[00:24:22] to show support to paris attack victims, the world federation of color blind people raised an irish flag [00:37:26] ++ [10:24:49] is there anything in phabricator about language fallback not working properly for statements and the search dropdown? I'm not finding anything but maybe I'm not searching for the right thing [10:27:43] I do get fallbacks on the history tab, special:search and special:usercontributions, but the same items are just showing as qids in statements and the search dropdown [10:30:52] Think that's a known problem [10:31:02] but can't look right now, got to go [11:05:50] Lydia_WMDE: patch is uploaded \o/ [11:05:54] * benestar is off now :) [11:06:26] benestar: \o/ [11:06:41] <3 [11:10:54] in case anyone is wondering which patch: the one to make merging of two items that link to each other harder [13:17:24] Jonas_WMDE: https://integration.wikimedia.org/ci/job/mwext-mw-selenium-composer/19/console [14:31:12] Lydia_WMDE: is there a ticket that Special:Merge and the widget should allow ignoring conflicts? [14:31:22] maybe after a conflict happened, the users should be able to resolve/ignore it [14:31:49] benestar: not yet. i'd like to get it live without for a bit and see if people really need it [14:32:02] k [14:33:23] jzerebecki: aude https://grafana.wikimedia.org/dashboard/db/wikidata-api-wbgetclaims [14:45:26] addshore: what did you do so I can't edit it anymore? [14:45:44] bah, I untickets a box.... and as far as I can tell there is no way to re tick it.... [14:45:51] I also can not edit them.... [14:46:18] heh nice [14:54:13] https://gerrit.wikimedia.org/r/250162 is getting old... although it's pretty trivial [15:15:58] addshore: pretty :) [15:16:43] would be nice to have formating for the property ids (e.g. "Commons category (P373)") [15:16:56] hahahhaaaaaaa, that would indeed be nice ;) [15:17:11] would also be nice to have an ordered list of use when you over over one of the bars! [15:17:17] or hover over the ids or in the graph and get the name [15:17:43] yeh, unfortunatly I'm basically 100% sure that is not possible in grafana ;) It would be possible in shiny though.. [15:17:48] :/ [15:18:08] would have to be a chrome extension or something :P [15:18:20] you might be able to write extensions for grafana [15:37:22] FYI https://github.com/wmde/grafana-dashboards [15:38:17] \o/ [15:38:32] addshore: btw you can define a grafana dashboard in puppet via grafana::dashboard [15:38:44] ooooooohhhh *goes to look* [15:56:34] Hi, I want to merge item Q20968093 to Q6389609 and the system won't allow it to me. Can somebody help to me? Thank you, Urbanecm [15:59:43] Urbanecm: are you using the merge gadget or special:mergeitems? the merge gadget should be able to merge them (you can enable it in your preferences if you're not using it already) [16:00:13] I am using Special:MergeItems. I'll try to merge it using the gadget. [16:01:57] Nikki: Thank you, it's merged. [16:02:02] :) [16:27:49] aude: can you remind me how your legacy entity usage data is stored? [16:42:07] addshore: mysql [16:42:11] YAY [16:42:18] it's pretty dumb [16:42:22] can I have access to it / a dump of it? [16:42:24] but works [16:42:37] thats literally so perfect :P [16:43:07] aude: just starting this one https://grafana-admin.wikimedia.org/dashboard/db/wikidata-entity-usage [16:44:53] https://tools.wmflabs.org/audetools/wikidatastats-20151117.sql [16:45:01] epic! many thanks! [16:45:02] \o/ [16:53:26] aude: in thats db I guess I just want the stats_usage_tracking table? [16:53:43] and where aspect = 'total' that used to be X ? [16:53:56] L.gsw used to be L_gsw ? [16:54:42] i think so [16:54:46] coolio! [16:55:08] should match http://tools.wmflabs.org/audetools/wikidata-stats/usage-tracking/2015-11-09 [16:56:32] *exports as a json* [16:56:44] *goes to write another crappy php script* :D [16:57:00] do you want json? [16:57:07] I already converted it :) [17:02:19] heh ok [17:08:43] ahh, total can be ignoered actually! :) [17:31:37] right, aude its all in graphite! [17:32:13] just got to wait for the rendering to catch up :P [17:32:24] any idea how many days worth of tracking you had there? [17:32:53] \o/ [17:33:05] 3 snapshots [17:33:24] coolio [17:33:30] roughly once a month [17:39:02] hmmmmm [17:44:09] aude: does this look reasonable? https://grafana.wikimedia.org/dashboard/db/wikidata-entity-usage I'm a bit confused by the dip... [17:47:05] ignore todays data (not finished adding it yet) but the other data looks a bit odd :p [18:02:18] addshore: it's somehat difficult so see what colors represent what :/ [18:03:09] yeh, I think I actually just found a bug in graphite though :P [18:03:39] *in grafana.... [18:03:49] This is the same part of the graph in graphite directly https://graphite.wikimedia.org/render/?width=1000&height=600&target=daily.wikidata.entity_usage.