[00:23:57] hi everyone [00:24:17] i need help in configuring the thumbnails for images in my wiki page [00:24:33] the server it's hosted has passthru() function disabled [00:24:34] where are the links in the tabs made? Changing makelinkobj didn't seem to affect them. [00:25:38] so instead of the thumbnails appearing, an error message appears: [00:25:40] Error creating thumbnail:
[00:25:42] Warning: passthru() has been disabled for security reasons in /home/khsy/domains/khairul-syahir.com/public_html/wiki/ib/includes/GlobalFunctions.php on line 1839
[00:25:54] can anybody help me?? [00:26:25] i've tried asking the host to enable passthru(), but they seem to be reluctant as that would cause a security vulnerability [00:27:06] try exec? [00:27:37] if passthru is disabled, it seems unlikely exec would be allowed [00:27:52] silverks_: you need to use GD instead of imagemagick [00:28:31] ok, how do i change from ImageMagick to GD? [00:28:45] GD is the default, you should undo whatever you did to enable imagemagick [00:29:41] (probably, make sure $wgUseImageMagick is not set in LocalSettings.php) [00:29:55] ok, i'm commenting that now [00:30:46] ok, that fixed the thumbnailing for normal images [00:30:52] now i'm having one more problem [00:31:09] the thumbnailing for svg images still shows the error message [00:31:19] you can't render SVGs without an external program [00:32:36] i thought ImageMagick can do it, even though without great accuracy? [00:32:50] imagemagick is an external program [00:32:56] that's why it doesn't work when passthru is disabled [00:33:04] your host doesn't let you execute external programs [00:33:10] ah...i see.. [00:33:33] then, is there any workaround for this? [00:33:44] like, installing svg renderer on cgi-bin for example [00:33:58] i did that for mimetex [00:35:52] 03dale * r39374 10/branches/MetavidWiki-exp/frontend_pages/index.php: auto complete search fix [00:45:20] by the way, thanks twincest for the help [00:45:32] at least i got the thumbnailing for the other images working fine now [00:48:55] 03dale * r39375 10/trunk/extensions/MetavidWiki/ (159 files in 20 dirs): (log message trimmed) [00:48:55] merged experimental branch: here is a draft of the official release notes: [00:48:55] == Unified Search == [00:48:55] * new unified search model groups and aggregates relevant semantic metadata per search [00:48:55] * advanced search improvements [00:48:56] * improved autocomplete suggestions/display [00:49:00] == Improved Skinning Support == [00:49:30] big merge committed ;) [01:16:34] 03(mod) List previously deleted pages as Recreated instead of Created ( N; New) - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=15131 (10N/A) [01:50:01] 03(mod) alphabetical order method for DynamicPageList - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=14971 +comment (10mediazilla) [01:55:13] 03demon * r39376 10/trunk/phase3/ (4 files in 2 dirs): [01:55:14] * Bug 12976: Use $WebResponse->setCookie() rather than raw setcookie() calls. [01:55:14] * Moved all of the debugging/logic to WebResponse so it can be properly used elsewhere. [01:55:14] * A bit of cleanup so cookies set by $wgUser->setCookie() use $wgCookiePath as they should. [01:55:14] * Bug 14887: $wgEnablePersistentCookies has been added to allow for disabling of persistent cookies. [01:55:30] 03(mod) WebResponse->setcookie() should handle prefixes - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=12976 +comment (10innocentkiller) [01:55:38] 03(FIXED) WebResponse->setcookie() should handle prefixes - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=12976 (10innocentkiller) [01:55:48] 03(FIXED) Configuration option to disable persistent cookies - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=14887 +comment (10innocentkiller) [02:02:24] I get 404 errors when I try to get to work the short urls (no root access)... I did exactly what they told in manual:short url, still get 404... Can someone help me? [02:02:44] try a different method? [02:03:25] well