[00:23:35] to whom do I report: PHP fatal error in /usr/local/apache/common-local/php-1.18/extensions/CentralAuth/CentralAuth.i18n.php line 19586: [00:23:35] Allowed memory size of 125829120 bytes exhausted (tried to allocate 7864320 bytes) [00:23:59] occurs while trying to view all translations of a banner in CentralNotice [00:26:03] btw that is is on meta.wikimedia.org ^ [00:40:09] awjr: done anything for the fundraising triage? [00:40:25] nothing more than putting together a list of bugs [00:40:42] i was hoping to talk to you this week actually to see what else i should be doing to get ready [00:40:50] and to figure out how to publicize [00:40:54] thought you mentioned some publicity you wanted to do [00:41:09] awjr: foundation-l ftw! [00:41:11] :) [00:41:12] :) [00:41:13] :) [00:41:22] lol ok [00:42:09] awjr: seriously that depends on the bugs [00:42:46] I did have good success w/ posting to the VillagePump-like pages for the Wikibooks/Wikisource pages [00:43:19] but only a small amount of devs [00:49:04] interesting - i like that idea [00:53:26] oh wow, i guess oct 5 is coming up pretty quickly... [01:11:51] hexmode: is there anything else i should be doing in prep for the bug triage other than publicity? [01:12:33] awjr: if you have a list of bugs and some publicity, we're good :) [01:12:43] cool [01:12:54] and I did most of my publicity the saturday before [01:13:00] alright [01:13:03] and was satisfied [01:13:10] awesome [01:13:14] but you may be harder to satisfy than me [01:13:15] thanks! [01:13:17] ;) [01:13:20] np [01:13:30] i have a feeling getting people psyched about helping out with FR stuff is gonna be tricky [01:14:06] sure, but they might have some ideas that they want us to know about [01:14:17] who knows... worth a shot :) [01:14:53] definitely! [12:48:04] ^demon: how much do you know about search and OAIRepository? [12:48:22] <^demon> OAIRepository is an extension. [12:48:30] sure [12:48:36] <^demon> That's it. [12:48:40] <^demon> You've hit the end of my knowledge. [12:48:46] haha [12:48:48] heh [12:49:06] Reedy: do you know how it is involved in search? [12:49:19] I'm trying to figure this out... [12:49:24] I know very roughly [12:49:30] I suppose I need to ask Brion [12:49:30] OAI is like a feed of changes, or something [12:49:36] right, got that [12:49:41] And search uses that to reindex stuff after it's changed [12:49:42] <^demon> Oh, I also know that mysql search used to suck for < 3 chars, but then Brion fixed it and it doesn't anymore. [12:50:02] I gathered this from the fact that fatals in OAI stop the reindexing process, it gets stuck trying the same request over and voer [12:50:30] RoanKattouw_away: does OAI feed a raw lucene or our MWSearch'd lucene? [12:50:53] No clue [12:50:59] I think OAI is a generic format [12:51:26] RoanKattouw_away: thinking about that search bug from yesterday and trying to figure out if OAI would have to generate expanded wikitext or if MWSearch would [12:51:39] suppose it doesn't matter right now [16:36:33] hrm [16:47:04] hexmode, so what do we want to do with that etherpad from now? Start annotating it? [16:51:46] Reedy_: yes, but only if you have some thoughts on those bugs... what you're doing is what I wanted... [16:52:23] Reedy_: so when I get to it in a few min, I'll understand which ones have what concerns. [17:03:31] robla: http://etherpad.wikimedia.org/ShellSort if you have any input... Right now I'm just trying to get a first take from Reedy_ [17:03:33] Back later, dinner time [17:03:42] nom nom! [17:12:27] RoanKattouw_away: https://bugzilla.wikimedia.org/16655 -- you looked at this and then reedy made some changes... looks ok now? [17:13:57] I looked at it 2009 [17:14:06] So I have no idea what I saw back then, or what it's like now [17:14:08] heh [17:14:20] ok, didn't look at dates... sorry [18:28:58] Ryan_Lane: What's the timeline for the HTTPS depoy? [18:29:23] after the servers are installed, and configured [18:29:28] :) [18:29:45] I don't know when ma rk is doing that, so I don't really know [18:30:10] Hmm [18:30:55] He suggested he was gonna do it yesterday, but I guess that didn't happen [18:31:05] Just wondering whether you were planning to snatch me tomorrow [18:31:18] well, this part you can help much with [18:31:22] it's just changing the DNS [18:31:29] everything else is ready otherwise, right? [18:31:38] I guess we need to use the pr-cdb [18:31:45] and maybe that needs to be regenerated? [18:32:08] so let's plan on the morning tomorrow [18:32:10] Yes [18:32:13] OK [18:32:20] I expect I'll be the firefighter, if anything [18:32:38] *Ryan_Lane nods [20:10:35] hi [20:13:37] hi hashar [20:13:39] how's it going? [20:14:34] great :-) [20:14:40] my daughter is happily sleeping [20:15:17] :) [20:15:17] and I have a lot of paperwork ongoing so my mediawiki productivity is a bit low :/ [20:15:36] still have to fix changes I have done to CodeReview, but since we are probably moving to git ... I am not sure it is worth it. [20:15:47] how is the MediaWiki architecture book