keep using it [21:45:39] ↑ wtf