[00:00:03] which AntiSpamMeta also works off of [00:00:41] Someone just deleted mine... [00:00:55] don't do that [00:01:00] it's annoying, funnily enough [00:01:12] What is? [00:01:18] then only list ops that care [00:01:41] !morbo | Annemarie [00:01:41] --mwbot-- Annemarie: http://img514.imageshack.us/img514/330/morbomediawikiaf9.jpg [00:01:55] !this_is_intentionally_null [00:01:55] --mwbot-- [00:02:18] I'm gonna alias it to that. [00:02:46] 03(NEW) 'Try Beta' needs a dismiss/hide option - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=20152 minor; Normal; MediaWiki extensions: UsabilityInitiative; (overlordq) [00:02:58] Done and done. [00:04:10] i was beginning to wonder how long until someone asked for that [00:04:40] 03(mod) 'Try Beta' needs a dismiss/hide option - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=20152 +comment (10agarrett) [00:07:42] I can't find a bug for global gadgets. [00:07:46] Surely there's one somewhere... [00:07:49] Anyone know? [00:09:31] <^demon> Annemarie: If it exists, it's not filed under gadgets, and hasn't been resolved LATER, INVALID or WONTFIX. [00:10:20] Hmm, okay. Thanks for looking. [00:10:23] I'll go file one. [00:15:34] 03(mod) 'Try Beta' needs a dismiss/hide option - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=20152 (10overlordq) [00:17:55] 03(NEW) Implement global gadgets - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=20153 enhancement; Normal; MediaWiki extensions: Gadgets; (public) [00:19:58] Annemarie is a global gadget [00:20:49] he's something... [00:21:00] 14(INVALID) Having both lang and xml:lang attributes not identical to $wgContLanguageCode - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=5790 +comment (10innocentkiller) [00:21:25] MrZ-man is a maintenance script. [00:21:26] 03(NEW) Add rollback permission for editors on Polish Wikipedia (pl.wikipedia) - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=20154 enhancement; Normal; Wikimedia: Site requests; (danny.leinad) [00:21:40] <^demon> Annemarie: 14950 is definitely a blocker. [00:21:45] 03(mod) Implement global gadgets - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=20153 (10innocentkiller) [00:21:48] 03(mod) Global preferences - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=14950 (10innocentkiller) [00:21:51] K. [00:23:41] 03(mod) 'Try Beta' needs a dismiss/hide option - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=20152 +comment (10public) [00:33:01] 03demon * r54692 10/trunk/phase3/ (includes/DefaultSettings.php skins/common/shared.css): Fix some misnamed IDs from the AllMessages refactor. [00:35:38] 03(FIXED) Colours in the mediawiki Allmessages page - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=7006 +comment (10innocentkiller) [00:38:13] https://bugzilla.wikimedia.org/show_bug.cgi?id=7006#c11 Haha, <3 Nikerabbit. [00:39:19] is there any way to see the visit statistics for a page over time? or does mediawiki just keep the total visitor count by default, without timestamps? [00:39:48] The latter [00:40:18] nanotube_: I would suggest a service like Google Analytics for over time data. [00:40:26] You can break it down per page. [00:52:29] 03(NEW) Enable flood flag on Polish Wikipedia (pl.wikipedia) - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=20155 enhancement; Normal; Wikimedia: Site requests; (danny.leinad) [01:03:30] GreenReaper: hmm, how easy is it to stick that into a mediawiki install? [01:20:51] Guest720: There's an extension that does it, GoogleAnalytics [01:21:11] http://www.mediawiki.org/wiki/Extension:Google_Analytics [01:21:18] or http://www.mediawiki.org/wiki/Extension:Google_Analytics_Integration even [01:30:19] *Simetrical extensions, feh, uses the skinAfterBottomScripts hook [01:33:08] Wow, merging changes selectively in SVN is a PITA. [01:33:35] MessagesHil.php MessagesLij.php MessagesSco.php MessagesVep.php [01:33:52] Are people just making up these message names or what? :P [01:33:57] Er, language codes. [01:43:01] 03simetrical * r54693 10/trunk/phase3/ (RELEASE-NOTES includes/Linker.php): [01:43:01] Don't use summary attribute in TOCs [01:43:01] The summary attribute is obsolete in HTML 5. It wasn't serving any [01:43:01] useful purpose anyway, since it duplicated the h2 inside the table. (Of [01:43:01] course, we should really stop using for TOCs altogether!) [01:45:50] 03(NEW) Make the voter list searchable/filterable - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=20156 enhancement; Normal; MediaWiki extensions: SecurePoll; (Emufarmers) [01:46:25] <^demon> Simetrical: I've tried removing that table before...it's really hard to get it displaying just right :-\ [01:46:41] I know, shrink-to-fit is a pain in CSS if you have to support IE < 8. [01:54:26] 03mrzman * r54694 10/trunk/extensions/IndexFunction/SpecialIndex.php: revert r54605 per comments on CodeReview [01:55:45] Simetrical: ^ [02:00:25] GreenReaper: thanks for the nfo :) [02:01:04] np [02:06:34] 03simetrical * r54695 10/trunk/phase3/ (RELEASE-NOTES includes/OutputPage.php includes/Skin.php): [02:06:34] Remove extraneous type="" for CSS/JS in HTML 5 [02:06:34] Removed in a bunch of places, but of course there are zillions left! [02:06:34] This is one of the things that would be a lot easier if we could just [02:06:34] drop support for non-HTML 5 output. [02:17:40] 03mrzman * r54696 10/trunk/extensions/IndexFunction/IndexFunction_body.php: Fix bug in which indexes wouldn't be deleted if all of them were removed from the page [02:30:27] Simetrical: and fyi, the TOC changes need parser test updates [03:04:35] 04(REOPENED) New toolbar breaks wikEd - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=20134 +comment (10oq9pscb02) [03:51:51] 14(INVALID) Enable captcha for anonymous page creation on Spanish Wikipedia - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=8668 (10Phoenixoverride) [04:39:40] has anyone here done much indexing of dump files for offline querying? [04:42:04] i guess not then (-: [04:51:21] I guess yes but not if you leave. [04:54:36] i just wanted to swap a few ideas if anyone else has done this type of thing [04:55:29] Oh, that was nanotube, my apologies. [04:55:40] no worries (-: [04:56:03] Honestly I've just used the established tools, like sdict. I know XML is not an optimal representation. [04:56:57] If you can load it into a database you might be able