[00:09:57] http://toolserver.org/~magnus/ts2/geneawiki/?q=Q855749 <-ok, I killed it :P [00:13:44] Whoa. [00:14:29] Nice, Mr. Manske. [00:16:06] pretty awesome [00:19:42] Next round of confirmations began [00:20:27] did someone do the stuff? [00:20:57] seems wikidata site is good [00:21:11] and it's on meta [00:22:47] cool, that went pretty smoothly [00:22:52] duh, added :D [00:22:57] yay! [00:23:17] and just fyi, the hidden comments i added were just do i would know which is which, they arent required [00:23:39] yeah, i know, but i added them for fun :P [00:24:03] and i've quality checked all the country IDs (to make sure they're right) and categories (to make sure they don't have any unrelated crap) [00:24:47] awesome [00:24:50] rschen7754: Us admins split the work :P [00:24:57] :) [00:25:05] once the bot is approved by someone, ill copy them over and load them up [00:25:14] duh: Which bot? [00:25:18] Legobot 4 [00:25:25] all concerns have been addressed [00:25:37] and its been open for +1 week i think [00:25:38] was it clarified somewhere whether neutrals get counted in RfA? [00:25:47] neutrals aren't counted [00:25:57] duh: Shouldn't your RfA end later? [00:26:01] only on commons do neutrals count as a semi oppose or whatever [00:26:05] Ajraddatz: for what? [00:26:12] RfAs/RfPs [00:26:19] Hazard-SJ: yeah my RfA closes tomorrow i think. This is for a bot request. [00:26:22] Ajraddatz: ok, I was thinking in the Commons way [00:26:31] https://www.wikidata.org/wiki/Wikidata:Requests_for_permissions/Legobot_4 [00:27:16] what's consensus for bots? everyone's happy? [00:27:25] yeah [00:27:27] duh, i assume that when the bot finds an article in more than one of those categories, it will just add the necessary item to the correct property? [00:27:30] I don't think it's been defined yet [00:28:14] Jhs: yes, it treats each template row thing as an individual request [00:28:22] so if you [00:28:33] |ignore an article from one, and not another, it only uses the list for that specific request [00:28:41] duh, closed as done [00:28:49] thanks [00:29:33] nice nice :D [00:31:09] ok gotta go afk for 10 min, and then ill come back and fire it up :D [00:32:12] yay! [00:33:47] Ajraddatz: I know [00:34:06] duh: Care to run another trial? :P [00:34:15] Hazard-SJ, then why did you ask O_o [00:34:18] Hazard-SJ: for which bot? [00:34:52] Ajraddatz: I was wondering if I made a mistake in closing one of the confirmations [00:34:55] duh: 4 [00:36:25] yeah once i go upstairs, i got stuck helping someone in #wikipedia [00:37:01] what a nice guy [00:37:06] OK [00:38:13] duh: Actually...since Ajraddatz closed it, it wouldn't exactly be a trial :P [00:38:39] when it comes to duh, trials are a bit pointless [00:39:00] he just sprinkles his fairy dust everywhere and it works :D [00:39:07] :D [00:39:12] my bot has only screwed up once on wikidata :P [01:04:58] Hazard-SJ: the CennoxX confirmaion caught my eyes, it would be unsuccessful if it were on Commons [01:05:39] don't know how others naturally expect [01:06:35] whym_away: I just closed them per https://www.wikidata.org/wiki/Wikidata:Administrators [01:06:55] At least 8 supports, and >= 75% support [01:07:27] I've never liked how commons does it with the neutrals. If I vote "neutral" on someone, I don't want it being counted against the person. [01:07:34] duh: rivers of africa :/ [01:07:38] Feel free to propose modifications, whym_away [01:07:57] Ajraddatz: That's like counting them as weak opposes [01:08:18] yes, commons interepretation of "those who voted" include neutrals [01:08:25] a footnote on this might be a good idea [01:08:48] rschen7754: yeah i'm just reviewing the code since ive made quite a few modifications since i last ran it [01:08:53] ok [01:08:59] * Hazard-SJ blames duh's