[01:38:43] New patchset: Pragunbhutani; "Adds a link to corresponding Wikibase Item to action=info" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64644 [02:20:24] New review: Hoo man; "(Untested) The code looks good, just some nitpicks" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64644 [02:27:58] New patchset: Pragunbhutani; "Adds a link to corresponding Wikibase Item to action=info" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64644 [02:30:09] o.O Portal reference! :) [02:31:01] TCN7JM: yes. Are you an admin yet? [02:31:12] Tomorrow. [02:31:37] TCN7JM: then tomorrow I will give you the logs where the context for the nick change is [02:32:13] Okay. [02:32:19] I'll remember to ask you about that if you forget. [02:32:36] which I will, since I'm terrible about things like that [02:57:31] *ding* Ladies and gentlemen: There are currently a large number of open entries at https://www.wikidata.org/wiki/Wikidata:Properties_for_deletion , many of which have low levels of participation. Your participation would be appreciated. [02:58:34] I'm concerned about P131. [02:58:50] Is there a way, if it is deleted, to quickly switch to the correct properties in its place? [02:59:15] TCN7JM: kinda. It's called "the closer fixes it by hand" [02:59:37] Considering that I close many of them... grumble grumble [02:59:46] Well...uh...P131 is used in thousands of items, I would think. [03:00:10] ...grumble grumble [03:00:18] lol [03:00:18] I will ask legoktm to make me a tool [03:00:34] Why don't we just have AWB on Wikidata or something? [03:03:03] well I don't know how to use AWB... [03:03:29] Really? I would think you have the rights to it on enwiki. [03:03:36] also, AWB would need a major rewrite to work for WD [03:04:32] Which is why I said "or something", realizing it wouldn't work in its current form. [03:04:43] AWB is also windows-only [03:05:37] Windows is also awesome. [03:05:54] :) [03:06:15] rschen7754: I use Windows. I close RfD/Properties [03:06:25] but I don't know how to use AWB [03:07:10] How do you set up an edit notice [03:07:28] I was wondering the same thing. [03:07:33] err /Editnotice work? [03:07:35] Sven_Manguard: for RfD? [03:07:40] it's per-namespace [03:07:44] Jasper_Deng: Yes. [03:07:52] * Jasper_Deng was the one who set it up for the User and User talk namespaces [03:07:59] check my MediaWiki: namespace edits [03:08:31] Jasper_Deng: That's what I was wondering about. /Editnotice works for User and User talk namespaces? [03:08:44] yes [03:08:49] Okay, thanks. [03:09:44] * Hazard-SJ wonders what he is missing, but then decides he can't bother to read everything [03:11:36] Jasper_Deng: will /Editnotice work on the Wikidata namespace or not? [03:11:54] no [03:12:01] unless someone set it up in the meantime [03:12:03] what will it take to make it work? [03:12:09] * Sven_Manguard pushes Jasper_Deng to do it [03:12:43] * Jasper_Deng pushes Sven to learn how instead [03:13:17] Jasper_Deng: that'd take much longer [03:14:27] Well [03:14:37] I was working on a Python AWB for Wikidata [03:14:57] But like a bunch of other wikidata projects it got neglected :( [03:15:05] my other* [03:15:17] Jasper_Deng: {{#ifexist:{{FULLPAGENAME}}/Editnotice|{{{{FULLPAGENAME}}/Editnotice}}}} [03:15:19] [2] 04https://www.wikidata.org/wiki/Template:FULLPAGENAME [03:15:29] I don't get it [03:15:33] [07:58:50 PM] Is there a way, if it is deleted, to quickly switch to the correct properties in its place? <-- Oh, thats easy to whip up a script for. [03:15:50] Sven: if the editnotice exists, then transclude it, otherwise, no [03:15:53] ^^ [03:16:55] lego: what I don't understand is how account creators can create edit notices on enwiki.... how does MediaWiki do that? [03:17:48] Sven_Manguard: You want edit notices for project pages based on /Editnotice? [03:17:53] yes Hazard-SJ [03:18:07] * Hazard-SJ goes to do it slowly [03:18:29] * Sven_Manguard thanks Hazard-SJ slowly [03:18:54] Jasper_Deng: because editnotices are protected by a titleblacklist rule [03:19:03] accountcreators have the tboverride right [03:19:14] a dirty hack IMO [03:19:20] indeed! [03:20:13] * legoktm