[06:57:34] Hi, I am trying to retrieve sizes of pages in a given category through pywikipediabot, my script is at http://python.codepad.org/gW2bAweN, Can someone help me with this [06:58:20] tuxnani: I think you want api.php?list=categorymembers or something. [07:01:13] tuxnani: https://en.wikipedia.org/w/api.php?action=query&generator=categorymembers&gcmtitle=Category:Physics&prop=revisions&rvprop=size&format=jsonfm [07:01:22] Elsie: But that api call doesnt have length of page [07:03:03] Elsie, how about having a unlimited query? [07:29:22] not possible [07:30:12] tuxnani: You can use the cmcontinue parameter. [07:30:18] You can specify a higher limit than 10. [07:30:20] I assume. [07:30:22] The size is there. [09:19:01] 呀呀呀呀呀呀呀呀呀呀 [09:19:24] 哈哈哈哈哈哈哈哈哈哈哈哈哈 [10:22:40] Elsie: just pointing you towards this [[Wikipedia:Administrators' noticeboard#Special:BrokenRedirects]], I know you probably know about this issue, but you would be the best one to reply to this [14:51:39] andre__: hey [14:52:20] andre__: I think I might still be on the default Cc list for betalabs/deployment-prep [14:52:24] that or labs, one of the two :) [15:06:48] paravoid, you're on default CC for "Wikimedia Labs > Infrastructure" (and that's the only one left) [15:06:58] remove me [15:09:46] done [15:20:56] andre__: thank you! [15:23:43] np [15:28:36] fuzzybot dos not replace usage... hm... but i can't see a backlog [16:58:09] legoktm: Re bug 57931, there are other ways to create an empty page. For example, substing a template or parser function that produces no output (like [[en:Template:void]]). [18:20:40] greg-g: So, I've reviewed briefly and determined that contrary to what was in my head, the code on mw.org is not very broken at all [18:21:04] awesome? [18:21:15] I could *either* leave it be and wait for the deploy on Thursday patiently, *or* push some updates that have been in testing for a while and we're pretty anxious to get out [18:21:31] I'm game either way, you can safely release the deploy window if you need it [18:25:21] greg-g: The latter may be complicated enough to require scap, though, so if that's a non-starter then so be it [18:26:23] marktraceur: let's wait, sounds reasonable to do [19:36:44] I already had a git repo on my local machine which has now turned into a Gerrit repo, and am trying to push the most recent commit into Gerrit using git-review... I thought I had added a remote for Gerrit such that git-review should work, but I'm having trouble. qchris do you have a moment? [19:37:29] Sure. [19:37:39] I'll pastebin [19:37:43] Ok. [19:38:53] qchris: http://pastebin.ca/2491201 [19:39:32] The git-review/Gerrit instructions basically say I should just be cloning directly from the Gerrit repo FIRST so that its remote is already all set up properly [19:40:04] I guess I could do that -- wipe out this repo, git clone from Gerrit, verify that git-review works, then add my other remote (GitHub) [19:40:36] Add a .gitreview and then git-review -s? [19:40:41] but if there's a way to just make my gitconfig say what it oughta say so that I can avoid that, it would be nice. qchris let me know if you need any more info to help me solve this [19:41:01] oh, do I not have one? I'll check. I've done git-review -s and it didn't give me any errors [19:41:26] brainwane: still reading your pastebin :-) [19:41:36] :) [19:42:07] It thinks you're in a rebase too... [19:42:15] yeah [19:42:36] weirdly, my repo does not have a .gitreview file in the root directory, yet "git-review -s" does not complain [19:43:50] Reedy: yeah, I guess it automatically tries to rebase my change off the master branch of the "gerrit" remote [19:45:00] I guess the "lazy" way is to just do a clean clone from gerrit and