to use the same tools that Wikipedia uses to index it, but that seems a roundabout method. [04:57:00] i've done some for en.wiktionary but i need to more structured indexing of the article text to allow more advanced queries [04:57:43] no wikipedia entries are not very structured so the standard software doesn't have anything like i want to implement [04:59:05] thanks tho [05:06:50] So I'm doing the verboten, and poking around MediaWiki's db schema. [05:07:16] poke away it's nor verboten at all [05:07:27] I need a fast way to look things up by category, and MediaWiki::API is expensive enough that it's worth my time to rsync to my home system with more power and do the queries there. [05:08:40] There's something I find hard to understand/believe in there, though. It looks like category lookups are done by querying against categorylinks.cl_from. [05:09:12] Except cl_from is a *text field*...using a text field in a WHERE clause can't be cheap... [05:09:46] Is maintaining a lookup table too expensive? [05:10:33] hippietrail: The verboten part is depending on hidden internals... [05:12:00] yes categories are addressed by name since the actual category text page may or may not exist [05:12:11] no its all documented [05:12:34] though some might only be documented in the source (-: [05:12:40] and it does change from time to time [05:14:11] Even if the category text page doesn't exist, it's still plausible to maintain an text-to-int lookup table for the category name, isn't it? [05:14:21] do you intend to maintain a local mirror of the wiki database? [05:14:40] hippietrail: Yeah, actually. It doubles as my offsite backup. :) [05:14:42] for the actual wikis or for your own tool? [05:14:55] do you mirror live or from official dumps? [05:15:00] Not Wikipedia. [05:15:03] rosettacode.org [05:16:50] That's my site; A while back, I wrote a bot that maintains a bit of infrastructure. [05:17:26] This pretty much describes how the bot works: http://rosettacode.org/blog/?p=229 [05:17:31] basically im trying to figure out if you already have a local database, how you update it, and whether it has the offical schema or not [05:18:04] Cron job on my server runs mysqldump. Cron job at home runs rsync to grab a copy. [05:18:11] oh so you are talking about your own wiki hacking it a bit as a sysop/dev of your own site? [05:18:16] Yup. [05:18:50] ah ok well depending on your skillz you can pretty much change your local schema as you like but each mediawiki update there could be conflicts [05:20:03] I understand that. I don't like it, but it's the only way I can get the number crunching necessary for the infrastructure at the moment. [05:20:21] how many pages in your wiki? [05:21:05] Not incredibly large; The bigger problem is I only have a 256MB Slicehost slice. [05:21:35] http://rosettacode.org/wiki/Special:Statistics [05:21:39] i once made an extension that maintained some metadata about all pages in a separate table. that way it needed just read only access to the standard db schema [05:21:50] 547 content pages, but that doesn't include all the category pages which are actually important content... [05:21:54] i dont know what that means (-: [05:22:20] oh well that's tiny you sould be able to index that without fiddling with the schema at all [05:22:21] Near as I can tell, a "content page" is a page that's in the main namespace, not in a Talk, Category or any other. [05:22:39] yeah i'm not sure if "content page" is official terminology or not [05:23:12] hippietrail: The problem is that I have to query against not just two categories, but something around three hundred. [05:23:31] I have more categories than actual pages; It's how the site is organized. [05:24:01] do you need to query them live or just say once per day? [05:24:14] Currently, the cron job running the bot runs every four hours. [05:24:27] is it not working to your satisfaction? [05:24:58] Via MediaWiki::API, and without using any sleeps or rate limiting, it takes eleven minutes to complete, making the site incredibly slow in the process. [05:25:19] And that's using local filesystem caching to avoid slamming MediaWiki::API more than it needs to. [05:25:44] Although most of that happens during the page posts. [05:26:17] It posts more pages than it strictly needs to, and I haven't had time to properly clean that up yet. [05:30:39] why don't you query the database directly with SQL? [05:31:08] That's what I'm learning to do at the moment. [05:31:51] I was just concerned that the categorylinks page isn't very normalized, which means that the same string may be compared against multiple times when doing a search-by-name. [05:32:28] It depends on how MySQL handles indexes and text fields, I suppose, but it doesn't feel like it's going to be efficient. [05:32:48] you can do an sql query that tells you how efficient it will be [05:32:58] but my sql skills are not great [05:33:17] There's EXPLAIN, and there's also profiling. [05:33:22] still i have written category queries for both en.wiktionary and en.wikipedia which don't seem slow [05:33:25] yes [05:34:12] 03(mod) Having both lang and xml:lang attributes not identical to $wgContLanguageCode - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=5790 (10shinjiman) [05:34:14] Part of the problem I'm going to have is that my database is larger than my available RAM on my server. [05:34:33] Heck, it's larger than my total amount of RAM. [05:35:33] That's one reason I'm looking at doing the queries against my offsite backup; *That* machine has four cores and 8GB of RAM... [05:36:25] and your offsite backup also uses an actual wiki db that the same sql queries would work on unmodified? [05:37:56] hippietrail: mysqldump > file > rsync > file > mysql [05:38:17] where the dump is the server, and the restore is on my local system. [05:39:36] well i don't grok all that but it looks like it means you have the same db on both machines so should be no prob [05:39:49] though i'd be surprised if amount of ram is really an issue [05:40:23] db's were designed back when ram was expensive do they were implemented with algorithsms that didnt require much ram [05:40:44] Wait, what are you trying to do? [05:41:08] hippietrail: My slice has 256MB of RAM, 512MB of swap. It currently has 8MB RAM free, 178MB of swap in use, and 60MB of RAM devoted to cache. [05:42:14] yes what is the more specific