[07:05:39] any devs here? if so, could you pleeeaase take a look at https://phabricator.wikimedia.org/T115792 ? It's been months now, and it should be a very simple fix [08:03:25] sjoerddebruin: uhuh. when viewing wikidata in dutch? [08:03:28] nikki: thank you! [08:03:38] Lydia_WMDE: Good morning, yeah still. [08:04:04] Or didn't we deploy in the meantime? [08:04:29] hi btw :) i am back from traveling [08:04:43] sjoerddebruin: no deployment in the meantime. starting again tomorrow iirc [08:04:54] Okay, will keep a eye on it. [08:04:59] <3 [09:44:27] https://www.wikidata.org/wiki/Translations:Help:FAQ/36/en looks outdated [10:55:07] hello there! I added a new JS lib to the list of external tools: wikidata-lang, a small Qid->2-letters lang code dictionnary, quite handy to handle labels/descriptions fallback https://github.com/inventaire/wikidata-lang [10:56:47] Nice. [11:03:03] https://twitter.com/WikidataFacts <- anyone know who's running it? [11:10:48] Lydia_WMDE: nope [11:12:38] Maybe @ReaderMeter, first follower... [12:44:56] I was going to test the new math feature for wikidata but I had to realize that wikidata.beta was not updated http://deployment.wikimedia.beta.wmflabs.org/wiki/Special:Version but not on http://wikidata.beta.wmflabs.org/wiki/Special:Version [12:45:59] for testing we need 6c61ce1 of the math extension [12:46:40] moreover I have no permission to create new properties... http://wikidata.beta.wmflabs.org/wiki/Special:NewProperty [15:25:32] Lydia_WMDE: I don't know for the showcases items, no one seems really interested... [15:35:00] Thiemo Mättig (WMDE) ping [15:47:02] Math crown around? [15:47:06] crowd... [15:47:10] hi [15:48:07] I was guessing that the message key would be datatypes-type-math [15:48:12] but that did not work [15:48:23] neither locally nor on beta [15:49:12] datatypes-type-PT:math is at least translated to the correct key [15:49:13] It work. [15:49:35] does it work on your development instance? [15:49:43] It works right now. The message key ___IS___ datatypes-type-math and it ___WORKS___. Thats why I really, really do not understand what you are talking about. [15:51:03] what does ___IS___ mean? [15:52:59] from which file is this message http://wikidata.beta.wmflabs.org/wiki/MediaWiki:Datatypes-type-math/de generated [15:54:08] what "file"? [15:54:46] does this message comes from this change https://gerrit.wikimedia.org/r/#/c/263608/ [15:54:59] or from that one https://gerrit.wikimedia.org/r/#/c/263605/ [15:59:55] ... [16:00:05] Thiemo_WMDE: Anyhow feel free to submit the patch... [16:00:20] I just wanted to avoid that this makes things worse [16:00:44] When you changed the message key, you *introduced* a second message. This patch did not deleted the old one. Thats probably why it still works. [16:00:53] Otherwise it would have stopped working. [16:01:28] Also Raimond did not updated translatewiki, so it does not know about the now wrong message key "datatypes-type-PT:math". It still delivers the correct one. [16:01:46] Things are worse now. [16:02:03] Thiemo_WMDE: I just introduced that key, because that fixed my local testing environment [16:02:14] I do not believe that. [16:02:52] What do you see in your local environment when you call the page with ?uselang=qqx [16:02:59] Thiemo_WMDE: the good news is that you do not need to belive, you can convince yourself [16:03:37] I know what you mean. But I can not see your environment. I can not know whats wrong with it. [16:03:56] http://wikidata-math-de.wmflabs.org/wiki/Property:P4?uselang=qxx [16:04:18] http://wikidata-math-de.wmflabs.org/wiki/Property:P4?uselang=qqx [16:04:35] How does this end up there? [16:05:19] Why is this different on beta? http://wikidata.beta.wmflabs.org/wiki/Property:P59844?uselang=qqx [16:05:37] I do not know. I just wanted to share my experience with you. I'm not claiming to know anything better. [16:05:48] Oh, I see. [16:06:06] The Wikibase.git code is outdated on http://wikidata-math-de.wmflabs.org [16:06:28] Probably not updated for about a month, I guess. [16:07:14] These prefixes were introduced a few weeks ago. Your code oes not know about prefixes. That's why these message keys are generated. [16:07:18] What is the best way to select Items using SPARQL from query.wikidata.org ? '?s a