copy your changes over and push [19:45:03] brainwane: So your gerrit repo had a .gitreview file (see last commit on) https://git.wikimedia.org/log/mediawiki%2Ftools%2Fmissing-from-wikipedia/refs%2Fheads%2Fmaster [19:45:29] I NEED TO DO A GIT PULL duh [19:45:39] Maybe you ran the "git review" commands that gave no error on that part? [19:46:18] That would give you a .gitreview file. [19:46:40] I did a git pull [19:46:44] But in might merge unwanted things depending on how your repo currently looks. [19:46:47] Ah ok :-) [19:46:48] now "git review" actually works [19:46:49] thank you [19:46:54] I am sorry for bothering you [19:47:01] thanks Reedy, thanks qchris [19:49:25] (working on https://www.mediawiki.org/wiki/Gerrit/Advanced_usage#missing_Change-Id_in_commit_message now) [20:04:40] aaargh "missing change-ID". still working through https://www.mediawiki.org/wiki/Gerrit/Advanced_usage#missing_Change-Id_in_commit_message [20:06:14] brainwane: Look in 'git log' for commits that are missing a Change-Id: . Your hook might be correct but you may just need to trigger it (in the simple case where you only have one commit, git commit --amend does that) [20:09:39] $ git commit --amend 621f006 [20:09:39] error: pathspec '621f006' did not match any file(s) known to git. [20:10:34] brainwane: what are you trying to do? [20:10:47] modify the commit whose sha1 you pasted, i assume [20:11:01] right. I'm attempting to deal with https://www.mediawiki.org/wiki/Gerrit/Advanced_usage#missing_Change-Id_in_commit_message [20:11:06] git just told you that it has no *file* by the name of "621f006" :) [20:11:16] so, 621f006 is missing the change-id? [20:11:28] ahhhh [20:11:43] * brainwane was looking through man page for git-commit and the section on --amend, missed that part. [20:11:44] and it's not the top commit? [20:11:46] if you just commited something, and gerrit thrown you an error, you can try just git commit --amend [20:12:05] let's do an interactive rebase, then! *thunder* [20:12:26] there is one commit missing change-id - a merge commit 3 commits ago. [20:12:39] 621f00659148382f3911b5328fc416165731be29 is its sha [20:12:49] so, yeah, 621f006 [20:12:50] a merge commit? [20:12:53] yes [20:13:06] hmm. why do you have a merge commit? [20:13:15] gerrit-generated merge commits don't have change-ids [20:13:16] (I tried doing an interactive rebase and was asked "aren't you in the middle of an interactive rebase right now?") [20:13:22] unless it's your commit [20:13:24] huh. [20:13:28] aren't you? :P [20:13:54] i was going to suggest `git rebase -i 621f006^`, then change the "pick" to "edit" next to 621f006, save and that should be it [20:13:55] the offending merge commit is a merge from a remote [20:14:24] hmmmmmmm. [20:15:26] brainwane: you know, i have a bad feeling about this ;) [20:16:11] brainwane: can you copy-paste the first few lines of git log --oneline --graph --decorate? [20:16:35] * 03f4682 (HEAD) pep8 style fixes [20:16:35] * 868c601 lang shows up in result page [20:16:35] * 501af27 slightly better results msg [20:17:07] uh [20:17:12] are you not on any branch? :o [20:17:20] you're right! [20:17:24] I'm on no branch. good eye [20:17:26] maybe you are in fact in the middle of an interactive rebase [20:17:45] did git checkout master [20:17:46] now: [20:17:48] * 070e934 (HEAD, master) Merge branch 'master' of ssh://gerrit.wikimedia.org:29418/mediawiki/tools/missing-from-wikipedia [20:17:48] |\ [20:17:48] | * 54675e9 (gerrit/master) Add .gitreview [20:17:48] * | f4a9d40 README updates [20:17:49] * | 621f006 Merge branch 'master' of github.com:brainwane/missing-from-wikipedia [20:17:51] |\ \ [20:17:53] | |/ [20:17:55] | * 3246205 Merge pull request #4 from fureigh/master [20:17:57] | |\ [20:17:59] | | * a5ce87a webapp - PEP8 compliance. [20:18:03] | |/ [20:18:05] oookay [20:18:05] * | 03f4682 pep8 style fixes [20:18:07] |/ [20:18:09] * 868c601 lang shows up in result page [20:18:43] so, you want to submit this merge commit for review? [20:18:48] 070e934 (HEAD, master) Merge branch 'master' of ssh://gerrit.wikimedia.org:29418/mediawiki/tools/missing-from-wikipedia [20:19:26] if yes, i think it should be enough to just run `git commit --amend` and accept, that should run the hook and generate a change-id [20:19:32] basically what I really want is to just sync up the contents of https://github.com/brainwane/missing-from-wikipedia with the gerrit repo, and I am fine with completely rewriting history along the way [20:19:58] all right, I have now done git commit --amend [20:20:00] and accepted [20:20:04] let's see what git log says! [20:20:30] ah, there's still that commit 2 commits behind HEAD, 621f00659148382f3911b5328fc416165731be29, that does not have a CHANGEID [20:20:48] it looks like you already have some commits in the history of gerrit/master that were merge commits [20:20:55] (if i'm reading the graph correctly) [20:21:02] a GUI would really help here [20:21:15] ah, or maybe not [20:21:37] brainwane: i found these merge diagrams horribly confusing, especially for large projects. i don't use them, and i find i have no need for them. [20:21:45] not on the screen, and not in my mind. [20:21:52] how do I add a changeid to commit 621f00659148382f3911b5328fc416165731be29? [20:22:06] brainwane: i assume that you did a `git pull` in the past that generated these merge commits? [20:22:09] yeah [20:22:12] brainwane: git review should do so automatically when you commit. [20:22:24] run git review -s on your repo to set up the appropriate hook [20:22:48] DanielK_WMDE_: I did that, but only after I had already made this commit. So how do I now retroactively add a changeID to that commit? [20:22:51] DanielK_WMDE_: gitk is a good tool for visualizing merge graphs [20:22:57] (bebirchall is currently struggelign through a bothed git-review setup on #mediawiki) [20:22:59] +1 for gitk [20:23:08] brainwane: git commit --amend [20:23:13] then just save [20:23:14] brainwane: i was going to suggest `git rebase -i 621f006^`, then change the "pick" to "edit" next to 621f006, save and that should be it [20:23:18] brainwane: You do so by amending its commit summary through an interactive rebase. But if it's a merge commit, usually the solution is to rewrite the merge commit out of existence [20:23:19] that sould probably work [20:23:25] What MatmaRex said [20:23:28] but no idea what will happen if you have merge commits along the way [20:23:31] RoanKattouw: i know. i find them completely worthless :) [20:23:32] Except maybe with s/edit/reword/ [20:23:39] um yeah, i meant reword [20:23:48] git rebase -i just destroys merge commits [20:23:54] It even shows you by commenting them out [20:24:04] brainwane: we could also simply go with the 'tactical nuke' solution :) [20:24:19] brainwane: git reset --hard gerrit/master; git pull --rebase [20:24:24] actually, i tend to use cherry-pick to get out of a mess [20:24:44] checkout master, pull, create clean branch, then cerry-pick whatever i want on that branch [20:24:44] but okay, let's rebase like RoanKattouw said [20:24:55] got my out of any mess so far [20:24:58] probably less trouble as this point [20:25:12] also yeah, cherry-picks are nice [20:25:19] and this save-my-ass workflow is also pretty nice :D [20:25:26] (dear Konversation, why does everyone in this channel have the same color? it's confusing) [20:26:12] ok! I continued git rebase (git rebase --continue) and got Successfully rebased and updated