[01:21:37] Teams come and go, Channels Live Forever! [01:22:11] https://archive.org/help/wayback_api.php [01:22:36] hmm so that gives a JSON response [01:22:40] there's quite a lot of prior art to read up on [01:22:43] I wonder if there's a way to make them do the redirect [01:22:54] the "memento protocol"? http://timetravel.mementoweb.org/ [01:23:08] someone implemented it for mediawiki [01:23:40] This page is subject to change frequently, please check back for the latest info. Updated on September, 24, 2013 [01:24:01] :D [01:24:32] it looks like we'd want our own endpoint after all [01:24:40] because the list of providers is long [01:24:55] it's not just IA, as I thought. see http://timetravel.mementoweb.org/about/ : archive.today, Archive-It, Bibliotheca Alexandrina Web Archive, DBpedia archive, Canadian Government Web Archive, Croatian Web Archive, Estonian Web Archive, Icelandic web archive, Internet Archive, Library of Congress Web Archive, NARA Web Archive, Portugese Web Archive, PRONI Web Archive, Slovenian Web Archive, Stanford Web Archive, UK Government W [01:24:55] eb Archive, UK Parliament's Web Archive, UK Web Archive, Web Archive Singapore, WebCite [01:25:31] but it could probably fit in 50 lines of code or fewer [01:25:50] hmm is that for the memento API? [01:26:00] since that's just an API spec... [01:26:02] * YuviPanda clicks [01:26:51] http://eresources.nlb.gov.sg/webarchives/ is just s'pore related stuff for exampl [01:27:15] also how do we pick which one to send to? [01:28:01] instead of redirecting the user to an archived copy, we could present a page similar to the one we show for ISBNs [01:28:07] (example: https://en.wikipedia.org/wiki/Special:BookSources/9780809315826 ) [01:28:29] that's a pretty good idea [01:28:35] Special:WebArchive [01:28:49] although that might be confusing actually [01:29:00] why should I care if I'm getting it from IA vs Singapore Archive... [01:29:34] "As a User I want to see link I clicked" [01:29:39] * YuviPanda throws story points around [01:29:56] * ori exchanges them for mingle points. [01:30:16] another approach (I think legoktm may have suggested this) is to redirect the user to a page that has a top navbar frame and a main frame [01:30:35] the main frame is the page as it exists in the waybackmachine (or some indication of failure to retrieve the page) [01:31:03] whereas the top frame explains that you have clicked on a dead link, and that we queried the waybackmachine for you, but that alternate providers exist [01:31:19] that sounds nice too [01:31:28] except for frames and I don't know how frames work on mobile [01:31:42] There are a few archive providers that have been banned from Wikipedia, I think archive.today is one of them. [01:31:44] you mean this mobile fad hasn't blown over? [01:31:56] * ori goes back into hibernation for another 5 years [01:31:58] But the other ones seem like good ideas to link to [01:32:18] but why? if IA doesn't have it we can probably link to them [01:32:25] but I suppose the IA's api should do that for us instead [01:33:08] https://en.wikipedia.org/wiki/Wikipedia:Using_archive.is [01:33:43] isn't that a paid for service with ads unrelated to IA? [01:33:44] i agree that (a) there should exist an API that takes a URL and optionally a timestamp and that redirects you to a mirror of that page as it was at the time you specified, and that (b) all things being equal, IA is a better entity to host such an endpoint than we are [01:34:19] +1 [01:34:41] yep [01:35:06] anyways it's horrible when you have a nice idea and then people just invite themselves to opine on it, so i'll let you guys drive :P [01:35:35] this is all legoktm's idea, I just went along for free chocolate and to whine about one particular thing >_> [01:35:55] Uh, when was there free chocolate?? [01:36:03] hahaha [01:36:32] there is a great yiddish word for this btw that has since made its way into english dictionaries [01:36:32] legoktm: remember when I thought the skeletal hand was candy and ate it and it was plastic? [01:36:34] https://en.wiktionary.org/wiki/kibitz#English [01:36:42] "To give unsolicited or unwanted advice or make unhelpful or idle comments, especially to someone playing a game. " [01:37:04] thus also: a kibitzer [01:37:07] Ohh, the cupcakes [01:38:05] ori: :D nice [01:38:06] ori: your unsolicited input is always appreciated :D [01:38:45] +1 [01:38:53] ori: did you ever read that paper I sent you? :) [01:38:55] talk transcript [01:39:05] oh gosh [01:39:07] * ori blushes [01:39:09] i did start [01:39:11] at one point [01:39:20] do you have the URL handy? [01:39:24] yes [01:39:43] ori: http://www.cs.virginia.edu/~robins/YouAndYourResearch.html [01:40:44] * ori reads [01:42:00] ori: <3 getDependencyGraph() [01:43:39] :) peter is going to build an on-wiki page inspector tool of some sort that would show performance data [01:43:44] and this might be integrated into it in some ay [01:43:47] Nice! [01:43:47] *way [01:44:27] he's looking for feedback / ideas btw -- https://phabricator.wikimedia.org/T117411 [02:05:09] eep, cert for https://semantic-mediawiki.org/ expired today [02:05:16] * ori emails yaron [14:36:04] AaronSchulz: it seems that link tables are broken in last wmf version, investigate please? https://phabricator.wikimedia.org/T117679 [17:03:10] ori: my hack^Wwork around for the json+bad uft-8 monolog bug is ready for your critical eye -- https://gerrit.wikimedia.org/r/#/c/251679/ [17:03:29] Feel free to tell me that I'm approaching this entirely wrong [17:03:43] (if you have a better suggestion that is) [17:05:26] legoktm: re: https://phabricator.wikimedia.org/T116786, the initial set of proposed events (here: https://github.com/d00rman/restevent/tree/basic-events/schemas), includes edit, delete, move, restore, and `revision visibility set', any suggestions for how i should go about obtaining the right event type? [17:07:19] legoktm: if this corresponds to rc_type, then we have edit, new, log, external, and categorize (and i'm unsure what each of those entails, beyond what the name suggests) [17:32:42] urandom: rc_type of edit => edit, for delete/move/restore/revdel, it will be rc_type log, and then you'll have to look at the log type and log action [17:32:59] legoktm: gotcha [17:33:12] saw that, but want to be make sure it wasn't an abuse of some kind [17:33:23] want to be make sure? damn. [17:33:53] went yoda there for a sec. [17:34:23] external is like the wikidata stuff [17:34:46] and categorize is the new category watchlist feature [17:47:12] bd808: wtf is going on with the old logs in https://phabricator.wikimedia.org/T115686#1792653 ? [17:47:32] Oct 30 logs in Nov 7/8? [17:49:18] legoktm: yuck. syslog buffering? [17:49:37] for over a week? [17:50:58] that file is syslog all the way down [17:51:11] hhvm -> local syslog -> remote syslog -> file [17:51:17] *I think