*.*&from=-85d&lineMode=connected&areaMode=stacked [18:04:22] i have trouble to see how the numbers match http://tools.wmflabs.org/audetools/wikidata-stats/usage-tracking/2015-11-09 [18:04:25] e.g. for enwiki [18:04:55] https://graphite.wikimedia.org/render/?width=1000&height=600&target=daily.wikidata.entity_usage.*.*&from=-85d&lineMode=connected&areaMode=stacked looks better [18:05:09] yeh [18:05:10] https://graphite.wikimedia.org/render/?width=1000&height=600&target=daily.wikidata.entity_usage.enwiki.*&from=-85d&lineMode=connected&areaMode=stacked [18:05:19] thats enwiki [18:05:43] so the data is being stored /perfectly/ and grafana is screwing something up (or I have told grafana to screw something up).... [18:05:50] hmmm, so it's a little less usage in november v. october :/ [18:06:34] but doesn't seem to match my numbers [18:06:51] no it does [18:07:01] todays numbers for graphite are still running [18:13:13] aude: I think I may have fixed it.... switching from null as null to null as zero seems to make it look right [18:13:28] im guessing the stacking gets confused with null values [18:14:42] ok [18:15:41] cool, I'll set this to run daily then! [18:15:54] \o/ [18:24:17] Lydia_WMDE: ! [18:24:43] addshore: ! [18:24:49] https://grafana.wikimedia.org/dashboard/db/wikidata-entity-usage (still needs work) [18:24:54] also still a bit confusing... [18:24:57] \o/ [18:25:58] addshore: seems to be missing at least meta, mediawiki and wikispecies? [18:26:05] hmm, S + T + L + O = X, that doesnt seem to be right [18:28:04] *goes to see how wrote that code* [18:28:32] oh.... O = X - S - T - L of course.... [18:29:08] DanielK_WMDE_: ^^ neither of these add up :/ [18:36:24] <__mek__> Howdy folks, what's the best resource / documentation for getting an instance of wikidata up and running. Is this process considered "MediaWiki" installation? (https://www.mediawiki.org/wiki/Manual:Installation_guide) [18:50:52] addshore: what do you mean by "add up"? Add up to what? [18:51:17] __mek__: you'd start with the mediawiki installation, yesw [18:52:23] __mek__: also, I assume you are trying to get an instance of *wikibase* up and running. The instructions are here: https://www.mediawiki.org/wiki/Wikibase/Installation [18:52:31] NO guarantee though, I havn't looked at these in ages [18:53:11] there were a few substantial edits made over this year, so it should be reasonably up to date [18:53:25] but we don't (yet) do stable releases, so ymmv [18:54:23] DanielK_WMDE_: well, X currently sums to 21.3 million for all sites. but L sums to 63 million and S to 46 million :P [18:54:31] addshore: S + T + L + O = X is not the correct logic at all... X means "we don't know what this is". We only use X if we do not have more specific information [18:54:45] DanielK_WMDE_: I thought thats what O was? [18:54:51] If we know that only the label is used, we only record L [18:55:06] no, O means we know that it's not S, T or L. [18:55:24] ahh okay, so then X is when we just have no ideA? [18:55:32] (currently, O means "statement") [18:55:40] yes, exactly [18:55:52] okay, I think that phpdoc could really do with updating then! [18:55:57] https://github.com/wikimedia/mediawiki-extensions-Wikibase/blob/master/client/includes/Usage/EntityUsage.php#L53 [18:56:10] at least, it confused me [18:58:11] addshore: ah, right. Think of that as logic operators on a bitfield. [18:58:20] then it is correct [18:58:23] ahhhhhhhhhhhh! [18:58:27] :D [18:58:32] yes, that makes so much more sense.... [18:58:52] addshore: make a patch that clarifies this, i'll be happy to merge it. [19:01:58] <__mek__> Thanks DanielK_WMDE_ ! [19:07:23] DanielK_WMDE_: query for those task that were in the last and are not in the current sprint: https://phabricator.wikimedia.org/project/sprint/board/1594/query/rVzwH7yJ9_xE/ [19:11:14] jzerebecki: thanks. some of them