problem? [05:42:31] OverlordQ: Simplified: Take two wiki categories, find the pages that are in one of them, but not both of them, and build lists of the pages that aren't in both. [05:43:02] Add a category for each of those pages to identify that it doesn't belong in one of the two major categories. [05:43:32] http://rosettacode.org/blog/?p=229 [05:43:41] Describes it more or less in detail, as it's done now. [05:43:52] that shouldn't be that hard with a plain sql query [05:43:59] *shortcircuit sighs [05:44:01] I know. [05:44:08] and unless you got a copule billion pages, memory shouldn't be a problem [05:44:25] I was figuring out the SQL query when I discovered I'd be querying against a text field, which didn't feel right at all... [05:45:00] Which is what I came in here to ask about; Why that table isn't normalized so as to have a numeric field in categorylinks.cl_to. [05:45:12] text field queries are easy peasy. and you wont even need pattern matching from the looks of it [05:45:28] I *know* text field queries are easy to do. [05:45:37] My concern is that it isn't as efficient as it could be. [05:46:14] I'm been bumping against the performance limits of my slice for months, and I've been looking for and removing any inefficiency I can find, as I have time. [05:48:35] I'm so perpetually low on RAM that I'm beginning to think memcached is a waste of resources, and I'm wondering if switching to a purely disk-based approach would be better. [05:48:57] Dunno, somebody older will have to tell you why it's stored as text [05:49:43] that problem shouldn't tax your memory limits at all [05:49:54] why what's stored as text? [05:50:06] flyingparchment: categorylinks.cl_to [05:50:08] why categories are stored as text [05:50:24] presumably because the category page might not exist [05:50:31] so there's no page id to use [05:50:36] it still gets an entry in the category table even if the category page doesn't exist [05:50:48] I'd think categorylinks.cl_to would point to a separate table (categorynames?) that maps a numeric value to a string. [05:51:20] Or to the category table's pk. :) [05:51:35] categories dont have an id. category pages have an id but those pages can come and go or not exist at all [05:52:06] hippietrail: category.cat_id? [05:52:11] the brions and tims of this world will surely give you some good reasons why [05:52:20] some categories do have ids now, but it's very new, whereas categories are from 2004 [05:53:33] Figures I'd get activated for weather spotting and not see anything, go home when net's green, and then things would get noisy *after* the watch expired. [05:55:55] flyingparchment: So, on older installs, there might be categories that aren't in the categories table? [05:57:56] Aside from cl_to being a text field, the categorylinks table looks like a normal many-many bridge table, to me. [06:06:54] Ew. [06:07:16] The category titles appear in two places: category.cat_title and categorylinks.cl_from . [06:08:09] that would be how you JOIN category and categorylinks, no? [06:08:20] Looks like it. [06:08:26] good morning [06:08:29] could any one help with this [06:08:33] https://bugzilla.wikimedia.org/show_bug.cgi?id=12242 [06:09:40] *shortcircuit knows nothing about that part of MediaWiki. [06:10:22] 03philip * r54697 10/trunk/phase3/includes/ (8 files in 2 dirs): Update the Chinese conversion tables. [06:10:24] I haven't seen Tim in a while. :-( [06:16:18] I can't play with it any more tonight. Thunderstorm has caused power flickers twice, and I don't have a good UPS. [06:18:36] so you have UPS but it's bad enough not to protect you from power flickers? [06:19:00] Usama: No, I don't have a UPS. [06:19:02] :) [06:19:50] And I'm more worried about a close strike than an actual outage. [06:20:10] :) I'm still learning English. so I asked. [06:20:12] A surge would be a bad thing; I can't afford to replace any of my home equipment right now. [06:20:29] Usama: Your understanding was fine; My statement was slightly misleading. [06:20:40] Generally, the implication would be what you interpreted it as. [06:20:42] oh, thank you [06:21:04] I suggest you to turn the cut the power off [06:21:12] sorry, [06:21:48] Already did. [06:22:19] Fortunately, I've got a router I can fall back to, I'm currently on a laptop, and Comcast's cablemodem is horrible, anyway. [06:22:43] I see [06:22:55] good luck then [06:23:08] The desktop machine with the high specs is unplugged, now. :) [06:29:04] Hello ppl, i got an Error with the Widget Extension. Is there any guide how to install it, cause i dont know, if i installed it correctly. can i check it somehow? [06:29:36] I followed the instructions from the mediawiki page, but i get a few php errors when i start it: [06:29:49] PHP Warning: strftime(): It is not safe to rely on the system's timezone settings. You are *required* to use the date.timezone setting or the date_default_timezone_set() function. In case you used any of those methods and you are still getting this warning, you most likely misspelled the timezone identifier. We selected 'Europe/Paris' for '2.0/DST' instead in C:\Inetpub\wwwroot\mediawiki\extensions\Widgets\sm arty\Smarty_Compiler.class.php on line 400 [06:29:49] PHP Warning: include(C:\Inetpub\wwwroot\mediawiki/extensions/Widgets/compiled_templates [06:30:00] PHP Warning: include(): Failed opening 'C:\Inetpub\wwwroot\mediawiki/extensions/Widgets/compiled_templates/\%%A9^A9B^A9BDE9A6%%wiki%3ATest.php' for inclusion (include_path='C:\Inetpub\wwwroot\mediawiki;C:\Ine tpub\wwwroot\mediawiki/includes;C:\Inetpub\wwwroot\mediawiki/languages;.;C:\php5\pear') in C:\Inetpub\wwwroot\mediawiki\extensions\Widgets\sm arty\Smarty.class.php on line 1265 [06:31:28] It seems that there is something wrong with the timezones, i already had the problem when i tried to connect to my special:preferences page [06:33:56] the "problem"-line is: [06:33:57] $template_header = "_version.", created on ".strftime("%Y-%m-%d %H:%M:%S")."