[14:22:02] greetings [14:57:44] \o [14:59:33] you're up early ;) [15:01:06] about an hour earlier than usual, was intending to help ejoseph with some mw-docker things [15:01:53] can't help being up anyways, liam think's it's hilarious that he wont sleep in past 6 [15:07:38] Thank the gods for coffee [15:08:24] :) [15:32:49] mpham: I had a look at turnilo (https://w.wiki/4unR) and it looks like ApiFeatureUsage gets around 100 req/day. This seems super low, but those requests might be super important to someone. [15:33:09] I added a few comments to T302638 [15:33:09] T302638: Sunset ApiFeatureUsage (TDMP) - https://phabricator.wikimedia.org/T302638 [15:41:23] thanks for checking [15:41:34] that does seem low, but isn't zero [15:45:18] what i would wonder about is how many edits those are from. It's not impossible that those 100 requests come from users that make 2% of all edits [15:46:49] I'm getting that from the sampled_128, so there is only 6 requests from the last 7 days in that dataset. Hard to get any conclusion from only that data [15:47:27] And with a number that low, multiplying by the sample rate is only a very rough approximation [15:47:52] ahh, yea with that kinda of sampling it could be just about anyting. I suppose all we know is it's low :) [15:48:32] yeah, but knowing it is low is probably sufficient in this case [15:48:53] we need to relate that to how important those few requests are, but the logs are not going to tell us [15:49:59] stick a banner on the special page and let them tell us if they care? [15:50:09] I mean, at the end of the day, it's not really even a search feature, and it costs us time and effort to constantly maintain. If I understand, it is only on our plate because it was done in a somewhat hacky way. I think if it's important enough to keep around, it should be done properly by an appropriate team [15:51:25] it's about as searchy as wdqs :P [15:51:44] i'm not necessarily a fan of maintaing it, but my understanding is it was important to bot authors and they make the majority of edits [15:52:38] (separate debate of quantity != quality for another day :P) [15:53:23] but i guess i worry because noone at wmf seems to have the job of making bots work, so they have historically depended on individual teams making things work [16:01:13] I'll skip the unmeeting, conflicting meeting [17:29:56] haha, i'd love for the search team to not have to worry about wdqs either [17:31:07] I think that, like commons, if bots are a high priority for wmf strategically, we should make sure it has proper focus/resources, rather than having a lot of teams touch little pieces of it at the expense of the work they're still expected to do [18:06:45] have a great weekend, y'all!