can probably be dropped, but I'll try to go over them with lydia tomorrow [22:31:53] aude: After https://gerrit.wikimedia.org/r/252267 has been pushed tomorrow, we should verify and increase the dispatching resources [22:32:22] I'll do that tomorrow, if I find the time... just FYI [22:36:22] But I'm happy that bots take care of the lag by now [22:37:21] hoo: +1 [22:40:53] aude: ? do you have an idea when https://phabricator.wikimedia.org/T117524 can be resolved? [22:41:09] seems like you are the one to ask :) [22:42:06] Harmonia_Amanda: i'll have to ask Lydia_WMDE [22:42:27] ok, thank you [22:42:43] seems there's a bit of interest :) [22:42:49] yes [22:43:03] so maybe we can schedule this, but i have to check with lydia [22:43:04] the french wikinews really want to try wikidata [22:43:31] :) [22:43:38] :) [23:02:14] aude: there is something special to do for Wikidata team about this issue? If not, we can also deploy that as a regular config change during a SWAT deploy window. [23:10:40] hi, where can I find documentation regarding the SQL table structure of wikidata? [23:11:44] Andyrom75: I can give you some hints, but be aware that only some secondary (meta) information is stored in MySQL (or rather MariaDB) [23:11:50] Andyrom75: https://www.wikidata.org/wiki/Wikidata:Database_download#Database_schema [23:11:53] if you want the actual data, you wont find it there [23:12:15] It might be better, if you tell us your use case very briefly [23:13:34] hoo: I'm trying to make some querys through some web tool, but unfortunately I don't know the field in table X that store the ID field of table Y [23:13:56] I need such information in order to structure properly the querys [23:14:00] Andyrom75: What kind of queries and what kind of Ids do you already have? [23:14:24] more or less nothing :-P [23:14:33] All I can see is through SHOW and DESCRIBE [23:14:42] You're on tool labs? [23:15:10] I should have an account, but I'm using a web tool to interact with the DB [23:15:29] I still don't know what kind of data you're looking for [23:15:54] Statements? Sitelinks? Terms (labels, descriptions, aliases)? [23:16:07] just to give you an example [23:16:12] You probably rather want to look at https://query.wikidata.org/ which has all data and can be queried via SPARQL [23:16:15] ok [23:16:46] I've seen SPARQL few minutes ago but I'm more confortable with standard SQL [23:16:52] so I'm using Quarry [23:16:53] I see [23:17:21] but as told, we don't have most of the data queryable via SQL, so most probably you wont find what you're looking for [23:18:12] mmm I thought that SQL could access to anything, however I think that what I was looking for should be quite easy [23:19:08] To be concrete, what you can find in SQL: Sitelinks, terms (labels, descriptions, aliases) and meta information about edits, users etc. [23:19:30] should be enough [23:19:46] Just to give you an example, I'd like to extract all the article that missing on a language version of a wiki-project, but that is present on the other language versions, ordering the result for the count of such languages [23:20:15] what I miss in this case is the link between the table that contains the various information [23:20:20] That should be doable (if you manage to write a fast enough query) [23:20:34] Look at the wb_items_per_site table [23:20:43] I'll run it on Wikivoyage, that has a small db [23:21:52] Yeah, guess that should work [23:23:12] I've seen that table. There's a field that contains the Qxxxx info? [23:23:27] or should I arrive there from another table? [23:23:30] yeah wb_items_per_site.ips_item_id = wb_entity_per_page.epp_entity_id [23:23:39] where epp_entity_type = "item" [23:23:48] wb_entity_per_page.epp_entity_type that is [23:24:12] ips_item_id is the numeric item id (the part after the Q) [23:24:16] not sure what exactly you need [23:24:38] let me try... [23:33:26] one step further... by now thanks for tip [23:34:41] joining these two tables (wb_items_per_site, wb_entity_per_page) there a field (~ID) that put in common two articles of a different language version? [23:35:20] is it maybe ips_item_id? [23:35:44] I mean, the same ips_item_id for different ips_site_id? [23:36:58] just tried... I've answered myself :-P [23:38:02] I don't fully get your question, but seems you've figured yourself :) [23:38:48] Away for a moment