\n"; [06:35:15] i did a mistake @ copy and paste the php warnings, i re-do it (sorry for spamming) [06:35:17] PHP Warning: strftime(): It is not safe to rely on the system's timezone settings. You are *required* to use the date.timezone setting or the date_default_timezone_set() function. In case you used any of those methods and you are still getting this warning, you most likely misspelled the timezone identifier. We selected 'Europe/Paris' for '2.0/DST' instead in C:\Inetpub\wwwroot\mediawiki\extensions\Widgets\sm arty\Smarty_Compiler.class.php on line 400 [06:35:17] PHP Warning: include(C:\Inetpub\wwwroot\mediawiki/extensions/Widgets/compiled_templates/\%%A9^A9B^A9BDE9A6%%wiki%3ATest.php): failed to open stream: No such file or directory in C:\Inetpub\wwwroot\mediawiki\extensions\Widgets\sm arty\Smarty.class.php on line 1265 [06:35:17] PHP Warning: include(): Failed opening 'C:\Inetpub\wwwroot\mediawiki/extensions/Widgets/compiled_templates/\%%A9^A9B^A9BDE9A6%%wiki%3ATest.php' for inclusion (include_path='C:\Inetpub\wwwroot\mediawiki;C:\Ine tpub\wwwroot\mediawiki/includes;C:\Inetpub\wwwroot\mediawiki/languages;.;C:\php5\pear') in C:\Inetpub\wwwroot\mediawiki\extensions\Widgets\sm arty\Smarty.class.php on line 1265 [06:35:41] Ravex: don't paste here [06:35:46] ups, sorry [06:35:50] Ravex: use fpaste.org [06:35:56] okay, secon [06:36:28] http://fpaste.org/T061/ [06:36:30] like this? [06:36:40] yes [06:36:52] sry, didnt know that [06:37:06] np [06:37:23] this extension makes me crazy <.< [06:38:18] I hope some one will look into it and help you soon [06:39:44] i already tried to use different parameter, as explained in the error-message, but didnt help [06:39:55] you can ignore the strftime 'warning', the other part is where your problem is [06:41:00] ah i read it like: he couldnt create the first page, then it didnt create a file he should compile [06:41:20] and then the compile errors "appears" [06:41:34] but i dont know alot of php ;) [06:42:54] not a clue since it's on windows [06:44:19] http://fpaste.org/On6r/ (i uploaded the messages again, cause there was a space on sma rty , dunno why) [06:45:57] my guess is windows might not like the strange characters in the file name but that's just a guess [06:47:42] its a temp file, cause it doesnt exist (just a guess) [06:47:59] like he caches the file, compiles it, and puts it on the page [06:48:31] and cause he got an error with the timezone, he doenst find a file -> No such file or directory in C:... [06:51:13] i already had a timezone problem in \includes\specials\SpecialPreferences.php Line 1052. i had to change $now = date_create('now'); into $now = date_create('Europe/Paris'); [06:53:57] that's the wrong way to fix that [06:55:00] well, but it worked, so i didnt spend more time on it :/ [06:55:20] in your php.ini add date.timezone='whatever' [06:55:46] ah okay, second. i add it [06:58:13] iam getting this error then: http://fpaste.org/3Ncu/ [06:58:19] the one i got before [06:58:37] did you restart apache? [06:58:45] iam using iis [06:58:48] or IIS, or whatever you're using [06:58:52] sec ;) [06:59:23] iam just a trainee here in this company :P, but the sys-admin doesnt know much about it, too.. [06:59:49] Hi! I can't login to svn server to upload my extension there. I've probably forgot my password I used to generate a public ssh key (it was long time since my commit request). Also, my linux account name differs from the one I've requested at svn server, I wonder whether that cause a problem. [07:00:25] a) that'll be a problem b) no [07:00:46] worked ;) [07:00:57] OverlordQ: what if I'll regenerate a public key, can you or someone re-upload it? [07:01:15] OverlordQ: the key should be stored at server [07:03:13] probably need to talk with somebody in -tech [07:03:29] OverlordQ: perhaps I should write a email request to Brion Vibber? [07:03:44] http://www.mediawiki.org/w/index.php?title=Manual_talk:Opening_external_links_in_a_new_window&diff=269963&oldid=242656 [07:03:52] > When I am in a website and I click on an off-site link, without being notified that I am switching domains, I find it as an insult. [07:03:53] I would not dare to speculate on if emailing brion is a good or bad idea [07:04:43] OverlordQ: i restarted the server, after editing the php.ini. it worked now. it also removed the time-zone problem in the Extension:Widget problem i had. the new error message is now: http://fpaste.org/nnc7/ [07:04:43] OverlordQ: ok, I'll try wikitech list. [07:05:55] OverlordQ: Maybe I should try to remember the password, but I wonder whether about 20 or so unsuccessful login attempts will cause an throttle or some kind of alarm at server side. [07:06:15] it's not a password, it's a passphrase used to encrypt the key and is never sent to the server [07:06:20] you can try as much as you like [07:06:39] flyingpatchment: but the phrase should match, isn't it? [07:06:44] match what? [07:07:36] flyingpatchment: I've been asked for password when I try svn checkout svn+ssh://questpc@svn.wikimedia.org/svnroot/mediawiki/trunk/phase3/extensions wiki [07:07:47] if it asks for a pass_word_ then you're doing something wrong [07:07:55] it should ask for a pass_phrase) [07:08:15] yes it asks for password, not a pass phrase. [07:08:24] perhaps your key is in the wrong place or you didn't configure the client to use it [07:09:40] flyingparchment: I've read the configuration guide at http://sial.org/howto/openssh/publickey-auth seems to performed all steps. maybe there's some my mistake though [07:10:24] paste the output of ssh -vvv questpc@svn.wikimedia.org to dpaste.org [07:13:49] uggga [07:15:54] how can i align a table to the center of the page [07:15:59] align: center does not seem to do it [07:17:51] got an example? [07:18:09] flyingparchment: http://dpaste.org/OMxv/ [07:18:33] OverlordQ: omfg, i solved the problem... i just had to less previleges on the compiled-folder... [07:18:50] CorpX: margin:auto; [07:18:55] eg:

foo bar

