[09:57:41] <nikki>	 it seems that if you try to undo the last edit on an item, the edit summary now says you restored the old version rather than that you undid the edit by a particular user... does that mean that that undoing the most recent edit won't send a notification?
[09:59:59] <hoo>	 nikki: Works for me
[10:00:06] <hoo>	 are you sure you clicked the right link?
[10:00:20] <nikki>	 yes
[10:00:36] <hoo>	 weird
[10:00:45] <hoo>	 On wikidata.org?
[10:00:56] <hoo|away>	 Away for a bit, but I'll check back laer
[10:01:57] <nikki>	 yep, I just tried on https://www.wikidata.org/w/index.php?title=Q4115189&action=history making sure to click undo and not restore
[10:02:18] <sjoerddebruin>	 Isn't that the behaviour since they added the new way to enter summaries (always includes default one)
[10:36:22] <hoo>	 sjoerddebruin: No, shouldn't be
[10:36:40] <hoo>	 "wikibase-item-summary-restore"
[10:38:18] <sjoerddebruin>	 hm
[10:38:44] <hoo>	 ah
[10:38:52] <hoo>	 it happens when you undo the last revision
[10:38:52] <hoo>	 I see
[10:40:02] <nikki>	 I got a notification at least
[10:40:25] <nikki>	 which is good to know :)
[10:43:28] <hoo>	 nikki: sjoerddebruin: Filed https://phabricator.wikimedia.org/T147631
[10:43:32] <hoo>	 feel free to comment there
[10:43:41] * hoo heading for luch
[10:43:48] <nikki>	 thanks
[10:53:23] <sjoerddebruin>	 Still wondering why the quality of the suggestions decreases when adding more statements.
[11:02:11] <sjoerddebruin>	 Thiemo_WMDE: seems like most of our descriptions are too short to hit Google; however, the one of Michelle Obama seems to be okay https://www.google.nl/search?q=obama+site:wikidata.org
[11:03:04] <Thiemo_WMDE>	 sjoerddebruin: Yea. There is no guarantee Google will use these. Length might be a reason. Still the right thing to do for us, I believe.
[11:03:29] <Thiemo_WMDE>	 I want to give search engines as much hints as we can, and this is one of them.
[11:04:18] <sjoerddebruin>	 Would be great to add tags for Facebook and Twitter too, so they show up pretty there.
[11:12:12] <Thiemo_WMDE>	 What tags?
[11:13:47] <sjoerddebruin>	 hmm, i thought we've had a task for that
[11:14:23] <sjoerddebruin>	 For Facebook: Open Graph https://developers.facebook.com/docs/sharing/webmasters#markup
[11:15:22] <sjoerddebruin>	 Twitter: https://dev.twitter.com/cards/types/summary
[11:18:09] <Thiemo_WMDE>	 sjoerddebruin: This is cool, personally I did not know this. Can you please make sure we have a Wikidata specific ticket for this? Thanks!
[11:18:26] <sjoerddebruin>	 Found the MediaWiki-wide one: https://phabricator.wikimedia.org/T142243
[11:18:35] <sjoerddebruin>	 And here is the Wikidata one: https://phabricator.wikimedia.org/T51859
[11:19:22] <sjoerddebruin>	 This is also a great example why these tags are nice to have: https://phabricator.wikimedia.org/T142048#2541404
[11:55:56] <hoo>	 Thiemo_WMDE: Regarding my new SnakFormatter
[11:56:05] <hoo>	 I was thinking that through forth and back yesterday
[11:56:09] <Thiemo_WMDE>	 Hm.
[11:56:26] <hoo>	 if you want, I can make it a DataAccessPlainTextSnakFormatter or some such
[11:56:56] <hoo>	 I considered this yesterday, but then went for the general option as that's as cheap to implement
[11:57:47] <Thiemo_WMDE>	 We could merge the patches and continue working on the code.
[11:58:13] <Thiemo_WMDE>	 It's just that I have so many smaller issues, I wanted to discuss this first.
[11:58:37] <hoo>	 I can guess… I also had a very hard timing naming these yesterday
[11:58:39] <hoo>	 also the setting
[11:58:39] <Thiemo_WMDE>	 Main functionality is unused: the array parameter gets a trivial [ 'url' ].
[11:59:01] <Thiemo_WMDE>	 As said, I believe we will never use this generic class again.
[11:59:18] <hoo>	 Yeah, could be
[12:00:00] <Thiemo_WMDE>	 I believe I fully understand what it does. What about this: We merge the 2 patches, I add a patch on top (renames and such), you continue when I'm finished. Does this help? You can focus on other things then.
[12:00:28] <hoo>	 Yeah, we can do that
[12:00:41] <hoo>	 please keep in mind that this is a first step in the migration
[12:01:00] <hoo>	 we will soon rename the functionality to legacy or escaped plain text or something
[12:01:07] <hoo>	 but right now there's only on
[12:01:16] <hoo>	 e format, so this is fine for now
[12:09:41] <tarrow>	 Lydia_WMDE: odd request. Could you move Adam's UK to EU adaptor from his desk to a public area. I'm coming to the hack day tomorrow and I forgot mine... Thanks!
[12:18:19] <Lydia_WMDE>	 tarrow: lol sure
[12:18:35] <Lydia_WMDE>	 tarrow: i'll put it into the bar in our area
[13:46:50] <Thiemo_WMDE>	 hoo: I merged the DataAccessPlainTextSnakFormatter patches we talked about. Please continue. I will not have time to work on this now, but will look into this on Monday.
