[10:46:58] lunch [11:53:11] reminder that i'll be working from a train today. hopefully the wifi works, but i'll probably be low bandwidth even if it does [15:04:33] \o [15:17:54] meh, changing function names in an interface used in multiple repositories is annoying :S [16:37:37] dinner, back later [18:44:25] \o [19:30:56] weird python magic, apparently x[:0] = [1,2,3] is a valid way to prepend to a list [19:46:29] assignment to a slice? interesting. [19:48:49] I am trying to run the updater on a recently imported Wikidata dataset, and get this error when trying to run the "last update" query: https://paste.toolforge.org/view/raw/84047a51 [19:59:13] hare: hmm, i think that may be beyond me :( perhaps david will know something when hes working tomorrow [19:59:45] It's really weird, I only started to encounter this recently. It is always possible I have something set up wrong but I don't know what it would be. [20:02:02] hare: i see a few mentions of this in phabricator, although not exactly the same. https://phabricator.wikimedia.org/T215623 and https://phabricator.wikimedia.org/P17741 So at least it's been seen before :) [20:05:02] hare: maybe https://phabricator.wikimedia.org/T207133#4681278 [20:05:21] old, but the reported stack traces are from the same bits [20:05:30] suggests a bad import? [20:09:55] It's possible, I just wonder what made the import bad to begin with [20:11:29] hare: the suggestion there and in the T215623 ticket is that some urls were incorrect. Sadly i don't fully understand how all the url references in wdqs work :S [20:11:47] T215623: Problem with docker-compose2 (behind proxy) - https://phabricator.wikimedia.org/T215623 [20:11:50] That makes sense to me. I'm just trying to figure out what *I* did wrong [20:17:09] verify the concept uri's are correct? i think that has something to do with munging but not completely sure [20:19:03] That's a good idea, thank you [20:20:16] sadly while i can look at how our services run, actually knowing what blazegraph does is usually beyond me :P [20:23:25] for wikibase host is there a preference between wikidata.org and www.wikidata.org? [20:27:24] hmm, i think streaming updater has something about that, checking [20:28:02] hare: the references in streaming-updater all seem to use http://www.wikidata.org [20:28:09] iirc the http vs https is important [20:28:51] It's a little confusing because the website (at some point in the past) changed its URLs from www.wikidata.org to wikidata.org, but I guess the underlying semantics have not change (nor would I expect them to, trivially) [20:29:03] I know it's always been http, counter-intuitively [20:32:20] ...Yeah I think I know what I did. Munged in one environment, tried to load in another. That probably broke everything. [21:17:27] Anyways, I am now downloading the latest TTL dump (turns out the mirror I was using might not have been updated recently, which would've been another problem entirely). I hope doing everything in the same exact environment will avoid this problem again 😬 [21:17:31] Thank you [21:40:48] hope it works out! [22:10:40] how weird...the parser reports Category:1975_establishments_in_Tunisia on enwiki links to Category:1970s disestablishments in Tunisia, per ParserOutput::getLinks, but its not found in the html output [22:12:35] i haven't run strong numbers yet, but spot checking from a set of 6.6M enwiki pages that having incoming links but no matching page, many of them are categories of this nature