[07:19:40] QuestPC: are you sure that's the same key you provided when your svn account was created? [07:20:51] it doesn't look like it's reading the key right: debug2: key_type_from_name: unknown key type '-----BEGIN' [07:21:05] http://wiki.utexasclan.com/index.php?title=Chat_Topics_to_be_Avoided [07:21:06] flyingparchment: 1 sec I'll compare them again (I've provided only id_rsa.pub and it seems that it doesn't like id_rsa which isn't RSA1 key, but RSA2?) [07:21:11] im trying to align that to the center of the page [07:21:17] OverlordQ: i believe that's normal for an SSH2-style key [07:21:25] OverlordQ: it first tries to parse it as the old binary format [07:21:36] aaah, you're right [07:21:43] thnx [07:21:44] flyingparchment: strange private key seems to be RSA1, while public is RSA1. [07:21:46] margin auto fixed it [07:22:08] QuestPC: why do you think that? [07:23:02] id_rsa.pub matches the one I've sent in commit request [07:23:41] QuestPC: why do you think "private key seems to be RSA1, while public is RSA1" - and did you mean to say RSA1 twice? [07:24:34] flyingparchment: nope, that was a typo, I've meant one is RSA2 [07:24:41] okay, and why do you think that? [07:25:20] flyingparchment: due to debug message in log, but you're probably right that it just attempts to figure the correct format of file [07:25:38] QuestPC: which key do you think is RSA1? [07:25:41] Ignore what I said about format [07:26:22] flyingparchment: my public key is there: http://ingress.uniyar.ac.ru/~sdv/id_rsa.pub [07:26:31] QuestPC: which key do you think is RSA1? [07:26:41] flyingparchment: public one [07:26:50] no, that's an ssh2 rsa key [07:27:01] flyingparchment: ok, sorry. [07:27:04] ssh1 rsa looks like " " [07:27:19] or maybe only one larger number, i forget. [07:27:38] flyingparchment: so, can you tell from the log why doesn't it allows to login? [07:28:05] *OverlordQ 's only guess is the private key is wrong [07:31:59] how can i turn on cell borders [07:32:07] OverlordQ: Maybe I should regenerate the key pair? But then I probably should re-send a request to upload my public.key.. My guess that private key can't be easily fixed without regenerating the while pair of keys. [07:32:19] while=whole [07:32:25] http://wiki.utexasclan.com/index.php?title=Chat_Topics_to_be_Avoided right now, there's no borders, but if i add "border:1", the whole thing looks ugly [07:33:07] CorpX: border-collapse:collapse, perhaps? I use it a lot to make nice CSS borders [07:33:22] i just want a simple border like this - http://en.wikipedia.org/wiki/List_of_FIFA_country_codes [07:33:50] but border=1 does this -> http://wiki.utexasclan.com/index.php?title=Chat_Topics_to_be_Avoided&diff=859&oldid=858 [07:34:09] cell padding? [07:34:17] CorpX: have you tried for example something like this table {border-collapse:collapse}; td{border:1px solid black;border-collapse:collapse} ? [07:34:43] CorpX: yes, padding via css, too. I've made just a simplified example. [07:34:43] does that go in monobook.css? [07:35:20] CorpX: the style can be specified direcrlty at the wiki page, or my may want to change monobook.css to add new classes if you want [07:35:31] ok thanks [07:36:12] np, hope it helps. [07:38:09] flyingparchment: Should I try to regenerate key pair and re-send my commit request via email, or there's any way to fix it without bothering Brion Vibber? [07:48:43] OverlordQ: Does your id_rsa.pub looks similar to the one I've placed at: http://ingress.uniyar.ac.ru/~sdv/id_rsa.pub ? Does your private id_rsa starts with -----BEGIN RSA PRIVATE KEY----- and ends with -----END RSA PRIVATE KEY----- , too? [07:49:16] OverlordQ: I mean the format and/or size of the file. [08:01:24] 03philip * r54698 10/trunk/phase3/languages/LanguageConverter.php: Fix a bug of LanguageConverter::findVariantLink(). If the article has already existed, there is no need to check it again, otherwise it may cause a fault. [08:15:01] Hi. Could someone give me a hint where "keyword" meta tags are set in MW? OutputPage keeps them, but a case-insensitive text search for "keyword" in all MW files did not help me to find who sets them. [08:15:22] In particular, I wanted to adjust the default keyword set which is based on the links found on a page [08:15:40] nick p858snake [08:18:02] whats an easy way to import flags of all the countries [08:18:06] the little thumbnail [08:19:50] CorpX: import these from Image: (File:) namespace with xml dump (with the option --uploads and correct option --server=http::/yourserver.org [08:20:11] how does the template work [08:20:18] {{flag|Albania}} produces the flag [08:20:31] where is that template [08:20:34] CorpX: go to page Template:flag to see it's source text [08:21:40] can i just export that template over? [08:22:03] that wouldnt get the images though, would it [08:22:20] MarkusK_: http://www.mediawiki.org/wiki/Special:Code/MediaWiki/53482 [08:24:01] CorpX: if you click the option to export all dependices it might also get the files [08:24:07] but i can't remember for 100% [08:24:32] where is that optin [08:24:33] i dont see it [08:24:49] Sparlka: Thanks, interesting. So my wiki had the keywords still in cache, although its code did not generate them any more. [08:25:01] CorpX: you can export, the similiest way is go to Special:Export and specify Template:flag there. it won't export the related images automatically [08:25:02] Splarka: Thanks, interesting. So my wiki had the keywords still in cache, although its code did not generate them any more. [08:26:21] i dont really know how to do the xml dump [08:27:03] Splarka: I am not sure that keywords are absoultely ignored. in google webmaster tools there's a section which checks the keywords, it had complained at many wiki pages (MW 1.11) having too short or duplicate keywords. [08:27:43] Splarka: I've made my private patch to filter out too short and duplicate words then. [08:27:47] QuestPC: I'm not brion ^_^ [08:27:57] ...yet [08:28:14] could i just make my wiki grab the images not present locally from commons? [08:28:33] MarkusK_: are you using 1.16 off of trunk? (the tag was just removed a few weeks ago) [08:29:06] Splarka: ok. I don't know why the webmaster tools checks for these words if they don't use these? I think they keep actual ratings in secret to prevent too much of SEO optimization (which puts bad sites top and kills good sites) [08:29:18] nm i see it [08:29:52] QuestPC: I don't know or care about meta keywords, I just remembered brion removed them recently, so could show MarkusK_ (and as a bonus, it shows the code that generated them before they were removed) [08:30:11] but you can leave a comment on that revision [08:30:21] Splarka: yes, I am running SVN head on my local machine, which is where I looked for the code [08:30:25] CorpX: yes you can, you can use ForiengFilesRepo or whatever its called [08:30:50] MarkusK_: heh, fair enough, no wonder you couldn't find the code! try action=purge on a page with meta keywords and it should go away? [08:31:17] Splarka: Yes, I did that and it worked as expected [08:31:25] when i exported template:flag, it only got the poland one (shown in examples) [08:31:31] it