I don't know any other [02:03:27] ... [02:04:09] my mediawiki is hosted in w/ directory, and i'm using some kind of fix for php5 extension [02:08:56] no idea anyone? [03:37:01] ???Hook InfoboxDataCapture::save failed to return a value, why this happens with Infoboxdata.php ???? [04:04:58] I tried to cancel my "away" status and xchat abruptly exited [04:05:10] probably segfaulted, that's not very nice, is it? [04:05:41] Chatzilla FTW :) [04:05:56] can anyone confirm Roan's story about the duplicate post to wikitech-l? [04:06:30] there's no duplicate in the archives or on gmane [04:07:55] i saw a duplicate release announcement on wikitech and on mediawiki-l [04:08:10] you mean two posts to each? [04:08:13] yes [04:08:19] maybe there's an alias or something [04:12:30] brion will know [04:22:12] gcc doesn't seem to behave much like the documentation [04:22:17] gcc -MMD file.cc produces no output [04:25:54] 03river * r39377 10/trunk/tools/switchboard2/: new directory [04:27:18] 03river * r39378 10/trunk/tools/switchboard2/ (17 files): switchboard 2. faster, more scalable and less crashy. [04:30:10] I'm having trouble to configure short URL... I followed the steps in manual:short URL but it doesnt work [04:31:58] 03river * r39379 10/trunk/tools/switchboard2/main.cc: compile fixes for linux/gcc4 [04:33:27] proletaire: which procedure did you follow? [04:33:59] The one with no root access : http://www.mediawiki.org/wiki/Manual:Short_URL/wiki/Page_title_--_no_root_access [04:34:47] my mediawiki is hosted in a w/ directory and I use some kind of script to use PHP 5 [04:34:59] hmm [04:35:11] what results are you seeing? [04:35:15] error 404 [04:35:23] can't find the page [04:35:45] have you confirmed .htaccess is enabled on your server? [04:36:10] I'm sure it is, I used it for converting php files to ph5 [04:36:27] *php5 [04:36:29] hmm [04:36:40] could you pastebin your current .htaccess? [04:37:42] one sec please [04:40:09] you got it? [04:40:32] yes - http://pastebin.com/ for future reference, BTW :) [04:41:03] sry, didnt know about that pastebin thing [04:41:57] OK; where is your w/ folder located exactly? [04:42:13] is it in the document root, e.g. http://example.org/w/ ? [04:43:04] it's in the first directory (www.example.com/) while the rest is in www.example.com/w [04:43:10] as they said to do it [04:43:29] wait, what's in the root, exactly? [04:44:39] nothing execpt .htaccess and index.php that redirects to w/ [04:44:56] hmm, remove that index.php for one [04:45:02] .htaccess, as far as I know, affects all subdirectories [04:45:20] Uhm... just to note... [04:45:40] .htaccess stuff normally relies on nonexistance of files [04:45:51] If you have a redirecting file it might skew things [04:46:13] Dantman: yes, why I suggested he remove index.php in his root :) [04:46:21] RewriteRule ^/*$ wiki/ [L,QSA] [04:46:23] ^_^ yup [04:46:28] covers it [04:46:33] eeek [04:47:24] null paths ftl... ^_^ action paths ftw! [04:47:40] deleted the redirect, didnt change a thing [04:47:54] proletaire: one thing to try; in the first RewriteRule, add a / before w/ [04:48:06] it's a php redirect like location: http://www.example.com [04:48:35] so it'll instead look like: RewriteRule ^wiki/(.*)$ /w/index.php?title=$1 [PT,L,QSA] [04:48:51] ^/wiki/ ? [04:49:14] Dantman: no, ^wiki stays [04:49:44] tis the way mod_rewrite works, it' confusing -_- [04:49:50] bleh [04:49:57] I'll stick with nginx [04:50:08] adding "/" almost worked... it seems that it now finds the page, but without the short URL... I'll look further with that [04:50:11] thank you [04:50:28] the URL is : elcq.net/w [04:50:44] Did you set $wgActionPath? [04:50:50] Ack no [04:50:54] $wgArticlePath [04:51:14] yep, $wgArticlePath should be all you need now. :) [04:51:39] articlepath is set, so scriptpath [04:51:44] 03(NEW) User preferences: Date & Time - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=15169 15enhancement; normal; MediaWiki: User interface; (AheunHana) [04:51:57] $wgScriptPath stays as /w [04:52:02] what's it set to? [04:53:18] w/ ... actually, it works now but the short URL redirect to the longer one... is there any way to keep that adress short in the adress bar? [04:53:38] Eeek... [04:53:44] norelatives [04:53:56] proletaire: $wgArticlePath = "/wiki/$1"; [04:54:32] set that right under $wgScriptPath = "/w"; [04:54:58] it is set like that [04:55:22] Bah... then give out the right info when questions are asked [04:57:02] Thank you guys [04:58:27] 03river * r39380 10/trunk/tools/switchboard2/ (process_factory.cc process_factory.h version.h): [04:58:27] V-2.0.1 [04:58:27] re-implement missing features: [04:58:27] - kill old php-cgi processes after 30 seconds [04:58:27] - max-q-per-user config option [05:10:07] 03river * r39381 10/trunk/tools/switchboard2/ (request_thread.cc version.h): [05:10:07] V-2.0.2 [05:10:07] provide proper error messages instead of just closing the server connection [05:12:27] 03(NEW) Signature fails link to user_talk - 10http://bugzilla.wikimedia.org/show_bug.cgi?id=15170 15enhancement; normal; MediaWiki: User interface; (AheunHana) [05:17:56] 03river * r39382 10/trunk/tools/switchboard2/main.cc: ignore SIGPIPE [05:39:28] 03(NEW) New added pages aren't visible in the CategoryTree - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=15171 major; normal; MediaWiki extensions: CategoryTree; (sebastian.zuercher) [05:40:09] Hi, is there any articles about setting up mediawiki with linux default smtp servcer postfix? I installed mediawiki in openSUSE. Enabled all the mail options, but my maibox still cannot receive any emails. btw, I didn't put any information about my mailserver. I assume mediawiki should use localhost:25 to send email by default. [05:41:01] correction: I assume mediawiki will use sendmail to send emails by default. I tried sendmail < mailbody.txt which works fine. [05:43:41] 03dantman * r39383 10/trunk/extensions/SemanticForms/ (12 files in 2 dirs): [05:43:41] Attempt fixing bug 14357. [05:43:41] Extension messages are now all loaded dynamically. [05:45:15] 03(mod) Semantic Forms short-circuts the new wfLoadExtensionMessages system and tests incorrectly - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=14357 (10dan_the_man) [05:55:31] 03dantman * r39384 10/trunk/extensions/SemanticForms/ (16 files in 3 dirs): [05:55:31] Code style fixes: [05:55:31] * Remove remaining ?>'s [05:55:31] * Correctly indent Language classes [05:55:31] * Fix random leading spaces strewn around the code. [05:55:39] hi all [05:55:46] long time since ive been in here [05:58:12] question: on our categories page are many broken links [06:00:21] &action=edit [06:00:27] 03(NEW) Go button on Special:RecentChanges and Special:Watchlist - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=15172 trivial; normal; MediaWiki: Page rendering; (MZMcBride) [06:00:34] is placed at the end which breaks it [06:10:36] 03(mod) Retiring the developer user group - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=12569 (10N/A) [06:11:12] TimStarling: Could you comment on https://bugzilla.wikimedia.org/show_bug.cgi?id=12569 ? [06:11:47] 03dantman * r39385 10/trunk/extensions/SemanticForms/includes/ (SF_GlobalFunctions.php SF_ParserFunctions.php): [06:11:47] Finishing up applying MediaWiki's scalability standards to Semantic Forms: [06:11:47] * Move $wgExtensionCredits out of the way [06:11:47] * Autoload classes [06:11:47] * Use a StubObject for the $sfgFormPrinter global. Now we can autoload the SFFormPrinter object. [06:11:49] * Use ParserFirstCallInit so that SF parser functions are correctly applied to all parsers and they are offloaded. [06:19:23] 03river * r39386 10/trunk/tools/switchboard2/ (11 files): [06:19:23] V-2.0.3 [06:19:23] - support I/O timeouts [06:19:23] - support limiting the maximum request size [06:19:23] - don't try writing more than 64KB in one fcgi record [06:21:03] 03river * r39387 10/trunk/tools/switchboard2/util.cc: missing file [06:22:12] 03river * r39388 10/trunk/tools/switchboard2/request_thread.cc: fix