refs/heads/master. [20:26:56] * MatmaRex has `git log --oneline --graph --decorate` aliased as `git tree`, by the way, highly recommend would alias again ++ [20:27:38] (for when i can't use gitk like a sane person would, and so that i can copy this incantation when i need to give it to people :D) [20:27:44] I see now that 14ee1d30647391e6d7a9f5d08ffdceb2e934502a does not have a ChangeID. I'll try that git rebase thing again. [20:27:49] And yes the merge commits have disappeared [20:27:53] from the git log [20:28:09] \o/ [20:28:24] brainwane: no, just do git commit --amend [20:28:27] and just save [20:28:34] that should add the change id [20:28:44] brainwane: Instead of 'edit', use 'reword'. That'll open a commit message editor. Just save without changes. Having "edited" the commit message should trigger the hook [20:28:59] brainwane: if it doesn't, the commit-msg hook isn't set up right [20:29:02] Or, alternatively, do what DanielK_WMDE_ says and use 'edit' in combination with commit --amend, that's equivalent [20:29:13] ('reword' is just a shortcut for that) [20:29:44] I hadn't seen "edit" *or* "reword" for "git commit --amend" before, will have to remember that. So, I went ahead and did a git rebase and it workedish [20:30:03] brainwane: 'edit' and 'reword' are only valid in the context of git rebase --interactive [20:30:08] my log is now * a587ec9 (HEAD, master) Add .gitreview [20:30:08] * 14ee1d3 README updates [20:30:08] * 03f4682 pep8 style fixes [20:30:08] * 868c601 lang shows up in result page [20:30:15] Sweet [20:30:21] Do all of those have Change-IDs? [20:30:23] let's see whether git-review works now! [20:30:30] yeah, except for qchris adding the .gitreview [20:30:50] but that's already in the repository, right? [20:31:05] yeah [20:31:07] But, "Add .gitreview" is the top one?! [20:31:08] but even so [20:31:10] yes [20:31:11] Oh well [20:31:12] weird, huh? [20:31:13] * qchris reads backlog [20:31:15] You just pushed a bunch of stuff [20:31:17] So clearly something worked [20:31:27] IT HAS NOW GONE THROUGH. Thank you and my apologies for the giant IRC spam [20:31:32] brainwane: btw, you may want defaultremote=origin in your .gitreview file, so you don't have to worry about keepign the pesky "gerrit" remote in sync [20:31:42] brainwane: congrats :) [20:32:11] * valhallasw just realized new git submodules do not get shipped to people who pull by default. [20:32:48] yay! [20:32:51] brainwane: :) [20:33:49] this is eery. aude, brainwane, RoanKattouw, valhallasw, and MatmaRex all get colored the same pale blue by konversation [20:33:58] this makes things rather hard to follow :/ [20:34:34] DanielK_WMDE_: use irrsi :) [20:34:54] naw [20:35:00] so, DanielK_WMDE_, I'm thinking about how I want to manage GitHub vs Gerrit and I sort of have not decided yet which one will really be the main repo. So until then I will probably want to manage both of the repositories sort of manually... I think [20:35:00] might try quassel, though [20:35:15] (both the remotes) [20:35:27] eek :) [20:35:39] brainwane: your Gerrit repo gets synced automatically to the wikimedia/ repo on github [20:35:45] ok! back to my learning project in which I painstakingly put together 1/100 of the functionality that Magnus could do in about 5 min while asleep [20:35:53] brainwane: https://github.com/wikimedia/mediawiki-tools-missing-from-wikipedia [20:36:12] brainwane: my take: gerrit is more powerful for reviews, and i like the review model better [20:36:18] but the ux stinks [20:36:53] github: nice & shiny, lots of extras, but a bit dumb [20:37:03] valhallasw: ah, hmm, right now the GitHub repo I mean is https://github.com/brainwane/missing-from-wikipedia . I could probably move stuff around.... [20:37:28] gerrit