goes afk [03:25:22] Sven_Manguard: Try it now [03:38:34] Hazard-SJ: I'm going to be out for the next 20 mins [03:39:21] OK [03:55:05] back [03:55:07] let me try [03:56:18] Sven_Manguard: I tried for the sandbox before you left initially, and I g2g soon :P [03:58:53] Hazard-SJ: it works [03:58:57] thank you [03:59:53] You're welcome :) [04:00:30] Sven_Manguard: Might be useful if you categori[sz]e whatever edit notice you're making ;) [04:00:36] * Hazard-SJ says bye [04:00:44] bye [04:11:20] Jasper_Deng: will langswitch work in /Editnotice? [04:11:39] which use? [04:12:02] * Jasper_Deng would leave it up to each individual user [04:12:34] Jasper_Deng: my question is whether or not anything would stop that tempate from working [04:47:12] * TCN7JM cues Queen music. [04:47:26] *bass guitar* Another one gets the mop! [04:47:33] * TCN7JM cuts Queen music. [05:47:35] New patchset: Aude; "(bug 48224) Format site link already used message with correct link" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64881 [05:48:57] New patchset: Aude; "(bug 48224) Format site link already used message with correct link" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64881 [05:49:47] New review: Aude; "see https://gerrit.wikimedia.org/r/#/c/64881/ which fixes the site link already used error message" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64083 [07:38:10] New patchset: Aude; "(bug 48473) validate claim guid in api and return error" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/63787 [07:39:16] New patchset: Aude; "(bug 48473) validate claim guid in api and return error" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/63787 [07:42:34] New patchset: Aude; "(bug 48473) validate claim guid in api and return error" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/63787 [08:32:48] Silke_WMDE: http://noc.wikimedia.org/conf/ [08:33:01] scroll down to see db lists [08:34:53] New review: Aude; "I have installed squid on some of my test wikis, but still cannot reproduce the issue we had during ..." [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64274 [08:41:25] New review: Tobias Gritschacher; "(1 comment)" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/63787 [08:51:42] New patchset: Daniel Kinzler; "Don't use flush(), it interferes with output buffers" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64274 [08:52:47] DanielK_WMDE: i think you are running out of excuses to check this one and have it merged :) it passes the tests now [08:53:57] New review: Daniel Kinzler; "@aude: the current code is broken, this change *probably* fixes it, and is unlikely to cause harm. I..." [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64274 [08:54:42] Denny_WMDE: "this ohne" being? perhaps i should un-ignore gerrit :) [08:54:58] * aude wants a second opinion on https://gerrit.wikimedia.org/r/#/c/64274/ [08:55:00] i suppose it's you patch from yesterday. [08:55:03] will look at it in a minute [08:55:12] looks sensible though [08:55:17] didn't get around to reviews yesterday much, so i'll start with that today. [08:55:38] we need this or something in the branch today (what's on master is broken) [08:55:49] DanielK_WMDE: my fault, didnt add the link https://gerrit.wikimedia.org/r/#/c/64538/ [08:55:51] aude: as i said, i don't see a reason not to merge it even if it turns out that this isn't sufficient to fix the original problem [08:56:01] not flushing there is the right thing, regardless [08:56:28] someone want chocolate :) [08:56:57] aude: can you just grab tobi or someone and get a second oippinion out of them? [08:57:08] * DanielK_WMDE joins the hangout now [08:57:41] (oh god i need a hair cut) [08:57:43] Tobi_WMDE: ^ [08:59:04] https://gerrit.wikimedia.org/r/#/c/64794/ please merge [09:01:00] Change merged: Tobias Gritschacher; [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64794 [09:18:07] New patchset: Tobias Gritschacher; "Removed word 'proleptic' from message" [mediawiki/extensions/DataValues] (master) - https://gerrit.wikimedia.org/r/64915 [09:19:03] Denny_WMDE: please chack: https://gerrit.wikimedia.org/r/#/c/64915/ [09:22:08] Change merged: Denny Vrandecic; [mediawiki/extensions/DataValues] (master) - https://gerrit.wikimedia.org/r/64915 [09:22:35] DanielK_WMDE: the thing with the haircut should be realized by a mirror, not by a hangout :) [09:22:43] Tobi_WMDE: thanks! [09:24:18] New patchset: Daniel Kinzler; "changed namespaces a bit, added license" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64538 [09:25:06] Denny_WMDE: what does redundant web storage have to do with my hair cut? [09:26:07] DanielK_WMDE: :D [09:26:29] Denny_WMDE: btw - when generating a dump of many/all entities, perhaps the "meta data" section should be skipped (or rather, there should be only wone for the entire dump) [09:26:44] no need to skip it [09:26:54] it just describes the individual dataset [09:26:57] *sets [09:27:03] which are available at the given URI [09:27:07] so i think that's fine [09:27:17] it's quite redundant [09:27:21] oh, well [09:27:33] doesn't hurt [09:28:15] actually, i think we really need (meta)info about two document URIs: the canonical uri (latest), and the perma-uri (specific revision) [09:28:41] i havn't come up with a way to express their relationship nicely, but perhaps you got an idea... [09:29:00] New review: Tobias Gritschacher; "I don't see any harms here. But I'm not able to test it close to the production setup. So, we should..." [mediawiki/extensions/Wikibase] (master) C: 1; - https://gerrit.wikimedia.org/r/64274 [09:29:48] New review: Tobias Gritschacher; "btw. is there no bug for that?" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64274 [09:30:45] approved, jenkins should merge [09:31:41] Change merged: jenkins-bot; [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64538 [09:32:23] New review: Tobias Gritschacher; "@hoo: this is not an issue of this changeset. see aude's patch for the backend" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64083 [09:37:25] DanielK_WMDE: thanks [09:38:46] DanielK_WMDE: I do have an idea, but I am unconvinced of the need :) [09:39:32] any idea why www.wikidata.org/w/extensions/Wikibase/docs/ontology.owl is not visible? [09:39:47] but www.wikidata.org/w/extensions/Wikibase/docs/summaries.txt is? [09:40:02] filter on the file-extension? [09:41:36] probably [09:41:39] ask ops [09:41:47] k [10:11:55] New patchset: Aude; "(bug 48473) validate claim guid in api and return error" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/63787 [10:13:08] New review: Aude; "if https://gerrit.wikimedia.org/r/#/c/64881/ can be merged along with this, then okay." [mediawiki/extensions/Wikibase] (master) C: 1; - https://gerrit.wikimedia.org/r/64083 [10:25:35] New patchset: Daniel Kinzler; "Config setting for cache duration of entity data." [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64281 [10:29:02] DanielK_WMDE: Tobi_WMDE poke [10:29:23] can you please take a look at https://gerrit.wikimedia.org/r/#/c/64037/ and give me your opinion on how i'm injecting the settings there? [10:30:39] aude: seems fine [10:30:43] will do more reviews later [10:30:45] hmmm [10:30:46] lunch first [10:30:56] similar issue i'm having with the claim guid validation thing [10:31:03] the api needs to know a setting [10:31:25] * aude would like a consistent way of doing that [10:32:11] aude: since we don't control initialization, we can't use real injection. initializing in the constructor using the singleton is fine, i think [10:32:19] hmmm, ok [10:32:46] alternatively, set the field to null, provide a setter, and use a getter to access it. when the getter is called while the field is still null, it initializes it using the default instance [10:33:13] that's a bit more work, but may help to defer and avoid initialization [10:33:14] ok [10:33:27] New review: Tobias Gritschacher; "ClaimGuidValidator is now depending on repo." [mediawiki/extensions/Wikibase] (master) C: -1; - https://gerrit.wikimedia.org/r/63787 [10:33:29] i'll do the way i did in the client api thing for claim guid validation [10:33:37] sure [10:33:39] we can always adjust later if we decide a better way [10:34:16] one thing though: it's usually preferrable to inject