I32LP64 compile [06:37:29] 03river * r39389 10/trunk/tools/switchboard2/ (fcgi.cc request_thread.cc request_thread.h version.h): [06:37:30] - nicer errors when a request fails [06:37:30] - remove debugging code [06:38:52] 03river * r39390 10/trunk/tools/switchboard2/ (21 files): svn:keywords, svn:eol-style [06:46:00] 03dantman * r39391 10/trunk/extensions/SemanticMediaWiki/ (16 files in 2 dirs): Autoload SMW_Language so that we do not need to include it in every single language file. [06:47:17] 03dantman * r39392 10/trunk/extensions/SemanticMediaWiki/includes/SMW_GlobalFunctions.php: *BOOM* Wrong filepath [06:55:45] 03dantman * r39393 10/trunk/extensions/SemanticMediaWiki/ (16 files in 2 dirs): [06:55:45] Undo my autoloading changes. [06:55:45] :/ SMW does direct inclusion of an arbitrary language file early on (even before [06:55:45] enable semantics is called where the AutoLoading is set [this is Poor Design [06:55:45] IMHO]) meaning that it is impossible to autoload the languages. [07:03:35] Hi all - can anyone help me hide tabs above pages (on monobook) for anon users not logged in? solution in FAQ at http://www.mediawiki.org/wiki/Help:FAQ#How_do_I_remove_the_article.2Fedit_etc_tabs_for_users_who_are_not_logged_in.3F does not work [07:03:46] Hi, I'm having some problems getting MediaWiki to cooperate with imagemagick. [07:04:06] When it tries to generate a thumb, I get this instead: [07:04:08] Error creating thumbnail: convert: unable to access configure file `colors.xml'. [07:04:21] (I've looked that file, it doesn't exist, and I can't find anything in the docs about it). [07:04:37] convert: no decode delegate for this image format `/Users/Audacitor/Sites/wiki/images/e/e8/Xcodeicon.png'. [07:04:48] (delegates.xml does exist, however). [07:05:01] convert: missing an image filename `/Users/Audacitor/Sites/wiki/images/thumb/e/e8/Xcodeicon.png/180px-Xcodeicon.png'. [07:05:46] (I think this is essential a 404; it couldn't find the thumb to display, which makes sense since the thumb was never generated). [07:07:22] 03dantman * r39394 10/trunk/extensions/SemanticCalendar/includes/SC_ParserFunctions.php: Use ParserFirstCallInit in Semantic Calendar [07:10:04] 03dantman * r39395 10/trunk/extensions/SemanticCalendar/includes/SC_GlobalFunctions.php: Autoload the only class in the extension [07:21:44] 03river * r39396 10/trunk/tools/switchboard2/ (6 files): [07:21:44] V-2.0.5 [07:21:44] - don't use lots of CPU doing nothing in main() [07:21:44] - a resource leak could occur if a failed failed. make it not occur. [07:38:51] 03siebrand * r39397 10/trunk/extensions/MetavidWiki/languages/MV_Messages.php: [07:38:51] * lower case [07:38:51] * revert removals [07:48:41] 04(REOPENED) Make Special:Watchlist look and behave like Special: RecentChanges - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=15157 +comment (10siebrand) [07:52:07] <_wooz> lo [07:57:24] Duesentrieb, you were so kind to respond yesterday - i still seem to be stuck on mediawiki upgrade & character sets ;) [07:57:30] some pages have their titles mangled [07:58:01] that's bad... how did you import the dump? [07:58:04] i tried some tricks for mysql to convert tables/columns to utf8 (like converting it to binary first, to utf8 later), but that still failed [07:58:37] plain mysql import from commandline [07:58:51] mysql has was to *convert* or to *redefine* - you deinitly should NOT convert, because it's already utf8. [07:58:59] "plain" how exactly? [07:59:09] as far as i understand, latest mediawiki with a recent mysql should work just fine with utf8 data types, and is in fact preferred [07:59:17] mysql < dump :) [07:59:41] and yes, converting to binary first was supposed to just redefine [07:59:57] then applying utf8 to it still fails with the mangled string & duplicate key error [08:00:51] latest mediawiki with a recent mysql should work just fine with utf8 data types, as long as you do not have 4 byte utf8 codes. which are still not supported by mysql. [08:01:52] it looks like mw has a