wikibase:Property' seems to work but '?s a wikibase:Item' doesn't even though I can see it in the rdf on https://www.wikidata.org/wiki/Special:EntityData/Q42.ttl. [16:09:48] I have to go now. [16:10:27] Thiemo_WMDE: Finally I removed my -2 [17:57:40] wikibase gurus: in globe-coordinate data type, is it ok to have precision of "null"? [17:57:45] e.g. https://www.wikidata.org/wiki/Special:EntityData/Q5939824.json [17:57:53] is it normal data or a bug? [17:59:53] SMalyshev: not ok but we used to not enforce requiring it [18:00:50] aude: ok, I see. so what would you think be the best way to handle it when exporting to RDF? [18:01:12] because now it produces ""^^ and that's just wrong - "" is not decimal [18:01:22] i am not sure... [18:01:35] maybe have a default precision? [18:01:57] if i edit it, it probably gets some default [18:04:58] it would be better to have a bot fix these [18:14:13] physikerwelt____: around? [18:24:51] aude: bot would be nice, but RDF export probably has to handle it too... [18:33:34] SMalyshev: agree and agree [18:33:57] we used to also allow globe to be optional or any string [18:34:12] but bots fixed that + we disallow that now [19:56:19] physikerwelt____: Is there a way to run the Math extension w/o Restbase? [20:24:39] physikerwelt____: https://gerrit.wikimedia.org/r/263657 Please merge ASAP [20:26:46] hi [20:27:04] hey :) [20:27:15] I (with a ahck) managed to get it running [20:27:34] on chrome? [20:27:42] Chrome? [20:27:53] I didn't look at rendering [20:28:04] just making sure we don't break the Wikipedias tomorrow [20:28:11] that's why we can't deploy it tomorrow [20:28:44] which wikipedias? [20:28:49] All of them [20:29:02] ok what was the issue? [20:29:32] Well, datatypes only on Wikidata, but not on the client's breaks the clients hard [20:29:42] And Wikidata deploys are earlier [20:29:48] Tomorrow [20:29:57] while most Wikipedias only get the new code on Thurs [20:29:58] day [20:30:10] ah I see so you are in group1 [20:30:18] also it seems there are other open issues, but the most pressing one is making the Wikipedias fatal [20:30:38] uh where is wgMathEnableWikibaseDataType used? [20:30:44] I think merging https://gerrit.wikimedia.org/r/#/c/263657/1,publish is good in any case [20:30:52] jzerebecki: hm? [20:31:24] I'll merge that now [20:31:26] It is defined (going to be) in the extension and set to false (for now) via the production config. [20:31:43] oh ok it is not merged yet [20:32:02] Yeah, will take care of backporting everything [20:36:09] "but the most pressing one is making the Wikipedias fatal" what do you mean by that [20:36:19] * not fatal [20:36:44] Well, showing blank pages to Wikimedia error pages to Wikipedia users is about the worst thing that we can do [20:36:57] - blank pages [20:37:09] ok makes sense [20:38:07] how did you realize that this could happen? [20:38:21] Well, I saw the mail earlier today and then panicked [20:38:31] because that problem is well known (to me at least) [20:38:44] we always had to keep that in mind when rolling out new data types before [20:39:29] What a activity. <3 [20:41:07] hoo: I think you have to manually merge, Jeninks already started failing this afternoon [20:41:23] Ok for me [20:42:05] but you did not submit? [20:43:48] I wanted to give jenkins another chance... but force merged it now. [20:44:10] I think something was changed in the way how transactions are handled [20:44:22] this leads to failing tests [20:44:26] I have to look into that [20:44:45] (transactions in the database) [20:45:17] Git log includes/db/ is often a good starting point [20:45:21] (on core) [20:49:16] hoo: Do you still need anything from me... there is a upcoming paper deadline [20:49:38] I think we're done here, but it would be nice if you could hang around for another few minutes [20:49:42] until things are deployed [20:50:03] Sorry for the mess, but I just didn't know about the details until a few hours ago :S [20:51:09] finally I do unstand why people propsed to keep the new datatype in the wikibase repo [20:51:26] at least deployment problems sould like a very good argument [20:52:23] since all WMF deployed extension can now register new datatypes, one should cosider to make the process a little bit more robust, i.e. not fail if a datatype is unavilible [20:52:57] Yeah, there is a patch for that already [20:53:01] but it's not merged yet [20:56:10] https://gerrit.wikimedia.org/r/#/c/263660/ ^^ jenkins will fail again [20:56:21] Also on the branch? Yikes [20:56:27] Will wait for it anyway [20:59:28] https://integration.wikimedia.org/ci/job/mwext-testextension-zend/19531/console it's somehow related to the database tests and a geo table [21:10:10] physikerwelt____: Looks good now :) Thanks for your help. [21:10:27] It would be nice, if Math were usable without Restbase (at least in a very rudimentary way) [21:12:15] isn't that only used as a cache? [21:12:41] No... also for validation in some way [21:12:46] think that was mathoid before [21:12:56] Relevant task: https://phabricator.wikimedia.org/T121173 [21:31:26] hmm... we have some proofread pages on wikipedia :/ [21:32:00] :O [21:32:29] aude: You are talking about the ProofreadPage extension? [21:32:35] Tpt_: yeah [21:32:39] Yes, it's installed on some small Wikipedias [21:32:52] we don't restrict which badges can be used where, which is generally ok i guess [21:32:55] e.g. https://www.wikidata.org/wiki/Q21189843 [21:33:13] hoo: What exactly do you mean by usable? [21:34:11] fun [21:34:27] Save values, display them (no matter how rudimentary) [21:35:03] hoo: OK. That's a good idea, but would be a different task [21:35:14] the other task was more about not throwing an exception [21:36:58] We could fall back to source rendering mode. ie printing $\TeX$ [21:37:14] Better than nothing [21:39:09] didn't hovering over a badge used to show the name of the badge, not the q-id? [21:39:29] Q-ID is not so helpful... [21:39:35] exactly :) [21:41:10] hoo: do you have a change ID of the config change for me? [21:41:40] Not on my desktop atm [21:41:51] See gerrit or wait for me coming back [21:42:47] ... searching for config in gerrit did the trick [21:42:50] https://gerrit.wikimedia.org/r/#/c/263658/ [21:54:50] Ah, nice :) [22:12:12] aude: Any reason https://gerrit.wikimedia.org/r/#/c/263434/2 depends on the context change? [22:12:18] * context related change [22:13:56] Oh, I see WikibaseRepo::newRedirectCreationInteractor calls newEditFilterHookRunner [22:14:41] it's probably easier to review and see how the patches relate [22:15:00] Yeah, I probably wouldn't have notice the type mismatch otherwise [22:15:04] * noticed [22:16:52] aude: Looks good to me... do you want to amend or don't bother? [22:17:18] i can [22:18:27] then i'm going to look at the other projects sidebar patch [22:18:38] Awesome :) [22:26:03] Oh nice, we ahve a hooks.txt [22:33:28] hoo: i can create external-identifier properites on test.wikidata [22:33:51] oh [22:33:52] but the clients won't know about the data type yet (until thursday) [22:33:59] i think we made the code more robust against that so maybe ok [22:34:12] but we could explicitly define the data types in our config [22:34:20] aude: No, not yet [22:34:36] what, not yet? [22:34:39] I tested it for the Math thing earlier today and client's still die if they don't know a data type [22:34:44] oooo [22:34:52] Yes [22:35:01] I think Lydia_WMDE wanted it to not go live yet [22:35:03] but not sure [22:35:15] I mean it shouldn't be available [22:35:21] ok, i can add the config [22:36:54] aude: For reference https://gerrit.wikimedia.org/r/259001 would fix that [22:40:52] oh, it's stuck in gerrit :/ [22:41:13] yeah :S [22:51:59] aude: Think you ahve to create a feature flag, like I did for Math [22:52:24] ugh [22:53:04] I'll create a patch for removing the setting [22:53:08] it's unused and misleading [22:54:22] would be nicer if the setting was respected [22:54:43] Don't think so [22:54:49] why should we have two definitions? [22:55:36] one is what is available [22:56:02] Hm... I see [22:56:20] feature flags can work also [22:56:21] But taht would need to be properly implemented and documented [22:56:32] I think feature flags are maybe a little nice [22:56:40] just more ugly to have a feature flag for each type [22:56:44] that we introduce [22:57:39] Also true [22:58:05] We need a solution now... but we should probably open a bug for a mid term solution [23:11:48] aude: I'll be back in a bit (briefly). Can then review a patch for a feature swithc for external ids [23:12:05] Or something similar [23:12:17] ok