individual settings instead of the settings array. but that's not a big deal, i think [10:34:33] and in some cases, the initiator can't know all settings used internally, so it has to be a Settings object65 [10:34:34] that's what jeroen prefers [10:34:51] yea... i see the point, but not always [10:37:55] New patchset: Daniel Kinzler; "Purging of cached output from Special:EntityData." [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64289 [10:49:46] New patchset: Aude; "(bug 48473) Validate claim guid in api and return error" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/63787 [10:50:52] New review: Aude; "recheck" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/63787 [10:52:38] New patchset: Aude; "(bug 48473) Validate claim guid in api and return error" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/63787 [10:57:30] Change merged: jenkins-bot; [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64274 [11:13:29] New patchset: Tobias Gritschacher; "(bug 48473) Validate claim guid in api and return error" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/63787 [11:14:08] New review: Tobias Gritschacher; "PS 13: added more testcases and removed ALL OF THE WHITESPACES" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/63787 [11:17:05] legoktm: ping [11:18:54] anyone want to review https://gerrit.wikimedia.org/r/#/c/64472/ ? :) [11:19:03] * aude prefer it not get stale [11:22:51] New patchset: Aude; "(bug 47726) Adds a link to corresponding Wikibase Item to action=info" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64644 [11:29:37] aude: q999$b90d9fd1-4df9-dc97-9b56-7947848a0e43 [11:41:43] New patchset: Aude; "(bug 48473) Validate claim guid in api and return error" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/63787 [11:55:03] aude, Denny_WMDE: i would really have liked to get https://gerrit.wikimedia.org/r/#/c/64281/ and https://gerrit.wikimedia.org/r/#/c/64289/ into wmf5 [11:55:30] aude, Denny_WMDE: without these, entity data will be cached in squids for 31 days with no way of purging it [11:55:33] * aude needs https://gerrit.wikimedia.org/r/#/c/64881/ [11:55:39] DanielK_WMDE: look in a sec [11:55:56] we'll have to tag for demo, but can branch after [11:56:10] Change merged: jenkins-bot; [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64083 [11:56:46] aude: looking at your patch [11:57:00] k [11:57:23] * aude reviewing multichill's patch first [11:57:51] Change merged: Denny Vrandecic; [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64289 [11:59:33] Change merged: Denny Vrandecic; [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64281 [12:00:04] meh, always this revision stuff. seriously, i am very much doubting this will be used anywhere [12:02:40] Change merged: jenkins-bot; [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64477 [12:03:50] Tobi_WMDE, aude: the error message looks much better, but i suggest to not hide it under "details". [12:04:04] true [12:04:12] it seems like whatever is in details is the message we really want to show. the generic "edit failed" stuff is just distracting and useless [12:04:12] can we have a bug for that [12:04:20] or do we? [12:04:35] we talked about it before... i thought we did [12:04:37] ask the ui folks :) [12:04:42] ok [12:05:03] * aude tagging datavalues and diff [12:05:12] merged your patch [12:05:16] k [12:06:07] Change merged: jenkins-bot; [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64881 [12:11:01] aude: how do i call the clientinfo module? action=query&meta=wbclientinfo doesn't seem to work... [12:11:47] ah, meta=wikibase [12:11:54] k [12:11:59] * aude tryign to tag [12:13:11] aude: there's something wrong with the clientinfo: [12:13:14] [12:13:31] i should be able to append the scriptpath to the base URL to get the full URL. [12:13:43] that doesn't work here, because both contain the /daniel/ part [12:14:28] i guess this is how we currently have this info in the settings.... [12:14:34] huh [12:14:35] but it feels inconsistent, like asking for trouble [12:15:09] i think the confusing bit is the repoUrl setting: that's not the base of the