mix of latin1 encoded utf and non-utf data... [08:02:05] it breaks on Equipment/ string [08:02:09] if anything is encoded as latin1, it's broken [08:02:22] TimStarling: What do you think about rewriting replaceInternalLinks to use the preprocessor and work similar to braceSubstitution? [08:02:26] mediawiki does not use lating1 for anything, though it might declare some columns as latin1 for compatibility reasons [08:02:36] but it does not use latin1 since 1.5 [08:02:42] not for anything. ever. [08:02:49] yeah. i have a dump from 1.5.6 ;) [08:03:06] so if you see latin1 codes, it was mangeled on export [08:03:18] don't rewrite replaceInternalLinks until I've committed my changes to it, ok? [08:03:29] oh heh... [08:03:58] Well... not like my local edits changed much there... [08:04:52] I'm trying to introduce a new concept... Link Hooks... A lot of extensions are starting to extend the link syntax... and normally it leads to substandard implementation, or a regex that causes server errors. [08:06:33] Richlv: try using --default-character-set=latin1 with mysqldump as described in http://www.mediawiki.org/wiki/Backup - that's the only thing i can think of. [08:06:43] Duesentrieb, thanks, will try that [08:06:56] the most important thing with the parser is backwards compatibility [08:07:00] 03siebrand * r39398 10/trunk/extensions/intersection/DynamicPageList.php: (bug 14971) Added method for alphabetical ordering ('alphabetical') (patch by Ramac) [08:07:03] Richlv: also... if you converted from 1.4 previously, is it possible that that conversion dodn't work correctly, and you have broken titles from that? [08:07:21] the syntax has to be precisely the same in every little edge case [08:07:37] 03(NEW) API Query Links not db agnostic - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=15173 normal; normal; MediaWiki: API; (overlordq) [08:07:43] 03(mod) PostgreSQL/pgsql support (tracking) - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=384 (10overlordq) [08:07:52] 03(FIXED) alphabetical order method for DynamicPageList - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=14971 +comment (10siebrand) [08:08:01] Duesentrieb, as far as i know, it was converted from 1.4 - but titles work fine in the existing 1.5.6 installation [08:08:04] mhmm [08:08:50] obviously there are a few problems with that if you want to take the links into the preprocessor [08:08:59] Richlv: hm... well, try that lating1 trick then. [08:09:18] Richlv: fyi, here's the bug report on mysql not supporting full unicode: http://bugs.mysql.com/bug.php?id=14052 [08:09:22] Hmm... well maybe an offshoot [08:09:25] the preprocessor, after all, is called a preprocessor because it happens *before* something [08:09:31] will not be fixed before mysql 6, it seems. [08:10:25] TimStarling: Well, ParserBeforeStrip and all the other plaintext hooks still work fine even with your new preprocessor... there are no hooks working directly with links... [08:11:06] I'm not talking about hooks [08:11:07] Duesentrieb, thanks. what characters would require 4 byte utf8 encoding ? [08:11:15] I don't mind breaking extensions, I just don't want to break wikipedia [08:11:26] Richlv: rare stuff. gothic, some obscure chinese... [08:11:51] Richlv: the problem is that people *can* input it. and if they do, bad things happen. [08:12:35] not sure if they'd just get stripped... when trying to import a dump with those chars, i had mysql fail on me before. [08:12:58] Well, when you preprocess something, you usually expand off that? I only see the difference being turning links into nodes then expanding rather than mucking with a pile of text... either way they should end up treated the same? [08:13:15] Richlv: acording to the bug report, "utf8 truncates the string at the point the char appears" - which means data loss. [08:14:53] the difference is that preprocessing allows you to create links which didn't syntatically exist in the first place [08:15:17] hmmm? [08:15:21] {{openlink}} link {{closelink}} [08:15:34] where