bot [01:09:30] Hazard-SJ: maybe we should schedule our bots to run at different times? [01:09:40] mine runs on the hour and apparently takes 3 minutes for the query [01:09:52] i can switch it to :30 [01:10:14] but [01:10:22] my bot does check that the item hasnt been reported yet [01:10:28] but what if it takes over an hour for the run? [01:10:35] oh, that's for the other task [01:10:59] rschen7754: since i use SGE, if a version of that script is already running, it wont start another process [01:11:05] oh [01:14:49] duh: Mine takes over 4 hours to run, and just stuffs on everything at once [01:14:56] ah [01:15:15] maybe check right before saving that the items havent been reported yet? [01:15:28] also, [01:15:34] why does it take that long? [01:16:07] duh: I'll ask it kindly to check that the item still exists just before requesting deletion [01:16:16] eh [01:16:16] duh: It has many wikis to check [01:16:20] i dont think thats a good idea [01:16:24] :/ [01:16:35] if you just check if it exists, you're not checking if it has any sitelinks [01:16:36] etc [01:16:49] and its possible a human already reported it [01:16:53] or my bot [01:17:08] its easier just to check if 'Q42' on rfd.get(): [01:17:16] duh: The sitelink check would have already been made by then [01:17:26] yes but its possible someone re-adds it in that much time [01:17:29] or something [01:17:36] im more concerned about the four hours part [01:17:54] my redirect bot takes 30 minutes to run an expensive query against all wikis [01:18:31] duh: Do you want to see the code? [01:18:41] yes please [01:26:04] * Hazard-SJ uploads code [01:27:10] duh: The statistics from the last run are in https://www.wikidata.org/?diff=6733513 [01:27:52] yeah thats a long time... [01:29:05] also [01:29:12] find a better spot than the sandbox to do that :P [01:31:53] duh: https://github.com/HazardSJ/Hazard-Bot/blob/master/Wikidata/delsitelinks.py [01:32:02] ok will look in a few minutes [01:32:07] OK [01:37:52] https://en.wikidata.org/wiki/Special:Contributions/Legobot :D [01:40:02] :D [01:40:40] great... [01:40:48] i just figured out theres a bug when the bot stops [01:41:56] How do you add a discription when there appears to be no edit button. I am able to type in the discription but have no way of saving it, closing the page loses the edit see "SpamCop" that has no description [01:42:27] dbiel: have you tried hitting "enter' [01:43:36] Yes, it appears to do nothing as well [01:43:44] dbiel: Just so you know, you can't be editing descriptions and, say sitelinks or labels at the same time [01:43:47] link to the specific item? [01:44:01] dbiel: Tried reloading the page? [01:47:24] Hazard-SJ: i think i found your problem [01:47:45] also your unicode support is…well bad. [01:48:21] if rfd != []: [01:48:22] ewww [01:48:24] if not rfd: [01:49:05] anyways [01:49:06] your main problem [01:49:07] I just found the problem as well, IE 8 does not seem to be supported. going over to Google Chrome the edit fields appear [01:49:14] dbiel: ah right yes [01:49:18] there are issues with IE8 [01:49:24] theres a thread on project chat about it i think [01:49:38] Thanks for the help [01:52:44] bye [01:54:53] anyways, Hazard-SJ your issue is with your generator [01:55:06] duh: You mean if rfd:? [01:55:08] you're just fetching say the last 200 deletion entries [01:55:14] Hazard-SJ: oh right, yeah if rfd: [01:55:58] you should check based on timestamp [01:56:12] like "get all entries since blah" [01:56:23] you should also filter undelete entries and such [01:58:53] duh: How do I even know if it is an undelete entry?:S [01:59:14] There isn't an undeletion log [01:59:27] you need to filter by logaction vs logtype [01:59:31] logtype="delete" [01:59:39] logaction="delete" or "undelete" [02:01:25] duh: pagegenerators.LogpagesPageGenerator(number = number, mode = 'delete', site = site) [02:01:48] pagegenerators.LogpagesPageGenerator doesn't support anything of the sort [02:02:05] yeah [02:02:19] you're goign to have to directly use the QueryGenerator [02:03:22] * yurik doesn't like pywiki implementation :( [02:03:25] duh: You mean query.GetData, right? [02:03:36] no [02:03:41] i mean the generator [02:03:42] let me find it [02:03:50] the generator automatically handles the query-continue for it [02:04:00] yurik: So fix it [02:04:27] Hazard-SJ: just like in biology, after a certain point, it is easier to have a new kid than to fix an old person :( [02:04:48] ….. [02:04:53] that was funny [02:05:09] there are many conceptual problems that i see with pywiki [02:05:40] a) pywiki overconcentrated on wrapping everything into objects and methods, at the expense of flexibility and efficiency [02:05:49] ^ [02:06:12] yurik: Would the new kid be the rewrite and the old person be the trunk? :P [02:06:16] it is almost impossible to use api to the fullest - like use the generator AND get langlinks AND get categories [02:06:23] unfortunatelly no [02:06:34] i looked closelly at it. it is much better, but the core problem remains [02:07:02] Hazard-SJ: https://github.com/legoktm/pywikipedia-rewrite/blob/master/pywikibot/data/api.py#L842 [02:07:04] yurik: the query.GetData allows that, doesn't it? [02:07:21] yeah but then you're just making direct API queries [02:07:22] Hazard-SJ: what do you mean? [02:08:51] duh: all those generators are not really generators - they simply do generator=list & prop=pageinfo, after which you have to do another query to actually get the properties you need [02:09:03] duh: That isn't in trunk as far as I'm aware [02:09:15] yurik: yeah but it does what Hazard-SJ needs [02:09:21] according to ori, most bots don't get more than one page at a time, resulting in a huge inefficiency [02:09:22] Hazard-SJ: ouch. well then GetData works I guess [02:09:27] yes! [02:09:36] thats the main issue [02:09:43] rewrite kinda helps with that [02:09:57] duh, i don't doubt that it "works", it is "how" it works is what causes grief :) [02:10:06] :P [02:10:21] the reason for all this is that pywiki has a huge legacy of dump processing and page scraping :( [02:10:50] rschen7754: can you "approve" your roads jobs and Jhs's rivers jobs (I checked them) by copying them over to the main page. I don't feel like logging into the bots account to do it [02:11:03] ok let me just vote for a bugzilla bug [02:11:13] …you do realize that does nothing right? [02:11:17] aww [02:11:23] it adds me to teh cc list [02:11:25] no one looks at those really [02:11:34] is there a log somehwere? [02:11:38] of completed jobs [02:12:04] yes [02:12:15] IMO, the api should have been something like this: get a site object that self populates with the meta info, and expose a raw api object to do actions, and for query do an extra query object that returns merged page objects (in case the continue was called) [02:12:23] rschen7754: https://www.wikidata.org/wiki/User:Legobot/properties.js/Archive [02:13:43] duh: in other words: in python code it should look like this: [02:13:47] for result in pywiki.query( {'generator':'allpages', 'prop':'links'} ): [02:13:48] # process result data [02:13:51] yes [02:14:42] duh: done [02:14:48] where "result" is the dict returned from the server [02:15:10] rschen7754: er can you remove the html comments [02:15:24] this way there could be a wrapper that yields one *complete* page dictionary at a time [02:15:26] :( [02:15:38] just use find+replace [02:15:54] find: replace: '' [02:15:58] with regex enabled [02:16:47] i'll just do it by hand [02:17:02] api would handle continuing between requests, merging of incomplete