pathes [12:17:10] Change merged: jenkins-bot; [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64037 [12:23:18] hey, newbie here, is there any api python sdk for wikidata? [12:27:59] yupbank: the pywikipediabot framework has some wikidata support, but i don't know how well developed that is [12:29:11] yupbank: best ask on https://www.wikidata.org/wiki/Wikidata_talk:Bots, or someone who's active there [12:29:49] Abraham_WMDE, aude: demo time now? or do i have time to get myself a coffe? [12:30:23] hi johl [12:30:23] not much toay [12:30:25] today [12:30:28] hey. [12:30:33] time is the main thing [12:30:36] aude: not much time to get a coffe? [12:30:40] no [12:30:48] good :) [12:30:57] Have you had a look at this? http://matthewmueller.github.io/date/ [12:30:58] afk for 5 [12:31:02] demo time is over DanielK_WMDE [12:31:08] started 14:00 [12:31:12] didn't miss much [12:31:18] ah, right, sprint start [12:31:19] unless you have something to show [12:31:25] sorry i missed the demo, confused the times [12:31:43] no, rdf stuff is still incomplete [12:31:49] i'll branch little bit later [12:32:01] see if the claim guid validation thing can go in, though not essential [12:32:02] Abraham_WMDE: so, sprint start now? [12:32:14] not yet [12:36:40] DanielK_WMDE: thx :) [12:36:43] DanielK_WMDE: are you around? [12:39:46] Abraham_WMDE: now [12:41:41] aude: you were actually undoing all my changes of ps 13 in https://gerrit.wikimedia.org/r/#/c/63787/ [12:41:50] huh [12:41:54] ok [12:42:21] * aude try again [12:47:07] DanielK_WMDE: Denny will poke you about sprint planning [12:47:14] ok [12:53:31] DanielK_WMDE: any idea http://wikidata-dev-repo.wikimedia.de/wiki/Special:EntityData/Q1.nt why this is unsupported? [12:53:32] it used to work last week [12:54:47] Denny_WMDE: because i implemented a whitelist of formats; some were just too messy, i didn't really want people to request svg from easyrdf [12:55:04] Denny_WMDE: n3 and ttl will work. we can easily add nt [12:55:17] nt would be great [12:55:22] one sec [12:55:26] thx [12:55:57] the whitelist is a good idea, thx for that [13:02:10] New patchset: Daniel Kinzler; "Allow ntriples in Special:EntityData" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64924 [13:02:33] Denny_WMDE: there you go [13:06:42] Change merged: Denny Vrandecic; [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64924 [13:06:51] DanielK_WMDE: very much obliged, thank you [13:07:15] DanielK_WMDE: what you want to do for the next sprint? you already know, or need suggestions? :) [13:08:32] suggestions are welcome [13:08:51] the sprint will be a bit short with the hackathon and my free day on monday though [13:09:07] merge/redirect is something i would like to at least discuss [13:09:14] yea, a discussion item for that would be good [13:11:57] New patchset: Aude; "(bug 48473) Validate claim guid in api and return error" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/63787 [13:12:08] Tobi_WMDE: https://gerrit.wikimedia.org/r/#/c/63787/ [13:12:10] Denny_WMDE: job-queue based dispatch and coalescing via diff-merge are also still pending [13:12:32] made it more lenient about getting and removing claims [13:12:51] kept stricter validation for setting [13:13:12] is the diff merge usable already? [13:13:25] i would prefer "serializing claims in rdf export"? [13:14:37] dont know if the diff merge is ready. last i looked, it wadnt [13:14:50] claims i9n rdf would be good, yes. [13:15:05] i already have a draft for that [13:15:24] and some stuff stashed locally [13:15:40] let's get your stash empty then [13:15:56] and there are a few smaller random stuff that could need some love [13:16:02] #48260 [13:16:02] New patchset: Jeroen De Dauw; "Updated depenedencies in composer.json" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64810 [13:18:06] Denny_WMDE: oh, and i want to finish the content negotiation, of course [13:18:39] for the hackathon, i might need your assistance to work with ops to get the url redirects implemented [13:18:47] including for the entity data stuff [13:18:53] DanielK_WMDE: ^ [13:23:19] aude: sure [13:24:00] if i got it