[14:17:20] Funny how gerrit keeps saying something happened "in the future" [14:17:42] Hm, or is that a problem of clock sync on this PC [14:18:01] Looks correct [14:30:00] i noticed that too [15:19:01] legoktm: global css/js working for me too! no idea what greeting/praise could be sufficient for you today [15:19:07] :))) [15:19:30] perhaps a wikipedia holiday [15:25:54] legoktm, wait... it's live???? [15:26:01] yup [15:26:14] * Cookies52 gives legoktm cake. lots of cake [15:26:22] :> [15:29:29] legoktm, congrats [15:29:44] legoktm, the migration script appears a bit strangely on the wikis [15:29:49] oh, how so? [15:29:52] e.g. https://be.wikibooks.org/wiki/%D0%90%D0%B4%D0%BC%D1%8B%D1%81%D0%BB%D0%BE%D0%B2%D0%B0%D0%B5:RecentChanges [15:30:05] like an anonymous user on that page [15:30:08] the account doesn't actually exist [15:30:22] But if you go to Special:Log, you see nothing... [15:30:57] hm :/ [15:31:51] It has log_user = 0 [15:31:57] hi, short question about the API title-to-URI relation... is it better to store both (so on API requests get inprop=url and store the fullurl) or is it better to use some client lib (and which one would you recommend for python?) later on and re-create the URI from the title? [15:32:32] nowadays special:contributions checks for ID [15:32:35] yeah [15:35:53] Krenair: do you think I should have it make an account? [15:36:47] or just reuse MediaWiki_default? :D [15:37:07] jorn: you probably only need to store the title. https://www.mediawiki.org/wiki/API:Client_code#Python is a list of Python API libraries [15:37:11] Don't we have some solution to this in core? [15:37:11] I could do that too [15:39:18] legoktm: i've seen those libs, but for example it seems that "wikipedia" does an API request to get the url [15:40:05] well, you can also assume the url is going to be en.wikipedia.org/wiki/$1 where $1 is the page name [15:40:08] i'm not sure about the others, that's why i asked ;) [15:40:15] if you just want to map titles --> urls, I would just hardcode it [15:42:14] legoktm, take a look at what AbuseFilter does in AbuseFilter::getFilterUser [15:42:34] yes, but that is a bad hack [15:42:54] I'll just use MediaWiki default [15:43:02] legoktm: well, there's encoding stuff going on... and it's not just urlquote it seems to me (" " becomes "_")... i just don't want to guess on this thing :-/ is there a "spec" how to transform a title -> URI (and maybe URI -> title) somewhere that i don't know of? [15:43:25] actually, that doesn't have accounts on all wikis? https://tools.wmflabs.org/quentinv57-tools/tools/sulinfo.php?username=MediaWiki+default [15:43:49] jorn: I think " " --> "_" is the only thing. let me see if it's documented anywhere... [15:44:20] should be. [15:44:46] the other limitations are illegal characters. but those shouldn't be there. [15:45:36] Nemo_bis: https://tools.wmflabs.org/quentinv57-tools/tools/sulinfo.php?username=MediaWiki+default doesn't have an account on all wikis? [15:45:44] and https://en.wikipedia.org/w/api.php?action=query&meta=siteinfo&siprop=interwikimap&sifilteriw=local [15:45:57] has all the urls that you can put the title behind. [15:47:31] and not that /wiki/$1 has url part encoding and ?title= has querycomponent encoding usually [15:50:32] ori: on the rise last month? https://gdash.wikimedia.org/dashboards/frontend/ [15:51:20] legoktm: possible, for wikis created after its last use? [15:52:01] I guess. I'm not really sure how that account works [15:52:35] I also need to globalize it so it doesn't get renamed... [15:53:46] bbl in 10, finding food [16:04:10] thedjNotWMF: hmm ok, i guess url part encoding refers to php's way of doing the pathname encoding? cause for example the python.quote method will do this differently: it will quote parenthesis: ( -> %28 [16:05:36] (sorry to be so pedantic about this, but correct URIs matter for Linked Data stuff, as they're only considered equivalent if byte