going on? [20:16:01] hashar: unfortunately not enough people have been contributing their thoughts [20:16:08] hashar: I need to ping a bunch of people today [20:16:26] last time I have read it, it was more focused on MW history [20:16:36] might need to write about the overall architecture [20:16:39] yeah [20:16:51] (something we are really lacking to attract new volunteers) [20:34:14] hashar: right now unfortunately all the questions are more history-y [20:34:39] hashar: I don't know how to structure the questions to ask people to give a short architectural overview without that sounding like a forbiddingly large task [20:35:14] isnt there an author for this book? [20:35:24] I mean, he is probably expecting specific points [20:35:53] something like : what is the overally workflow of a user request? [20:36:00] hashar: actually/....... no [20:36:04] ahh [20:36:08] hashar: the editor welcomes varying approaches [20:36:27] so you will have to establish a plan :-) [20:36:31] hashar: http://www.aosabook.org/ you can read the previous volume ahead of time [20:36:38] hashar: sigh, I am asking you for help! [20:38:57] well maybe if nobody is interested, it should be postponed for the next volume :-[ [20:39:26] hashar: 4 people != nobody. [20:39:48] hashar: I know that if I structure my questions right, I can get useful answers, so I am asking you for advice on that [20:41:28] sumanah: is http://www.mediawiki.org/wiki/MediaWiki_architecture_document/text the actual text? [20:41:55] <^demon|away> hashar: I think we'll probably do one last push of trunk CR between now and the git conversion. [20:41:58] not yet, hashar. https://secure.wikimedia.org/wikipedia/mediawiki/wiki/MediaWiki_architecture_document has links to the initial ideas, brainstorming stuff, etc [20:42:14] hashar: the bulk of the text that now exists is https://secure.wikimedia.org/wikipedia/mediawiki/wiki/MediaWiki_architecture_document/How_and_why [20:42:23] the collection page [20:42:44] ^demon|away: in the end, we will probably stop using the CodeReview extension and switch to gerrit [20:43:02] ^demon|away: so I should probably not invest anytime to finish up my inline commenting "feature" [20:43:06] hashar: won't be for another few months I believe. did you see the new roadmap? [20:43:25] nop [20:43:34] hashar: "what is the overall workflow of a user request" is good, thanks [20:43:49] sumanah: that one can be solved with a simple diagram [20:43:54] <^demon|away> hashar: We could just back it out, if it's not usable as it is. [20:44:47] hashar: http://www.mediawiki.org/wiki/Roadmap Erik wants to polish it a bit more before heralding it as The Way Things Are Going To Be [20:45:01] sumanah: another interesting thing is the user input sanitization. We have webRequest to grab parameters given by a user and make them safe. So the question could be: "how do you avoid code injection?" [20:45:35] ^demon|away: we could revert it for 1.19 for sure. [20:49:53] thanks for the ideas hashar [20:52:02] sumanah: last one for tonight: a quick note about the installer would be good. During the early days (ask Tim Starling), you add to run a shell script to install MW. Then it was made to just upload file and run the /config/ wizard. Lastly, it is now a complete wizard :) [20:52:15] :) [20:52:18] cool [20:52:48] sumanah: so the question could be : "how did you install MW? How did we improve it for external usage (aka non WMF) ? What is now (the new installer). [20:53:08] thx [20:53:30] lastly, the top question is: how do you handle billions of page views per days? (tip: HEAVY CACHING) [20:53:49] that is more or less like the "handling a user request workflow" question [20:54:10] the idea is to expose that everytime the workflow request data, there is a cache to speed it up [20:54:40] the first one being web caches (Squid, Varnish), then memcached for data/parser, then database query cache (not sure it is used) [20:54:54] <^demon|away> (If you're not caching it in 3 places, you're not caching enough!) [20:55:00] that could also lead to "scaling a top site" [20:56:04] so the plan could be: what is the workflow of a user request? How do you handle that workflow with billions of page views? (ie cache), how do you scale further ? (split queries accross multiples databases, partitions caches ...) [20:58:03] *sumanah adds more to https://secure.wikimedia.org/wikipedia/mediawiki/wiki/MediaWiki_architecture_document/Initial_ideas#Possible_key_points [20:58:51] honestly I just feel more discouraged now because I still don't know how to make these questions bitesize to attract developers to answer them. [20:58:58] these are good questions of course. [21:01:59] sumanah: maybe you can have some answered during the New Orleans hackaton? [21:02:40] hashar: you make a good point. this might annoy guillom because of the change to schedule but maybe it cannot be helped. [21:03:10] I wish I could help a bit more but it is not really going to happen before mid-october [21:03:21] I can try having a call with guillom if it can