[01:00:04] I've been getting complaints from users about vandalism occurring on wikidata and then being shown on wikipedia via the mobile site [01:00:13] Why does the mobile site show stuff from wikidata where the desktop site does not? [01:12:38] Both instances I've seen so far today have involved protected pages, where this feature allowed someone to bypass protection [01:13:53] Actually I think I'm going to report this [01:15:54] Although this is similar to commons in some ways [01:15:55] Krenair: if the WP page is protected shouldn't the WD item be protected, too? [01:16:45] bearND: in these cases the WD item was not protected [01:16:59] I'm more concerned that it shows on mobile but not desktop [01:18:18] mobile web and apps show WD descriptions in various places (article header, search results, explore feed) [01:18:37] well this was mobile web, article header [01:28:48] I just looked through Wikidata RC and didn't have to go very far into logged-out edits to find non-reverted vandalism [16:14:01] fyi, the git reviewer notification bot is on the fritz again at the moment. i've just added reviewers to all my outstanding patches that didn't have them. [19:21:57] bmansurov: https://phabricator.wikimedia.org/T152458#2973755 yes, i was planning to work with jdlrobson after metrics [19:24:19] cool, nzr [19:36:07] jdlrobson: we haven't replaced the notif icon either. i think i sent that to you. the problem with splitting 1 user story with smaller tasks is things fall through the cracks :(but let's talk about it after metrics [21:47:41] jdlrobson: mind reviewing https://gerrit.wikimedia.org/r/#/c/334432/ real quick, please? [21:50:38] mobrovac: just +1ed although Pchelolo makes a good point it would probably be wise to be conservative in bumping that value [21:50:46] thnx jdlrobson [21:50:50] for instance max_pages is 5000 [21:50:56] yup yup, just replied to his concern [21:50:57] we may want to tweak that [21:51:13] jdlrobson: 5k * ~1k per page is not that big a deal [21:51:46] but it is if you think in terms of quality.. maybe if there's lots of noise we'll throw away older one's that are actually trending in favour of new ones which get no edits [21:51:50] jdlrobson: im also wondering if the other purge params should be adjusted, like the inactivity one [21:56:05] mobrovac: Probably. If you look at https://trending.wmflabs.org/ shows Australia open was last updated 8 hours ago, Australia day 10 hours ago. Weekipedia uses 1 week as the inactive time and a min speed of 0.1 so probably a good idea to up inactivity to 24hr. Speed can take care of most of the purging. On Weekipedia I also protect certain things that hit a [21:56:05] certain threshold (this service would not do that... so it would be interesting to see how the results differ.. might be more relevant.. might not). [21:56:51] yup, makes sense jdlrobson, will tweak that param too [21:57:52] mobrovac: as we up max_age and min_inactivity we should be wary that we're allowing older stuff to hang around which give max_pages may impact the quality (i have no idea how many pages get edited in a 24hr period) [21:58:14] We may find a lower threshold like 8 hours gives us more fresher interesting content. [21:58:42] jdlrobson: yup. once we reach the 24h window for replay and retention we can play with the other params to get the best ration [21:58:47] s/ration/ratio/