equivalent for fast checking :-/) [16:06:25] jorn: quote does query component encoding, not path encoding [16:06:44] jorn: there is a difference between what happens in title=sometitle and what happens in /wiki/sometitle [16:07:07] hmpf, now why was I logged out everywhere [16:07:12] valhallasw`cloud: from the urllib.quote doc-string: "By default, the quote function is intended for quoting the pathsection of a URL." [16:07:52] Nemo_bis: you mean :? [16:08:01] jorn: doh, you're right. I'm confusing it with urllib.urlencode [16:08:42] no worries, i was just checking... i made a bug-report about this back in 2011 but they never fixed it... http://bugs.python.org/issue12910 [16:09:10] jorn: quoting () is not necessary, but also not wrong [16:09:37] jorn: it's an equivalent representation, after all [16:09:46] legoktm, right... I'm gonna admit defeat. Where does the global js go? [16:09:54] valhallasw`cloud: well, for performance reasons the RDF spec define 2 URIs to be considered equal only if byte equivalent [16:10:06] jorn: right, so you need to normalize them in some way [16:10:07] Cookies52: did you read https://www.mediawiki.org/wiki/Help:Extension:GlobalCssJs ? :) [16:10:10] so if one is a more quoted version of the other it doesn't work ;( [16:10:36] legoktm, I read another documentation entirely for something else... :/ [16:10:56] jorn: so basically the question is mainly what mediawiki uses for tags? [16:10:57] exactly... that's why i was asking about the spec for title -> URI conversions, as they don't seem to be following any simple rules in python [16:11:03] yupp [16:11:32] and not just what works (by wikimedia being so nice) [16:11:37] ;) [16:12:01] jorn: it's not wikimedia 'being so nice', it's the RFC that defines them to be equivalent :P [16:12:41] well, actually you even take UTF-8 encoded stuff and serve it [16:12:55] I'm not even sure if mediawiki sees the difference between ( and %28; [16:13:35] jorn: are you looking for https://www.mediawiki.org/wiki/Manual:PAGENAMEE_encoding ? [16:14:39] Nemo_bis: thanks [16:14:43] jorn: https://github.com/wikimedia/mediawiki-core/blob/555e0b4b3c517cfb565ad275d9600806cd3cd50a/includes/title/MediaWikiPageLinkRenderer.php [16:15:15] curl -I 'http://en.wikipedia.org/wiki/Gerhard Schröder' [16:15:20] which uses https://github.com/wikimedia/mediawiki-core/blob/555e0b4b3c517cfb565ad275d9600806cd3cd50a/includes/Linker.php [16:20:18] Krenair, Nemo_bis: So I'm just going to keep running it with "GlobalCssJs migration script"... [16:21:32] valhallasw`cloud, Nemo_bis: thanks, it's actually wfUrlencode... sorry i couldn't find this myself [16:28:19] valhallasw`cloud: about the 'being so nice': http://p.defau.lt/?0DrF2ZSdfkLR_gFjx30p2w so something actually even serves UTF-8 IRIs which haven't been % encoded as URIs... [16:28:36] and they don't seem to be the same thing served from the cache... [16:30:13] jorn: either nginx or apache [16:30:54] I thought maybe curl encoded it, but it doesn't [16:34:56] well, i don't know if it's a bug or a feature or if the feature includes a bug (as it doesn't seem to serve from the same cache), but i guess in the end it doesn't matter for wm to much as most browsers use magic ;) [16:37:41] you just notice it when checking SemWeb URIs for equality and they don't mach :-/ http://www.w3.org/TR/2014/REC-rdf11-concepts-20140225/#section-IRIs : "IRI equality: Two IRIs are equal if and only if they are equivalent under Simple String Comparison according to section 5.1 of [RFC3987]. Further normalization MUST NOT be performed when comparing IRIs for equality." [21:18:49] helderwiki: thanks for fixing https://bi.wikipedia.org/w/index.php?title=MediaWiki:Common.js&diff=prev&oldid=32938 (which I was now breaking again for impatience :p https://bi.wikipedia.org/w/index.php?title=Germany&diff=32970&oldid=32969 ) [21:20:00] :-) [21:20:33] Fixing things is addicting... [21:20:52] Please spread the contagion