didnt grab the templates for the other countries [08:32:30] i guess i should export the whole category? [08:32:48] CorpX: strange, template should be generic. I wish MW export has the option to "export this page and all imagelinks, templatelinks, categorylinks and so on) [08:33:09] CorpX: but maybe that would be too slow and too long. [08:33:29] yea [08:33:39] http://en.wikipedia.org/wiki/Category:Country_data_templates [08:33:41] have to grab that [08:34:19] 1mb file [08:34:36] CorpX: make a list and pase it to Special:Export. or, try to create a bot, though it would make a longer time to write and I don't think that every user is allowed to run a bot at wikipedia.. [08:34:48] pase=paste [08:34:59] hmm [08:35:12] QuestPC: i just did import category ;) [08:35:50] CorpX: it probably won't export/import the templates, but maybe it has been changed since. [08:49:06] *svip pokes Splarka. [08:50:55] *Splarka can't offhandedly think of a cogent paradigm to reply to that in the manner emulating a situational event in futurama [08:51:06] *Splarka beeps [08:52:37] O_O [08:52:39] Good reply. [08:52:57] Emufarmers: My ultimate reasoning for not voting for that right wing guy is because he likes South Park. [08:54:28] I like South Park, but don't watch it, because I hate Comedy Central [08:54:46] Ah. [08:54:50] svip: also, he's a criminal [08:54:52] What's wrong with CC? [08:54:58] werdna: How can he run then? [08:55:00] contrarily, I used to watch Futurama a lot, until it moved to Comedy Central, not because I necessarily liked it, but because it was on between Family Guy and Aqua Teen Hunger Force [08:55:11] Can't we just make a claim that he is a citizen of Kenya, werdna? [08:55:19] svip: there's no rules against criminal convictions [08:55:29] But Kenya! [08:55:31] svip: MST3K [08:55:45] Splarka: Awesome. [08:55:45] how can i do align=center for a whole column in a table [08:55:50] so i dont have to do that for every cell [08:55:57] Hopefully, then, FOX will take it back. [08:56:00] was the best show ever in the universe, and they cancelled it, and the Sci-Fi version was crap [08:56:14] heh [08:56:20] Splarka: On the other hand; I don't watch TV at all. [08:56:24] Neither can I get CC. [08:56:30] So I just download the episodes after they have aired. [08:56:40] And watch it while I code MediaWiki. [08:56:41] criminal! [08:56:54] unless you live in Kenya or Finland [08:56:54] Uh uh. [08:57:04] I buy the DVDs when they are released. [08:57:20] But I am not going to wait two years until that happens. [09:02:16] hippietrail, shortcircuit: Yes, cl_to is a text field, but that doesn't mean WHERE clauses against it are expensive. It's an indexed varchar field, which means things like WHERE cl_to = 'Foo' , WHERE cl_to IN ('Foo', 'Bar', 'Baz') and WHERE cl_to LIKE 'Foo%' are quite fast [09:03:06] RoanKattouw: yes it would only be expensive if you did an unanchored pattern match which i dont think he needs to do [09:11:22] 03nikerabbit * r54699 10/trunk/extensions/Translate/Groups.php: Simplify [09:11:34] Splarka: finland? [09:12:31] lawless savages [09:12:51] 03nikerabbit * r54700 10/trunk/extensions/Translate/FFS.php: Support different separators [09:13:29] 03nikerabbit * r54701 10/trunk/extensions/Translate/MessageCollection.php: Don't die for invalid titles, just warn [09:17:08] 03nikerabbit * r54702 10/trunk/extensions/Translate/groups/FUDforum/ (. FUDforum.yml README): Base for FUDforum support - does not work yet [09:20:26] Nikerabbit: I think he is confusing it with Sweden. [09:50:15] 03catrope * r54704 10/trunk/extensions/UsabilityInitiative/js/ (3 files in 2 dirs): EditToolbar: Prevent JS error occurring when trying to access selectionStart attribute of a hidden textarea, resulting in the encapsulateSelection event never being triggered and breaking wikEd [09:50:51] 03(FIXED) New toolbar breaks wikEd - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=20134 +comment (10roan.kattouw) [09:51:12] catrope is RoanKattouw? [09:51:16] Yes [09:51:21] This throws my entire perception of reality into question. [09:55:05] uggah [09:59:44] *werdna hsows RoanKattouw his prefs patch [10:00:15] shows, too [10:00:21] I looked at it briefly [10:00:34] But like you agreed, isn't it gonna cause cache fragmentation? [10:01:04] to no further extent than logged-in users. [10:01:22] Obviously there's no point in giving all those preferences to anonymous users [10:01:41] Right so it's gonna be restricted to just a few of them [10:01:58] I don't think we'd give them a full prefs screen [10:02:10] Just let the optin stuff turn individual prefs on/off [10:03:11] Still, given how many anons there are and if even a few percent of them opt in, it's gonna split the Squid cache into two [10:04:01] so you don't want people to opt in? [10:04:01] Because if enough anons use Vector, every (or nearly every) page in the cache has to have a Monobook and a Vector version [10:04:32] I'm confused, why are you objecting to people trying out the new skin? [10:04:36] Isn't that kinda the point? [10:04:48] No; in fact, we were asked by Naoko to implement this for anons and Brion said something about cache fragmentation, at which point I explained to Naoko that it would hurt the cache badly [10:05:06] So I'm not against the idea of it, just wary of the performance impact [10:05:29] Well, you'll get cache fragmentation by more people logging in to try the skin, too. [10:05:30] Of course, if one or more of Brion, Tim and Domas approves, I'll shut up :) [10:05:48] Why would you want prefs for anons? Isn't it sort of "tough luck, suckers, no account, no prefs". [10:05:59] Yes, but the size of the cache wouldn't double the way it would with anon prefs [10:06:09] svip: they want more vector supporters [10:06:10] svip: Why would you say that when you can implement them? [10:06:29] It's not a contest to be the biggest smartass, you know. [10:06:37] Splarka: Use cookies, when I say "useskin=" you set it. [10:06:55] Or rather, "setskin=". [10:07:09] You know, as a method to set the skin, without all the prefs thingie. [10:07:12] it isn't just vector, it's also the toolbar [10:07:31] Okay okay, then "setbeta=true" [10:07:45] What's better about that? [10:07:47] And a switch in a cookie, "yeah, I'm using beta, give me the code". [10:07:59] werdna: You avoid all the other prefs, like timezone and stuff. [10:08:14] Which