right, it's only one single redirect, from http://www.wikidata.org/entitiy/Q123 to http://www.wikidata.org/wiki/Special:EntityData/Q123 [13:24:09] code should be 303, i think [13:24:17] Denny_WMDE: --^ [13:24:38] http://www.wikidata.org/wiki/Special:EntityData/Q123 will then trigger another 303 to http://www.wikidata.org/wiki/Special:EntityData/Q123.rdf or whatever the Accept header likes [13:25:00] no http://www.wikidata.org/entity/Q60.rdf ? [13:25:02] hmm, that wouldnt be correct [13:25:16] * aude reads the specs again [13:25:39] entity/q123 should conneg with 303 to either Special:EntityData/Q123.rdf or wiki/Q123 [13:25:42] also goes to sepc [13:25:44] spec [13:25:48] correcting if necessary [13:25:48] aude: http://www.wikidata.org/entity/Q60.rdf isn't really valid, since the ".rdf" thing refers to the document, not the concept. but the redirect would/should work anyway. [13:26:02] DanielK_WMDE: sure [13:26:31] Denny_WMDE: "either"? when would it go to wiki/Q123? [13:26:40] it html is preferred? [13:26:51] Denny_WMDE: in any case, it's much easier to implement with two redirects. [13:27:13] the first one *might* be a rewrite and thus hidden. that's less robust, though [13:27:30] it's en.wikidata.org/wiki/New_York_City that goes to www.wikidata.org/wiki/Q60 (easy part) [13:28:00] New patchset: Tobias Gritschacher; "(hotfix) Fixed diff of claims with a time property" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64927 [13:28:06] Denny_WMDE: btw - should S:ED?id=Q123 redirect to S:ED?id=Q123&format=rdf or to S:ED/Q123.rdf? [13:28:11] it goes to wiki/Q123 when html is asked [13:28:17] isnt that the point of connegin? [13:28:19] Denny_WMDE: the difference being that S:ED/Q123.rdf is cached by squids, and may be stale [13:29:46] Denny_WMDE: to me, wiki/S:ED/Q123.html wouldn't be the same thing as wiki/Q123, because the latter has all the UI crap. It's not just an html representation of the data, but a *user* interface to the data [13:29:58] so i think a redirect would be inappropriate [13:37:08] i never know exactly waht redirect and rewrite etc mean [13:37:28] html is a user interface [13:37:51] if i ask for a representation of Q123 in HTML, then wiki/Q123 is absolutely fine [13:38:42] if i ask for the representation in rdf/xml, then wiki/S:ED/Q123.rdf is perfect [13:39:31] so I ask for entity/Q123 and accept HTML, a 303 to wiki/Q123 is the correct thing to do [13:39:50] needs a special case, but is easy enough to do [13:40:51] any request to entity/Q123 should be answered by a 303 (or 404), never by a 200 [13:40:58] but... what should happen if someone asks for wiki/S:ED/Q123.html directly? should it redirect to wiki/Q123 then? [13:41:29] undefined. I am fine with it not returning anything, or wiki/Q123 [13:41:49] i don't care - whatever is simpler [13:42:07] Denny_WMDE: rewrite vs redirect: a redirect is sent to the browser, the document's url changes. a rewrite is internal to the web server, the client doesn't notice. this can get tricky with relative pathes. [13:42:34] so any request to entity/Q123 should be redirects [13:42:38] we need entity/Q123 -> wiki/Special:EntitiyData/Q123 --> wiki/Special:EntitiyData/Q123.rdf [13:42:52] the step in the middle is needed because that's where the content neg happens [13:43:12] but the first arrow could be a rewrite, as long as we are sure to *always* reply to it with a redirect [13:43:33] it's simpler to just use two 303 redirects, though [13:44:03] but it sounds wrong [13:44:07] why? [13:44:27] also, the question how an id=Q123 should be treated is still open. [13:44:49] id=Q123 in what ase? [13:44:53] *case [13:45:09] Request for entity/Q123 , accept: RDF [13:45:11] wiki/Special:EntitiyData?id=Q123 [13:45:24] this should trigger a redirect [13:45:28] but to which url exactly? [13:45:29] answer should be: 303 See Other -> SED/Q123.rdf [13:45:44] ah, but then you go from an uncached URL to aq cached (and possibly stale) URL [13:46:05] (URLs with parameters are not cached by squid/varnish) [13:46:38] Denny_WMDE: https://gerrit.wikimedia.org/r/#/c/64927/ [13:47:11] Change merged: Denny