is not bad, imho and is/can improve [20:37:57] i should make a lit of pet peeves [20:37:59] DanielK_WMDE_: I would just wish there was an easy way to submit seperate patches that do not depend on eachother strongly [20:38:10] actually, we should make one at the office, and send it to chad some time [20:38:17] DanielK_WMDE_: my time outside the Wikimedia ivory tower has solidified my conviction that we have to engage with GitHub users on their own turf *in addition* to having our own review stuff we control. [20:38:24] DanielK_WMDE_: I now typically do git reset origin/master && git add -p && git review, but that's not really efficient [20:38:47] valhallasw: hm? i just checkout master and start from a fresh baseline [20:39:05] brainwane: I propose a sprint to write compelling short repository descriptions for all Wikimedia repos [20:39:26] ok, leaving the office. bbl [20:39:30] brainwane: the current default (which is unchanged for the vast majority of repos) is: [20:39:36] bye DanielK_WMDE_ [20:39:37] "Github mirror of PHP extension wmerrors - our actual code is hosted with Gerrit (please see https://www.mediawiki.org/wiki/Developer_access for contributing) https://gerrit.wikimedia.org" [20:39:47] ori-l: oh that hurts physically. [20:39:49] which basically makes you feel like you're looking at something epiphenomenal [20:40:03] it would be supererogatory for a person to contribute after seeing that [20:40:06] nonactual code [20:40:18] yeah [20:40:23] I gotta get back to this db stuff. thanks all [20:42:58] valhallasw: you can just submit the patches as depending on one another and then cherry-pick them on master from gerrit's interface. [20:45:33] MatmaRex: *hits head agains wall*. I have seen the button, but never stopped to wonder what it did. [20:45:41] MatmaRex: that's really helpful. Thanks. [20:52:01] bits.wikimedia seems to be having issues [20:52:29] (not getting CSS) [20:52:50] ori-l: ^ [20:52:55] Site is a little bit slow here. [20:53:03] Getting slower every minute. [20:53:12] ori ori [20:54:20] there was a brief peak but it seems to be subsiding [20:54:23] http://ganglia.wikimedia.org/latest/graph.php?r=hour&z=xlarge&c=Bits+application+servers+eqiad&m=cpu_report&s=descending&mc=2&g=load_report [20:54:28] i'll revert if it doesn't [20:54:52] sjoerddebruin: wctaiwan: still having issues? [20:55:00] Yes greg-g [20:55:14] yes [20:55:24] okay, reloaded and got it [20:55:27] but it's intermittent [20:55:43] it should be getting better [20:55:53] okay. [21:00:56] sjoerddebruin: wctaiwan ok, it really really should be back to normal now, is it for you? [21:01:05] Checking… [21:01:09] Yes, it’s better [21:01:10] seems to be [21:01:48] hi [21:02:20] hi ankry, wazzup. [21:02:31] problem with displaying scans on many pages in plwikisource appaered [21:02:49] is it temporary or you are doind sth?> [21:02:52] ankry: link? [21:03:04] eg https://pl.wikisource.org/w/index.php?title=Strona:PL_Franciszek_Karpi%C5%84ski_-_Pie%C5%9Bni_nabo%C5%BCne.djvu/03&action=edit [21:03:19] https://pl.wikisource.org/w/index.php?title=Strona:PL_Franciszek_Karpi%C5%84ski_-_Pie%C5%9Bni_nabo%C5%BCne.djvu/003&action=edit works [21:03:32] ain't seeing nothing [21:03:59] cache problem ? [21:04:15] not local browser page [21:04:28] not local browser cache [21:04:36] how long has it been like that? [21:04:43] 0,5h ? [21:04:47] or so [21:04:50] I'm not seeing any scans either. [21:05:18] ankry: before or after 17 minutes ago? [21:05:29] difficilt to say [21:06:37] OK. found: more than 1h ago [21:06:50] rather less than 2h ago [21:07:16] hrm [21:07:42] MatmaRex: are you familiar with Wikisource extensions / JS? [21:10:25] oh, well, it's clearly related to the deployments earlier [21:10:29] 