[13:49:24] <hoo>	 Thiemo_WMDE: Thanks a lot :)
[14:05:29] <tarrow>	 Lydia _WMDE:I've landed a little early; would it be alright to pop by this afternoon and pick it up/say hi?
[15:32:55] <hoo>	 Easy chocolate https://gerrit.wikimedia.org/r/314710 :P
[15:51:47] <multichill>	 sjoerddebruin: Andy seems to be acting even more dickish than usual....
[15:52:09] <multichill>	 Does he think that consensus==Andy likes it?
[15:53:03] <sjoerddebruin>	 He seems to have his own opinion about consensus.
[16:11:18] <hoo>	 sjoerddebruin: https://phabricator.wikimedia.org/T132839#2699674
[16:11:27] <hoo>	 Does that sound sane to you?
[16:11:32] <hoo>	 I put some thought into that this week
[16:11:50] <sjoerddebruin>	 Nice to see that you're actually working on this. :)
[16:12:07] <sjoerddebruin>	 Would be great to have a different average for qualifiers and references though.
[16:12:13] <sjoerddebruin>	 +too
[16:12:25] <hoo>	 Not yet… but if there are no strong objections, I might sit down for a bit and create a demo.
[16:13:15] <hoo>	 sjoerddebruin: These are very basic Property id => probabilities anyway
[16:13:20] <hoo>	 so there's nothing we could average there
[16:13:32] <sjoerddebruin>	 So then, if I use a range of properties in combination with a specific external ID, they will be showing up more on other items with the same external ID?
[16:13:42] <sjoerddebruin>	 external ID property*
[16:13:56] <hoo>	 Yes, a little
[16:14:03] <hoo>	 depending on how we do the weighting
[16:14:06] <hoo>	 if we do that at all
[16:14:18] <sjoerddebruin>	 At least we need to work on improvements.
[16:15:17] <hoo>	 yeah… my proposal is the most minimal change… we can probably keep almost all of the existing extension code
[16:16:51] <sjoerddebruin>	 I've talked about this before, but it seems like the average gets lower by the number of statements. First I thought it was the external ID's, but I see it on items with less of them too.
[16:17:54] <sjoerddebruin>	 even things like place of death disappear from the suggestions
[16:22:05] <hoo>	 well, the weight P1 => P2 is SUM(probability(P1 => P2)) / (number of statements on the Item)
[16:22:28] <hoo>	 * rating
[16:22:35] <sjoerddebruin>	 But it feels like things became worse recently.
[16:22:36] <hoo>	 So, yes… it actually decreases with the overall number of statements
[16:22:49] <hoo>	 could be, yes
[16:22:55] <hoo>	 we don't remove ext. ids there
[16:22:59] <hoo>	 so they still lover the rating
[16:23:15] <hoo>	 while not bringing more probability in
[16:23:23] <sjoerddebruin>	 Ah.
[16:23:26] <sjoerddebruin>	 It's annoying when adding VIAF identifiers, as it hides the last identifier I need to add for example.
[16:23:59] <sjoerddebruin>	 Oh, like we blacklist https://www.wikidata.org/wiki/Property:P143 but it would be much better if their use would be added to https://www.wikidata.org/wiki/Property:P248
[16:24:47] <hoo>	 That's five steps from now
[16:24:56] <sjoerddebruin>	 :P
[16:24:58] * sjoerddebruin mumbles https://phabricator.wikimedia.org/T77972 again
[16:25:01] <hoo>	 and the resources are limitted
[16:25:07] <hoo>	 but yeah… that would probably make more sens than just removing something from the results
[16:25:35] <sjoerddebruin>	 Could you do the P31/P279 thing for empty items for that? ^
[16:25:58] <hoo>	 This could be hacked rather easily
[16:26:08] <nikki>	 that would be awesome
[16:26:22] <sjoerddebruin>	 We say stuff like that every time, but nobody does it. :)
[16:26:27] <hoo>	 Indeed
[16:26:48] <hoo>	 especially on that extension you first want to refactor so many things
[16:26:52] <hoo>	 before going to that
[16:27:26] <multichill>	 hoo: Thanks for tagging me on https://phabricator.wikimedia.org/T147591
[16:27:44] <multichill>	 I've been playing around with wbformatvalue , but no complete code yet
[16:28:01] <multichill>	 Oh wait, that's the other one, wbparsevalue ;-)
[16:28:17] <hoo>	 multichill: I plan to heavily expand on format wiki
[16:28:28] <hoo>	 that's why I was on making sure I don't destroy anyone's tool/ template/ …
[16:28:44] <hoo>	 The old parser functions/ mw.wikibase.* things will stay the same
[16:29:43] <hoo>	 but the wiki format obtained by the API will change *heavily*
[17:23:03] <multichill>	 sjoerddebruin: http://tools.wmflabs.org/multichill/painters/index_combined_new.php?propa=650&propb=651 in case you get bored :P
[17:23:23] <sjoerddebruin>	 I'm never bored. :P
[18:00:42] <multichill>	 sjoerddebruin: And http://tools.wmflabs.org/multichill/painters/index_connect.php?propa=245&propb=650 to expand existing items with more external-id's
[18:00:59] <multichill>	 http://tools.wmflabs.org/multichill/painters/index_connect.php?propa=651&propb=650 is actually quite interesting
[18:11:08] <hoo>	 Amir1: https://gerrit.wikimedia.org/r/314710 trivial
[18:11:55] <Amir1>	 hoo: {{merged}]
[18:12:06] <hoo>	 Thanks :)
[18:24:40] <sjoerddebruin>	 multichill: http://www.narcis.nl/ looked interesting on first sight, but http://www.narcis.nl/terms/Language/nl made me sad.