Vrandecic; [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64927 [13:47:35] do we cache Special pages content? [13:47:41] New patchset: Aude; "(bug 48473) Validate claim guid in api and return error" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/63787 [13:47:44] via squid? [13:47:49] yes and no i think [13:47:51] aude: so, whats with the guid validator? [13:48:05] nevertheless, this is a very different issue [13:48:09] Tobi_WMDE: it does only basic checks for get / remove claims [13:48:20] aude: ok [13:48:20] in case there are invalid guids in the db [13:48:26] aude: good [13:48:28] for saving, it's full check [13:48:48] and I don't see how the two step redirect would improve on that [13:48:53] but accepts lower case letters like for references, as the ui does [13:49:07] aude: waiting for the selenium tests to finish.. [13:49:11] k [13:49:22] the Special:EntityID?id=Q123 can return whatever format you want, I would default to JSON, but that does not matter for the rest either [13:50:23] aude: and what's with the namespace thingy in the article-links of the error tooltip? [13:50:37] i think demo was not fully updated [13:50:45] works for me now [13:50:56] can't reproduce, etc [13:51:23] * aude has one tiny follow up for the tooltip [13:52:16] ok [13:52:23] just checking tests [13:53:59] ok, so we can branch now? [13:54:05] DanielK_WMDE: that fine with you? [13:56:26] Change merged: Tobias Gritschacher; [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/63787 [13:56:51] going to branch now [13:57:03] k [13:57:09] my minor fix is not critical [13:57:20] aude, Denny_WMDE: squid/varnish caches anything from the /wiki/ path with no parameters if it sends the right headers [13:57:26] whcih Special:EntityData does [13:57:34] we could prevent caching, but i don't think we want that [13:57:55] it will just say * in the bubble though the bubble shows duplicate messages for that anyway [13:58:07] aude, Denny_WMDE: remember that Q60 stayewd empty even after we fixed the flush bug? it was stuck in varnish. [13:58:43] we could either have a short cache on it or none [13:58:43] the single item export is not expensive [13:59:01] (compared to a normal wiki page) [13:59:03] even stuff like recent changes gets cached [13:59:06] as we dont have tempaltes etc. [13:59:09] New patchset: Jeroen De Dauw; "Add @throws to QueryInterface interface" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64929 [13:59:12] Denny_WMDE: aude merged the patch that controls the cache duration a few hours ago. [13:59:20] yes, that's fine [13:59:21] i'd suggest 1 hour or so [13:59:23] someone merged it [13:59:26] i did [13:59:31] oh, sorry :) [13:59:34] np [13:59:59] but this doesn't relate to the URL discussion [14:00:13] i want us to be on the same boat for this one [14:01:38] Denny_WMDE: ok, so never mind the id=Q123 form. about the double redirect: [14:01:57] it's easier to debug, and less prone to failure in cases something goes wrong [14:02:11] but there is no need for it [14:02:41] entity/Q123 rewrites to SED/Q123 redirects to (whatever, like SED/Q123.rdf or wiki/Q123) [14:03:03] e.g. if mediawiki catches an exception and wants to showq an error page (somethign that can always happen), serving that from /entity/Q123 (because of the rewrite) could look strange if the html uses any relative pathes [14:03:12] but then, mediawiki generally uses absolute pathes [14:03:23] so we can try the rewrite, until it breaks :) [14:03:35] New patchset: Aude; "Change EditEntity status failure from session-failure to sessionfailure" [mediawiki/extensions/Wikibase] (master) - https://gerrit.wikimedia.org/r/64930 [14:03:41] i don't see any issue with the double redirect. it's semantically sound. just a little bit of overhead [14:03:57] it's three HTTP requests until you get the data [14:04:00] ok, new branches should be there.. [14:04:04] people complain about having to do a second one [14:04:05] yay! [14:04:13] Tobi_WMDE: yay! [14:04:15] Denny_WMDE: *shrug* [14:04:18] someone else already created tags I noticed [14:04:20] cinqo di wikidata [14:04:23] * aude tagged [14:04:28] when? [14:04:34] during demo time [14:04:38]