19:59 logmsgbot: reedy rebuilt wikiversions.cdb and synchronized wikiversions files: wikisources back to 1.23wmf5 [21:10:29] 19:53 logmsgbot: reedy synchronized php-1.23wmf5/extensions/ProofreadPage/ [21:10:39] Reedy: what was that sync? [21:10:47] update to master [21:10:48] ori-l: nope, not at all :( [21:10:54] Tpt__ : is [21:11:07] i think [21:11:18] * ori-l looks at the diff [21:12:06] Tpt__: ^^ [21:12:36] Uncaught TypeError: Cannot read property '1' of null [21:12:49] it's this commit https://gerrit.wikimedia.org/r/#/c/98884/ [21:13:13] that was to fix hundreds of fatals [21:13:40] [03-Dec-2013 21:13:09] Fatal error: Call to undefined method WikitextContent::getHeader() at /usr/local/apache/common-local/php-1.23wmf5/extensions/ProofreadPage/includes/page/EditProofreadPagePage.php on line 137 [21:13:52] that's 30 seconds ago [21:13:58] Yeah [21:14:05] There's not hundreds of those though [21:14:24] And those were there beforehand anyway [21:14:33] I logged bugs for them last week or the week before [21:15:06] https://bugzilla.wikimedia.org/show_bug.cgi?id=57615 [21:15:07] https://bugzilla.wikimedia.org/show_bug.cgi?id=57616 [21:15:21] 98884 was to fix the flood fo https://bugzilla.wikimedia.org/show_bug.cgi?id=57942 [21:15:37] ah [21:16:00] ori-l: pages linked from index page: https://pl.wikisource.org/wiki/Indeks:Pie%C5%9Bni_nabo%C5%BCne_%28Karpi%C5%84ski%29 do not work, adding a page to index --> scan on it does not display. [21:16:05] ori-I: fix for 57615 and 57616 is waiting for review https://gerrit.wikimedia.org/r/#/c/97835/6 [21:16:29] It looks like the image isn't found [21:17:39] It's maybe because pages are members of a djvu file and their index page have not as name Index:NAME_OF_THE_DJVU [21:19:14] *It's the cause of the bug [21:20:49] ankry: Is there a lot of books in pl.wikisource that doesn't have as Index: page name the name of the DjVu scan file? [21:21:07] almos all [21:21:10] almost all [21:21:27] And it worked before? [21:21:32] yes. [21:21:57] few hours ago it was OK. [21:22:32] I didn't know that it was possible. I am going to make a quick fix. [21:24:16] Tpt__: we avoid .djvu in index page name as then links to previous/next page are(were?) not consistent with index page contents [21:25:03] Ok [21:25:32] Tpt__: and makes splitting index page into many ranges not convenient for most users [21:37:55] some links to "without text" pages (pagequality level="1") are shown red in index pages (eg. https://pl.wikisource.org/wiki/Strona:Materya%C5%82y_i_prace_Komisyi_J%C4%99zykowej_T.2.djvu/002) [21:46:28] ankry: It's an old known bug. I haven't found the cause yet. [21:46:40] :( [21:46:51] Reedy: Could you review https://gerrit.wikimedia.org/r/#/c/98968 ? It fix the bug with pl.wikisource? [21:47:16] ankry: I've made the change that fix the bug with scan image [22:20:29] anomie|away: sure, but thats different than creating a page with no input text :P [23:25:34] TimStarling: I hope the mass deletions at http://www.mediawiki.org/wiki/Special:Log/delete have not caused bigdelete-like problems (when I nuked the images by Piotrus, an error appeared saying the database transaction was restarted, although all deletions ultimately went through) because these are images, that when put together, involve a similar amount of [23:25:34] database transactions as bigdeletes. [23:26:44] can these uploads which werent reuploaded be movedto commons on server side? [23:27:13] poking users or using a bot might be not best solution [23:53:12] Krinkle: Do you have any opinion on https://gerrit.wikimedia.org/r/#/c/98869 [23:58:50] * JD|cloud wonders if lego could improve his abuse filter at https://www.mediawiki.org/wiki/Special:AbuseFilter/26 [23:59:39] * legoktm looks