[10:13:51] Hi. I'm always getting a "badtoken" error when trying a block via API. I've tried with several csrf tokens and none of them works. A bug? [11:20:04] Hi, on the dutch wikipedia, after a question at the helpdesk it seems some edits are marked as patrolled, but the patrol-log doesn't show [11:20:21] it is on the artikel Friesland the question is about [11:28:49] akoopal: was the edit rolled back? [11:29:13] https://nl.wikipedia.org/w/index.php?title=Friesland&action=history [11:29:26] it is about the ip-edits [11:29:41] the first isn't rolled back, the second only self-rolled back [11:33:27] Well, AFAIK the only thing which gets an edit patrolled without log entry is rollback [11:39:50] WikiPage::commitRollback() patrols the edit before saving the new revision, so in theory a race condition is possible [11:45:06] akoopal: this is sort of expected, there are few options other than logging all rollback patrolling (there's a report for that) [11:48:21] Nemo_bis: but that is not the case here [11:55:50] akoopal: what do you mean? [11:56:34] Or, how do you know [11:56:45] see the history-log, not a rollback [11:56:53] only a revert of the ip-user [11:58:00] akoopal: see above, that doesn't prove that a rollback didn't happen [11:58:27] hmmm? You mean the rollback isn't in the history log then? [11:58:44] ahhh, the rollback failed then [11:59:41] My theory is that the rollback started, then the new IP edit was saved, then commitRollback() selected both edits, patrolled both, attempted to save the preceding revision which was identical to the last and didn't make it to the history [12:00:19] ok [12:00:28] and the first anon edit? [12:02:34] akoopal: first which? [12:03:23] ahh, never mind, that was rolled back as well [12:04:03] Yes, that's a normal case [12:04:57] The main question is whether rollback is actually supposed to rollback an edit by the same user performed after the rollback button was clicked... perhaps yes, too many edit conflicts with fast vandals otherwise [12:06:00] well, to be honest, and the rollback patrols should be in the patrollog [12:08:20] that's https://phabricator.wikimedia.org/T42624 [12:11:56] thanks [12:13:11] left a comment [12:13:18] Meh, I thought I had documented this feature several years ago [12:19:03] akoopal: if you make a patch to add the logging, I think it has good chances of being accepted :) the biggest concern possibe is the DB load of saving many log entries at once (maybe the patrol log would need a "digest" :p) [12:20:00] Nemo_bis: only a sys-admin, not really a coder, to affraid to dive into code like this [12:20:54] akoopal: from your past deeds I suspect you're being too humble ;) but ok [16:22:30] hello. Does wikimedia use openstack swift ? https://gdash.wikimedia.org/dashboards/swift.eqiad-prod/ [16:22:40] or is it some kind of experimentation ? [16:23:17] orzel: https://wikitech.wikimedia.org/wiki/Swift may answer the question [16:42:47] orzel: The TL;DR is yes we do use swift in the WMF production cluster. It is used for storage of media files and their associated thumbnails. [18:00:37] anybody knows when AccountMerge tool will be available? [18:03:47] "soon" [18:05:22] few months ago I heard that Winter will be "soon"; so that "soon" or maybe little "sooner"? [18:05:43] [[Tech]]; 203.111.224.65; [none]; https://meta.wikimedia.org/w/index.php?diff=12045911&oldid=12033553&rcid=6387504 [18:08:28] tufor: Winter is more like "never" than "soon". [18:08:46] like Los Angeles [18:09:08] Winter Is Not Coming™. [18:09:19] tufor: hi, ~2 weeks is probably my guess. [18:09:42] The world forgetting, by the world forgot. / Eternal sunshine of the spotless mind! [18:10:03] legoktm: thanks :) [18:10:03] gilles: let's try it :) [18:10:34] <^d> James_F: I always preferred Spring anyway. [18:12:08] Nemo_bis, bd808 : thanks [18:12:45] ^d: Eww. [18:12:49] the (promising) "ganglia" link leads to nothing though :( [18:12:55] so yes, I was saying that I thought of starting the "no blocking JS in the head" quest by looking at mobile web, which probably has less real dependencies on those top blocking modules than the desktop site [18:14:22] it shouldn't be too hard to review what's currently in there and what it does [18:14:38] gilles: sounds like a good idea to me [18:14:47] there seems to be low-hanging fruit in CSS already, which is what I looked at today [18:14:58] inlined images in the render-blocking head CSS [18:14:59] <^d> James_F: Or Fall :) [18:15:22] gilles: jdlrobson would be quite excited by that [18:17:24] [[Tech]]; Uğurkent; comment not related to the subject; https://meta.wikimedia.org/w/index.php?diff=12045993&oldid=12045911&rcid=6387521 [18:20:12] gilles: I changed the performance weekly meeting to be 50 minutes by default, is that OK? I enjoyed the last installment but I felt that 1:30 was a bit excessive [18:21:09] ori: sounds good, it's a better default and we can increase the length specifically for the meetings around the end of the quarter/quarterly reviews [18:22:11] "... a final decision between the two to be taken in late 2013" <- i got the weir feeling that's this page is kinda outdated ;-) [18:31:15] gilles: see https://phabricator.wikimedia.org/T95152 for the existing task on render-blocking stuff [20:31:10] ori: a couple of weeks ago you showed me some animated gifs that were recordings of our pageload reflows [20:31:13] what did you use to generate those [20:31:14] ? [20:52:33] gilles: quicktime [20:52:48] gilles: and https://gist.github.com/dergachev/4627207 [21:32:39] ori: oh ok, so manually? I was hoping for some magic script to generate those automatically :) which would allow a before/after comparison [21:33:31] even a static film strip-like series of images would do for that purpose [21:50:53] Is there any specific reason why "Winter Is Not Coming™" [21:52:54] Negative24: designers are fickle creatures [21:53:16] Negative24: also, the main leading force behind it (Brandon) left the Foundation [21:54:37] MatmaRex: It was a nice initiative even though the copies that I tried out were buggy [21:54:47] *very [21:55:24] Negative24: I must explain, Winter was an umbrella project [21:56:10] TarLocesilion: Well yeah but I was mostly referring to the most obvious of them all, static top bar [21:56:42] it was mostly just the redesign of the interface wasn't it? [21:58:39] I'm not a specialist, but yes, it was, I think [22:05:00] [[Tech]]; 37.131.1.226; /* c */ new section; https://meta.wikimedia.org/w/index.php?diff=12047911&oldid=12045993&rcid=6388171 [22:05:13] Negative24: but the fact WMF abandoned (delayed to unforeseeable future?) so-called Winter project doesn't indicate they abandoned all/most constituent sub-projects :) [22:05:47] TarLocesilion: yeah [22:05:50] [[Tech]]; Matiia; Undo revision 12047911 by [[Special:Contributions/37.131.1.226|37.131.1.226]] ([[User talk:37.131.1.226|talk]]); https://meta.wikimedia.org/w/index.php?diff=12047938&oldid=12047911&rcid=6388186 [22:07:15] afaik, they prefer changes based on research, which is the only good direction. [22:11:30] TarLocesilion: and something like Winter probably wouldn't benefit much. They would have to work quite hard to prevent confusion [22:13:20] huh, confusions. in UX, confusions happen every day. sometimes if you aim to develop sth, you have to make sb to be confused.