page dicts (imagine if you request multiple properties and some get populated and some are only returned in the next call), and possibly even revision changes - api can detect if the page has changed between calls [02:17:26] but you get back a simple dict {} that describes everything you requested [02:17:42] hm [02:18:03] would that be done client side or server side? [02:18:17] https://www.mediawiki.org/wiki/Requests_for_comment/API_Future#Easy_continue [02:18:51] kinda both - the server will notify the client if the page is "incomplete" (although we can work around it even now with some extra python code) [02:18:51] oh right [02:18:56] thats on my to-read list. [02:19:07] see the light green samples [02:19:54] and the client will know how to handle basic things like errors (will raise exceptions), log warnings, but won't try to do complex object creation [02:20:29] i am seriously thinking of rewriting the framework from scratch, and simply reusing some components from pywiki [02:20:37] duh, should be good [02:20:45] thanks [02:20:53] np [02:20:56] soon ill write something in so the bot will ignore it [02:22:04] duh: since you wanted +2, and since i recently requested it too, it might be a good team project ;) [02:22:12] i dont want +2... [02:22:21] are you confusing me with hoo? [02:22:34] huh? [02:22:38] duh [02:22:45] hoo? [02:23:01] * hoo is now in confused mode [02:23:03] ok starting the bot again [02:23:27] * Hazard-SJ has to go [02:23:29] Bye [02:23:37] dope, there are two of you :( [02:23:41] :D [02:23:59] duh hoo! [02:24:04] i dont do much core hacking [02:24:37] mainly AbuseFilter stuff [02:25:04] duh: You could at some point go for +2 on AF... I'm atm doing that more or less alone [02:25:09] sorry for the confusion, although if you do bot hacking, you might have some feedback :) [02:26:11] hoo: eventually maybe. it would be bad to give +2 to someone who doesn't know php though :P [02:26:11] * yurik thinks it should be possible to comment on each revision people do... like in facebook :) [02:26:26] duh: You don't know php?! [02:26:33] php is evil, stay away [02:26:44] https://en.wikipedia.org/wiki/User:Legoktm/php-0 [02:26:55] * yurik is secretly planning to rewrite the entire MW in C# [02:27:06] yurik: someone is planning to do it in python actually [02:27:09] yurik: Go for Python and I'm in [02:27:11] :D [02:27:26] it was on labs-l [02:27:33] duh: Whoops, things I shouldn't have said loud :D [02:27:39] hahah [02:27:40] hehe. I spoke with .... the guy who is a dev in PHP, forgot his name, and he was thinking about it [02:27:59] i mean - he is a dev of the php itself [02:28:23] the idea was to start migrating by rewriting api in php [02:28:35] yurik: He was thinking of porting MW away from PHP? [02:28:36] and then some other services (parser, etc) [02:28:39] yep [02:28:44] :D [02:28:55] it is a dead end unfortunatelly :( [02:29:13] you just can't do enough CPR to bring to life :) [02:29:24] broken by design :) [02:29:31] my fav - the array() object [02:29:43] the most hated thing imo [02:29:56] * yurik stops ranting [02:31:00] i still think c# is better - it forces you to write proper code for large systems, and it handles upgrades more gracefully [02:31:06] * yurik stops ranting again [02:35:39] * Jeblad_WMDE yawns [02:36:45] rschen7754: https://www.wikidata.org/w/index.php?diff=6740311&oldid=6738657&rcid=6753075 [02:37:16] Jeblad_WMDE: hi, would you know when i can expect https://gerrit.wikimedia.org/r/#/c/49205/ to get deployed? [02:39:16] I'm not sure if it is on the list of things to get backported, if not it will take a little over two weeks [02:39:43] who do i need to talk to to get it backported? [02:40:06] because bots should be able to edit as bots [02:40:12] mine is filling up recentchanges