help him [21:03:44] I understand. hashar, that might indeed help - you mean he could interview you? [21:03:55] for exampl [21:04:20] I am not sure I am the best one for that but that could give a start [21:06:22] thanks, hashar, I'll ask him whether he's interested [21:06:37] I am writing him an email right now (in english, you are in cc) [21:06:49] ok, thanks! you can write in French, it's fine [21:06:56] I can read enough French to understand [21:07:16] hashar: I took French in high school so I know just enough to be dangerous as the saying goes [21:10:18] good to know :) [21:10:32] I think we will start a french working group including you and brion :-d [21:10:53] ah bien [21:11:00] je suis le tour eiffel [21:11:02] je dois brush-uper mon francais :D [21:11:09] ou sont mes ananas [21:11:13] bravo a tout les deux :-) [21:11:18] hehe [21:11:27] on y va! [21:11:42] toute suite! alors! [21:12:13] I'm hearing the Amelie music already [21:12:14] il faut changer le nom des fonction dans mediawiki :D wgRequest -> wgRequ??te [21:12:26] un grand canard a mang? mon ordinateur [21:12:28] *sumanah laughs [21:12:33] LOL [21:12:43] hashar, hey in javascript that would actually work ;) [21:13:05] sumanah: I have sent the email to guillom. Will wait for his reply and I will contact him on phone to arrange a meeting. [21:13:08] i think high school latin was the best for coming up with crazy things to translate [21:13:09] thank you hashar [21:13:27] especially since most of the words we knew were wildly inappropriate for conversational use [21:13:36] can only talk about farmers, legates, and sailors [21:13:48] latin helped me learn more french word or understand them. [21:14:22] brion: I'm remembering something Seth Schoen wrote about two hostages singing about the possibility of being ransomed [21:14:26] hashar: I once read PHP code written by a Frenchman, every SQL query was called $requete etc. [21:14:51] I have been reading such code for 4 years at work :/ [21:15:17] That was mildly interesting. Also, half of the PagedTiffHandler code comments used to be in German [21:15:32] I kept using english words for my variables wich some of my coworker did not understood. So had to translate my code (& comments) to french :d [21:15:39] Fortunately Daniel chose me to review it, so I could read them, but I did add notes like "why is this not in English" :D [21:15:41] pfft [21:16:00] RoanKattouw: dont you read german? [21:16:03] I quite deliberately always code exclusively in English, even when working with Dutch people [21:16:05] I do [21:16:12] My German is reasonable [21:16:43] I am probably going to take German lessons. Seems to be really useful for eastern europe [21:16:45] a large part of the post-fork LibreOffice code cleanup was translating old German comments from the original Hamburg team ;) [21:17:07] brion: aha! it was ancient *Greek* - the structure of its verbs "led to many British prep school students learning things like the present subjunctive first-person middle dual of "luo", no doubt so that, if two of them were ever captured by extremely ancient Greeks, they could sing a duet in which they discuss with their captors the possibility that they might ransom themselves." [21:17:25] haha [21:17:32] +1 [21:17:43] had to dig to http://web.archive.org/web/20080829233641/http://vitanuova.loyalty.org/NewsBruiser-2.6.1/nb.cgi/view/vitanuova/2001/04/09/2 [21:17:57] What the hell is a subjunctive middle dual?!? :P [21:18:10] RoanKattouw, it's a reason that ancient greek died out as a spoken language [21:18:14] hahaha [21:18:41] It's funny, I read something about how as languages develop, their grammar degrades and their vocabulary grows [21:18:55] brion: you might also enjoy http://web.archive.org/web/20060207071538/http://vitanuova.loyalty.org/NewsBruiser-2.6.1/nb.cgi/view/vitanuova/2006/01/19/2 .... and now back to writing email to inexplicable people [21:18:57] Or so linguists believe, due to what they've observed from rainforest languages etc [21:21:36] sumanah, bahahahaha [21:23:43] sumanah: great one :) [21:23:51] n@ai totally rocks [21:26:56] heading bed for now. Have fun! [21:48:18] sumanah: That blog post (A short address) is awesome [21:48:32] I don't speak Latin but I can usually understand most written Latin [21:48:49] RoanKattouw: Seth is awesome. He got me into open source, and into Ray Lynch & Dar Williams [21:49:18] Oh, same Seth Schoen? [21:49:32] yes [21:49:51] I owe a lot of who I am to him [21:50:54] *RoanKattouw reads his WP page [21:51:06] Oh HE's the guy that discovered the cold DRAM read attack? [21:51:30] Hah, Appelbaum too, of course [21:51:48] I am at the point where I recognize the names of these celebrities, but don't know who they are yet [21:52:41] RoanKattouw: he also I believe discovered something about the yellow dots in currency photocopies, or something like that. [21:53:09] What are you referring to here? [21:53:12] I didn't hear this story [21:54:03] RoanKattouw: http://www.nature.com/news/2005/051019/full/news051017-9.html