[00:07:07] halfak: I put the screenshots but I can't see the second one [00:07:15] can you check if it's there correctly [00:07:24] Looks good to me. [00:07:31] But regretfully, the prediction says "true" [00:07:33] lolk [00:07:39] * Amir1 thanks halfak for user.is_anon [00:07:42] :))) [00:07:45] Maybe we should choose a different example. [00:07:47] +1 :) [00:07:51] Amir1, see also https://gist.github.com/halfak/307e9980e873db134c5c [00:07:59] I reimplemented the wikidata.py [00:08:04] I actually tried to find a better example and this one had 36& [00:08:10] *36% [00:08:25] halfak: Amir1 happy new year :) [00:08:27] but I don't know why ORES gave this too me [00:08:46] YuviPanda: hey, thank you! I hope you have a great year ahead :) [00:08:59] full of ORES and PAWS [00:09:07] :D [00:09:39] I think most of my use to ORES is kinda done now, tbh. [00:09:50] it needs MW folks and people building models and client side stuff now [00:10:23] YuviPanda, we are getting pretty close to stable operations environment, I agree. [00:10:41] Will need advice long term though for sure. [00:10:47] * YuviPanda nods [00:11:10] Amir1, thoughts on that gist? [00:12:06] sorry I was skimming the history to find something useful [00:12:35] I'll scan for a bit [00:13:00] halfak: It's okay [00:13:03] but you need to change [00:13:06] revision.user.seconds_since_registration [00:13:09] to log(revision.user.seconds_since_registration + 1) [00:13:09] Why? [00:13:15] Oh! of course [00:13:26] and delete line 88 [00:13:29] BTW, 286961313 looks like a nice good example. [00:13:49] I found 274298410 [00:13:53] let me check :) [00:14:22] your example is way better [00:14:55] http://ores.wmflabs.org/scores/wikidatawiki/reverted/287879745/ [00:15:04] That one might be good too. [00:15:13] I've got to run. [00:15:20] I'll be on tomorrow for our hack session though. [00:15:37] Amir1, anything you need before I run away? [00:15:47] no [00:15:49] have fun [00:15:51] happy new year [00:15:52] o/ [00:15:52] :) [00:15:59] WoOO 2016 [00:16:06] o/ [15:43:43] o/ [16:10:27] * halfak hacks on documentation [17:20:36] o/ halfak [17:20:43] o/ Amir1 [17:21:01] It's published [17:21:10] (he fixed the problem) [17:21:18] Great! [17:21:26] we should advertise it :D [17:22:44] +1 BRB [17:25:00] ok :) [17:41:39] * YuviPanda waves [17:43:56] o/ [17:44:03] * halfak is having a new bed delivered. [17:44:27] I just took apart my old box spring. I didn't realize how broken it was. [17:45:02] halfak: I read that as 'baby' [17:45:17] I'm considering getting a real bed (have been sleeping on a really nice airbed) [17:45:21] probably won't, though [17:45:30] * YuviPanda has a plan that currently involves 3months in India for summer [17:47:25] my github stream is also at 41d [17:47:45] halfak: I am writing/ have written https://github.com/yuvipanda/python-wdqs for querying wikidata query service [17:47:53] I wonder if I should move it to mediawiki-utilities [17:47:57] yeah! [17:48:04] mwdqs? [17:48:18] Wait. [17:48:31] Is it Wikidata query service or Wikibase query service? [17:48:39] Seems like it should be wikibase. [17:49:01] Bad names :\ [17:49:12] it's wikidata-query service [17:49:18] wikidata specific, I think [17:49:24] or rather [17:49:26] OK. I guess that makes sense. [17:49:32] you can set up your own with a wikidata-ish JSON dump [17:49:37] including with the wikidata dump [17:50:22] * halfak finishes the last bit of docs for the new refactor [17:50:37] It turns out that sphinx has a lot of counter-intuitive behaviors. [17:50:45] I shall also add docs to this thing, maybe tests (although tests are hard for something that's a small wrapper, heh) [17:51:06] yeah. API wrappers are definitely hard. :/ [17:51:18] YuviPanda, do you think it would make sense to use pywikibase? [17:51:31] hmm? [17:51:31] *hard to test [17:51:43] ah, no, wdqs is SPARQL and hence has nothing to do with wikibase [17:51:48] ptwikibase is a data type wrapper around wikibase stuff that makes it way easier to deal with. [17:51:55] But items come back, right? [17:51:59] Or properties [17:52:00] so you're dealing with SPARQL data types [17:52:19] you only deal with proprrties when writing the query [17:52:26] Gotcha. [17:52:28] *properties [17:52:34] Bit items? [17:52:35] the output is almost always tabular [17:52:36] *but [17:52:46] Gotcha. So not the item JSON [17:52:56] yeah [17:53:39] halfak: look at http://tinyurl.com/noaet5w [17:53:57] the output is just columnar (in JSON format, but not wikibase JSON) [17:56:54] Gotcha. Is it JSON lines or one big JSON blob with a list? [17:57:15] Oh I see. There are options. [17:57:42] halfak: yeah [17:57:57] halfak: but the underlying system supports one big JSON Blob and one big XML Blob [18:20:25] Up to 92% test coverage [18:23:50] halfak: when ae you arriving in SF? [18:23:50] And now we have full documentation converage for features. [18:23:57] YuviPanda, tomorrow afternoon. [18:23:59] nice [18:24:08] I've got dinner plans, but nothing after dinner yet. [18:24:23] I'll be staying in Berkeley with Stuart until allstaff. [18:24:36] but that's on BART, so I'll be mobile. :D [18:24:47] ^ User:Staeiou for the curious [18:25:43] So there's still a problem with the documentation for new revscoring. sphinx is not picking up a lot of the documentation and I still haven't figured out why I'll need to do some digging to figure it out. [18:26:05] I've got a run for a couple hours (getting lunch) [18:26:11] But I'll be back on later. [18:26:29] My plan is to implement the edit quality models with new revscoring and *maybe* figure what's up with sphinx. [18:26:30] o/ [21:13:24] halfak: tweeted, posted on facebook, etc. [21:16:19] * halfak gets back and goes to the social media. [21:40:46] https://twitter.com/halfak/status/683399634526404608 [23:32:33] \o/ our error messages are going to be so much better. [23:33:13] Rather than saying RevisionNotFound, we now differentiate between Text/User/Comment being deleted and not being able to find the revision. [23:33:23] We also report which revision we can't find. :) [23:33:28] The current or parent? [23:33:37] The user's last revision or the page's creation revision. [23:33:39] \o/ [23:35:38] * halfak kicks off a tuning report for enwiki/damaging using the new features [23:47:25] And we're extracting. This seems to be working well. [23:47:39] * halfak hopes for tuning report in the AM. [23:47:42] :) [23:47:55] Might do wikidata while I'm flying around tomorrow. [23:47:57] o/ [23:49:29] awesome [23:49:31] o/