right now [02:40:37] I don't think its on the prioritized list, .. [02:41:59] you're the one who marked it as "high" priority :P [02:46:35] duh: ok cool [02:46:37] th [02:46:38] x [02:47:32] woot [02:47:34] my bot didnt crash [02:47:50] i was afraid Jhs's name would kill it [02:48:02] yay :D [02:48:31] rschen7754: can you copy over one of my jobs please? i want to make sure the bot can handle other people editing the page when its running [02:48:43] ok [02:49:40] done [02:49:41] duh: setting it to high doesn't mean it goes in, just that someone should do the necessary job before other stuff that is flagged as normal [02:50:34] Jeblad_WMDE: ok makes sense. but so how do i get it backported? its a rather simple fix that is needed on the live wiki soon [02:54:00] aaah duh abort [02:54:13] uhoh.... [02:54:15] i killed it [02:54:19] :( [02:54:22] whats up? [02:54:58] i somehow put in the property "is a" instead of "country" in the field� :/ [02:55:03] duh, the list of things to be backported is very long [02:55:52] And it is other things with higher priority [02:56:05] But perhaps, I don't know [02:56:47] Jeblad_WMDE: well i'm pretty there are way more important things [02:56:51] "is a" is problematic and I don't think it should be used without some clearification [02:57:13] Jhs: had the job already run/was running? [02:57:17] rschen7754 can rollback the bot [02:57:22] yeah, but only on a few pages [02:57:23] :/ [02:57:25] i can fix it by hand [02:57:29] ok [02:57:33] we need detachable rollback [02:57:43] New patchset: Hoo man; "Log API errors in JavaScript to console if in debug mode" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/48968 [02:58:39] New review: Hoo man; "Patch Set 4:" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/48968 [02:58:57] Jhs: ok can you let rschen7754 which jobs need to be fixed/removed before i start the bot up again? [02:59:00] New review: Hoo man; "Patch Set 3:" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/48968 [03:00:53] :o jhs can edit .js pages? [03:01:06] couldn't i? [03:01:17] non admins cant... [03:01:21] oops, i just misused global sysop rights [03:01:22] maybe you have it globally? [03:01:23] :P [03:01:24] oh well [03:01:33] i dont care [03:01:48] and moving them to the bottom doesnt work since the bot loads the job page all at once [03:01:57] so just stick a # in front of them for now [03:02:00] to comment them out [03:02:27] jhs mostly do things well, mostly.. [03:02:31] oh, that's who jhs is… :P [03:02:49] * Jeblad_WMDE plan to kick jhs but the nxt time he is in Norway [03:02:57] h�h� [03:04:30] jhs, it would be nice if you could take time to proofread the translations to norwegian [03:04:43] Jeblad_WMDE, yup. tomorrow maybe [03:04:50] Some of the new english source messages are a bit weird [03:06:21] Jhs: can you comment out the jobs you moved to the bottom so the bot wont accidentally run them? [03:06:27] or are those safe to run? [03:07:05] they are safe, yeah [03:07:17] ok [03:08:15] rschen7754: propose detached rollback on PC? [03:08:25] duh: it already failed [03:08:30] oh [03:08:46] One of the extraordinary weird messages https://translatewiki.net/wiki/MediaWiki:Wikibase-snakview-variation-datavaluetypemismatch-details/en [03:09:03] Its kind of cute [03:10:44] heh [03:10:56] "property's data type's data value type" [03:10:58] Jeblad_WMDE, lol [03:12:24] Verdiens dataverditype �$1� passer ikke med egenskapens datatyps dataverditype �$2� [03:12:40] det h�res ut som en �ystein Sunde-sang [03:12:42] its a property that has a datatype, but the datatype of the datavalue does not match up [03:13:01] It needs rewording [03:13:36] indeed [03:13:48] rschen7754: is https://www.wikidata.org/wiki/Q408192 a property of itself? [03:13:51] er like [03:14:03] ooh [03:14:05] no [03:14:10] oops [03:14:11] * duh goes to fix recursion [03:14:20] lol [03:14:35] recursive items are recursive [03:16:05] is https://www.wikidata.org/wiki/Q179976 right? [03:16:26] oh [03:16:28] no that doesnt work [03:16:29] ugh [03:16:42] Just as a note; "is a" -type of properties are transitive, and we don't have code to support them [03:17:07] Recursion: See Self-Referencing [03:17:24] Self-Referencing: See Recursion [03:17:38] To handle transitive properties are difficult and it is not done in a few days to implement them [03:17:51] Error. Stack overflow. [03:18:30] It is necessary to implement that part to make "is a" work for some types of properties [03:18:50] or rather parameters in templates on wikipedia [03:19:08] Thats why "is a" should not be used [03:21:23] hm [03:21:31] i cant figure out a better way to make the recursion joke [03:22:37] public static void recurse(){recurse();} [03:22:46] duh, i really think Google nailed it http://www.google.co.uk/#hl=en&tbo=d&output=search&sclient=psy-ab&q=recursion&oq=recursion&gs_l=hp.3..0l4.978.2105.0.2177.9.8.0.0.0.0.158.850.5j3.8.0.les%3B..0.0...1c.1.3.psy-ab.NTfEcP5VAqo&pbx=1&bav=on.2,or.r_gc.r_pw.r_qf.&bvm=bv.42553238,d.bGE&fp=27da8d583881e653&biw=1366&bih=682 [03:23:00] yeah, i was trying something like that [03:23:09] (hmmm� i have no idea why the url for a simple google search is so long) [03:23:25] most of it is analytics, its just the &q=recursion [03:23:42] i was only able to come up with is a [03:23:45] which doesnt work [03:29:29] Imagine implementing sizzle for Wikidata [03:29:39] Not sure if it is possible at all [03:30:52] * Jeblad_WMDE plan to do some work on the sleep protocol [04:06:31] New patchset: Hoo man; "(bug 45043) Overhauled site input field" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/49460 [04:10:14] New patchset: Hoo man; "(bug 45043) Overhauled site input field of wbclient.linkItem.js" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/49460 [04:50:36] does wikidata parse titles with lang+site properly? Something like [[w:en:API]] --> enwiki / API [04:54:00] duh: i am thinking of your idea, and i can't come up with any scenario useful for the client [04:54:13] What do you mean? [04:54:31] well, what kind of request would a bot/user need? [04:54:38] translation. [04:54:40] Its easy [04:54:47] can you give an example? [04:54:49] You store each Qid as an "object" [04:54:54] user clicks on it [04:54:58] depending on the language set [04:55:03] you show them the content in that language [04:55:16] wait wait, is that in a client or repo? [04:55:27] which site? [04:55:46] i'm failing to see the possible usage [04:55:48] 3rd party [04:55:50] I dunno [04:55:54] Just an idea. [04:55:57] Mainly [04:56:07] no,it might be good, i am just trying to figure it out [04:56:09] If my bot has the item # [04:56:15] And I want enwiki's page content [04:56:25] I need to look up the enwiki sitelink, then another request for page content. [04:56:54] ok, but how would you have that Q#? [04:57:02] Because of properties [04:57:03] how would bot get that list [04:57:26] Say I want list of males born in DC or something stupid [04:57:33] That would return me IDs. [04:57:55] who would return the ids? wikidata will give you the list of titles in any lang you want [04:58:50] I'm not sure [04:58:58] I thought there was going to be a query interface for stuff like this, [04:59:19] there will be, i'm just trying to understand which queries will be useful [04:59:40] maybe the whole wbtsearch (as described https://www.mediawiki.org/wiki/Requests_for_comment/Wikidata_API ) [04:59:44] can be made remotable [05:00:04] so that anything it can do you can use from any site [05:00:12] and get pageids back [05:00:17] of the local site [05:00:30]