is exactly what I've done, except that I'm storing any preferences that other parts of code may choose to set [10:08:16] And, when you change some stuff in the beta, oi, you already had it set. [10:08:29] I'm not opening up the whole preferences window [10:08:31] as I said before. [10:08:33] werdna: Where do you store it? [10:08:34] how about a method with no cookies, no prefs, no sessions [10:08:35] skin for anons? [10:08:40] how do people imagine implementation of that? [10:08:40] svip: in a cookie. [10:08:44] Good. [10:09:03] Splarka: How about, get a browser with cookie support? [10:09:05] domas: werdna has a patch for that *finds link* [10:09:17] svip: got one thanks, not related to the point at all, thanks [10:09:21] domas: whipped thisup last night (untested) https://bugzilla.wikimedia.org/show_bug.cgi?id=20151#c2 [10:09:23] does werdnas patch disable caching at squid level? [10:09:47] domas: I haven't fixed the X-Vary-Options header yet [10:09:49] Splarka: Well, perhaps you could just add "&usebeta=yes" to the link? [10:09:50] but that's what I'd do, I assume [10:09:52] Like useskin. [10:09:59] no, but what you could do [10:10:20] is just temporarily, while pimping it, have all links in vector append ?useskin=vector by default (for anons only) [10:10:36] self cascading [10:10:41] But then the others said it also about the toolbar! [10:10:41] and cacheable (by URL) [10:10:52] so have the toolbar a button on the edit page [10:11:00] [click me to load a toolbar, oh yah, click me all night long] [10:11:03] Splarka: that would fragment the squid cache just as much as a cookie storing preferences. [10:11:04] *Splarka smarms smarmily [10:11:29] Splarka: and it would have the disadvantage that it would go away when somebody comes back [10:11:35] sure, but then you don't need to mess with anon prefs [10:11:49] werdna: if only "try beta" would go away... >_< [10:12:05] then that might be a concern, heh [10:12:11] *headdesk* when doing remote work on machines, dont accidently bring down the network interface [10:12:20] Splarka: But I've already implemented anon prefs, it's much easier than making all of those links cascade useskin properly. [10:12:35] Splarka: Cascading ?useskin=vector breaks parser cache [10:13:06] Oh, I know, don't allow anons to use the beta. [10:13:12] svip: why would you do that? [10:13:16] It's stupid [10:13:22] especially if you show a 'try beta' link for them [10:13:23] werdna: Make a fake domain then. [10:13:27] e.g. beta.en.wikipedia.org [10:13:28] svip: ? [10:13:30] Roan: mmm, sounds like Tim needs to write a patch for that, or include URI parameters in X-Vary-options [10:13:30] With BETA [10:13:38] Yes [10:13:39] 03vasilievvv * r54705 10/trunk/phase3/maintenance/parserTests.txt: Update parser tests for r54693 [10:13:57] that could work, but it would fragment the parser cache just as much [10:14:05] Splarka: it's obvious you don't know what you're talking about there. [10:14:09] werdna: By it would more transparent. [10:14:23] And, I'd say, more "controllable". [10:14:41] svip: a button to opt in, a button to opt out... [10:14:44] werdna: fair enough, but would the parser cache need to be invalidated for different skins? and if not, why not allow useskin not to invalidate it? [10:15:05] Splarka: it wouldn't invalidate, it'd fragment [10:15:09] Splarka: Yes, the pcache is per-skin [10:15:11] werdna: Obviously, when you are on en.wikipedia.org, you are out, when you are on beta.en.wikipedia.org you are in. [10:15:15] there's a separate parser cache per skin-date format-something else [10:15:26] Though en.beta.wikipedia.org may be more interesting. [10:15:39] But appending ?useskin= to links dynamically won't work without disabling the parser cache on every ?useskin= hit [10:15:44] svip: Why are you coming up with a million alternative solutions to something that THERE IS ALREADY A PATCH FOR? [10:15:57] RoanKattouw: or fragmenting it. [10:16:13] Your concern is still fragmenting! [10:16:27] I realise that cache fragmenting is unavoidable here. [10:16:32] Squid cache fragmentation is unavoidable [10:16:43] Whichever way you dice it, you have to cache a beta version and a non-beta version. [10:16:44] Wait, I just said that. [10:16:46] There is no way around it. [10:16:52] Indeed. [10:16:59] is the patch committed? is the commit "okay"d in code review? is it syched to deployment branch? is it scapped live? [10:17:15] you do not scap a patch [10:17:24] no, it's sitting mostly finished on a bug report [10:17:25] It's not even tested xD [10:17:32] then there is room for debate [10:17:33] (or so werdna said) [10:17:56] especially if domas is in a fetal position on the floor sobbing quietly [10:17:58] All you're doing is asking people to do more work for no apparent reason [10:18:40] I prefer the preferences patch because it's a generic solution to a whole class of problems [10:19:02] Yeah, but... hm, I dunno, I realise why a beta will be nice to try out on enwiki, but to avoid too many cache issues, wouldn't it -- sort of -- make sense to try on a limited enwiki. [10:19:05] nobody's given me a shred of reason to think that any other solution will be better. [10:19:18] Still edtible and whatnot, but you know, just not as many pages. [10:19:21] svip: That's prototype.wikimedia.org [10:19:27] I know. [10:19:38] But that is like more alpha than beta. [10:19:51] Aren't they still configuring that thingie? [10:20:52] with respect, neither you nor I have the ability to decide whether or not it's a good idea to try the beta on the live site or on a limited subset. [10:20:58] https://bugzilla.wikimedia.org/show_bug.cgi?id=14602 [10:21:26] werdna: I guess not. But I guess you did tell them about the cache fragmention? [10:22:48] tell who? [10:23:20] Dunno, whoever made the decision of trying a beta on a live site, and later wanted anons to do so as well. [10:23:43] I'll discuss with the usability folks and Brion when they're up later today [10:24:11] Hm. [10:24:13] 'kay. [10:24:13] Splarka: that bug was closed by Siebrand and Simetrical. [10:24:26] With respect to both of them, I don't think their decisions are binding. [10:24:44] In addition, it may make sense for non-Wikimedia wikis. [10:24:55] Well, some. [10:25:54] werdna: fair enough, but the issues still exist, the only difference is a developer wants it now instead of a user ^_^ [10:25:58] 03nikerabbit * r54706 10/trunk/extensions/Translate/ (_autoload.php ffs/Gettext.php): Read part of GettextFFS implemented [10:26:02] (also, you should dupe that to your bug) [10:26:21] *Splarka waits for "not just a developer actually..." [10:26:31] Splarka: I'm sure Brion will know what cache fragmentation is. [10:26:44] Have at least some faith that between Brion and I there is some semblance of competence :) [10:26:57] it's closer to Brion than I, though ;) [10:27:36] oh, I wasn't pointing out incompetence, just "a shred of reason" [10:27:46] (and a bit of hypocrisy) [10:28:06] ((and a small salad, please, no onions)) [10:28:57] werdna: have time to fix the annoying warning in gif.php? [10:29:18] svip: No, we're not still configuring prototype. We're using it to stage alpha stuff which gets updated sort-of-daily (usually whenever we've got something new) [10:30:59] or did you fix it already [10:31:22] Oh I love that class name: "GettextFFS" [10:31:33] hilarious [10:31:45] of all the commit messages, I'm missing 54703 [10:32:10] *Nikerabbit demands compensation [10:32:50] RoanKattouw: why you love it? [10:32:52] Splarka: I've succeeded in reading most of the sed stuff, but I can't find the one stripping newlines, must be one of those I don't understand. I'll get Trevor to look at it when he's in [10:33:24] *Splarka gives Roan cookies (persistent, fragments his cache, but chocolate chips) [10:33:36] Nikerabbit: Because FFS = for fuck's sake? [10:34:03] 03(mod) Change of way to initialize editToolbarConfiguration structure - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=20125 (10liangent) [10:34:04] *RoanKattouw sets proper X-Vary-Options so cookies will get through and asks Splarka to resend [10:34:37] 03(mod) Change of way to initialize editToolbarConfiguration structure - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=20125 (10liangent) [10:35:00] RoanKattouw: ironically, that mirrors the feeling while developing that part of code [10:35:12] 03(mod) Change of way to initialize editToolbarConfiguration structure - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=20125 (10roan.kattouw) [10:35:14] 03werdna * r54703 10/trunk/phase3/includes/media/GIF.php: Moar error handling for r53546 [10:40:14] 03(mod) Change of way to initialize editToolbarConfiguration structure - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=20125 (10liangent) [10:42:33] 03catrope * r54707 10/trunk/extensions/UsabilityInitiative/ (5 files in 2 dirs): EditToolbar: Add special characters for Telugu, thanks to Veeven for providing them and to GerardM for asking around for them; bump style versions for r54704 [10:42:55] https://bugzilla.wikimedia.org/show_bug.cgi?id=12242 [10:42:58] any help? [10:45:14] Usama: action=render is not meant to be used as it [10:46:19] yeah, it is a raw render of the wikicode. [10:46:24] it's used when fetching pages internally, e.g. interwiki transclusion, description page from commons on local wikis, ... [10:46:35] it has no directionality. only the interface around it does. [10:46:53] and thus putting extra stuff arround it would break that [10:47:14] that's a wontfix for sure. [10:47:16] Usama: just wrap it in

or the page in ? [10:47:33] thedj: +1 :) [10:47:43] (wherever you transclude/include/use it that is) [10:47:51] can I insert any thing in ifram [10:48:52] I tried it localy but couldn't do any thing [10:49:11] ahh, well that goes back to Usama: action=render is not meant to be used as it [10:49:19] then don't use an iframe :D [10:49:21] an iframe loads it as if you were visiting it by direct URL [10:49:29] and is one of the ways you can't use action=render really [10:49:38] (without breaking everything) [10:49:51] in that case, you might better use a special skin [10:50:07] &useskin=myskin is pretty boring, er good [10:50:11] that will only present those interface elements that you want to present [10:50:34] but will keep most of the required css. [10:50:58] &printable=yes is another method that might be useful here. [10:51:06] there maybe should be an option like &action=noskin or something [10:51:17] which strips the UI (rather than hides it with CSS or gives no headers) [10:51:33] though that would give hotlinkers a bit of a free ticket [10:52:44] btw [10:53:15] pessimistically speaking, different skin for anons means twice more cache requirement on squids [10:53:29] (or twice less efficient cache) [10:53:50] we don't have too high hitrates at the moment either [10:54:10] so, getting that feature with pessimist attitude would need serious squid expansion [10:54:16] why not high? as a percentage of all requests, all page views, or all action=view ? [10:54:27] action=view [10:54:38] besides [10:54:43] it would double backend load [10:54:48] for pageviews [10:54:52] =) [10:55:23] thats optimistic though for feature use :)) [10:56:02] if that feature is rolled live [10:56:09] I'd like that every interested party knows possible costs [10:57:25] domas: Brion advised us (usability) against it about a month ago, he knows it'll at least fragment the Squid cache. Dunno if he thought about double anon-load on backend or doubling the size of the Squid cache, though, but knowing him he probably does [10:57:39] does he? [10:58:02] well, it will not double the size of the squid cache [10:58:07] it will double requirements for size of squid cache [10:58:25] thats a bit different [10:58:58] Yes, and if people actually start using the feature, it'll come pretty close to doubling the actual size [10:59:09] I told you [10:59:16] it will not double the size of the squid cache [10:59:19] because it is fixed size [10:59:46] anyway, I told very pessimist scenario (which is very optimistic for usability team) [10:59:52] Ah, I guess that's a Squid implementation detail than, I know nothing about that [10:59:54] Yeah [11:00:03] which is, that the use will be high enough to actually spread the cache set [11:00:13] why squid implementation detail [11:00:17] which 'cache' does not have fixed size? [11:00:53] You're right that's optimistic, but a pretty low percentage will probably spread the cache set enough (20-80 rule) [11:01:15] or contrary [11:01:19] I don't know, I've never configured caches before xD I just have some vague clues about how they work [11:01:19] see, the trick is [11:01:30] that majority of our users hit same pages [11:01:42]