openlink = "[[" and closelink = "]]" [08:15:37] that's the simplest example [08:16:21] O_o I thought the preprocessor was created to stop that kind of stuff, not create it? [08:17:23] Duesentrieb, ok, we got relatively common cyrillic broken, so i guess it's not that issue :) [08:17:50] probably not [08:17:58] Well I guess I'm looking for something similar to the preprocessor then... just a small subprocessor for replaceInternalLinks [08:18:01] stop it? [08:18:14] Richlv: but... *some* nonenglish titles are ok, but others are garbled?... [08:18:28] no, the preprocessor was created to be exactly the same as a section of the old parser, except faster [08:19:12] Duesentrieb, i managed to break the db with alter table stuff ;), but i think all i saw were broken [08:20:39] did you verify the dump contained valid utf8? [08:20:51] you can do that by runnign an utf8-to-utf8 conversion using iconv [08:21:28] Duesentrieb, at least non-english parts were ok, will run it through iconv like that right now [08:22:15] nope, that fails - so i guess it contains something fishy [08:24:40] even the dump created with --default-charset-whatever? [08:24:43] 03(mod) User preferences: Date & Time - 10http://bugzilla.wikimedia.org/show_bug.cgi?id=15169 +comment (10niklas.laxstrom) [08:24:45] that's bad... [08:26:32] Duesentrieb, no, i haven't yet gotten such a dump, the person who manages the db has not yet responded today (might be sleeping or something, i think we have different timezones :) ) [08:28:03] 14(INVALID) Signature fails link to user_talk - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=15170 +comment (10lejonel) [08:31:15] TimStarling: ^_^ I think I love the current way the parser handles obscure links http://dev.wiki-tools.com/wiki/LinkHook#Old_Tests That works perfectly with my concept... [08:37:49] 03river * r39399 10/trunk/tools/switchboard2/ (6 files): [08:37:49] V-2.0.6 [08:37:49] - properly handle errors in the child php, should reduce the number of FastCGI protocol errors [08:37:49] - up the listen backlog a little [08:39:20] 03(mod) maintenance/updateSpecialPages.php fails with PostgreSQL - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=14414 +comment (10overlordq) [08:46:42] TimStarling: Can you comment on the last testcase here? http://dev.wiki-tools.com/wiki/LinkHook#Old_Tests [08:47:20] I can understand not having a media embed within another media embed... but is killing both of them the optimal way to handle this here? [08:51:37] heyho, is it possible to include the text [[Category:Foo Bar]] verbatim in a page? [08:52:07] sure [08:52:15] [[Category:Foo Bar]] [08:52:15] ooor [08:53:05] No entry for terminal type "tty"; [08:53:05] using dumb terminal settings. [08:53:05] [[Category:Foo Bar]] [08:53:10] ... [08:53:17] ignore me :) [08:53:20] lol [08:53:22] cool, tnx :) [08:53:35] and sorry for being a lazy piece of sh*t :) [08:54:28] ^_^ If you're going to be lazy in the domain of links... at least give me a good testcase for stupid syntax that parser changes need to keep the same... [08:56:02] Dantman: that's the way it is [08:56:02] Dantman: Has anybody ever challenged you not to type '^_^' for a day? :) [08:56:12] heh [08:56:13] hmm, waht about [[Category:Foo Bar|baz]] as a silly attempt to link to a category page? :) [08:56:19] [[/stupid_is as/stupid#does.|what he said]] [08:56:27] gmc: [[:Category:Foo Bar|baz]] [08:56:40] to link to a category, instead of using it, prefix the link with a colon, as I did in that example. [08:56:51] i knew that! [08:56:57] image link inside image link doesn't work [08:57:16] ie: Technical limitation... or... it should never work? [08:57:20] or was the problem the third nested link? [08:57:34] I know, don't embed an embed inside of an embed [08:58:18] I'm meaning, rather than [[Image:Filename.ext|[[Image:Filename.ext|[[link]]]]]] outputting the Image: stuff verbatim and the [[link]] being linked [08:58:39] you're asking hard questions [08:58:40] Can't the first Image: be an embed, the [[link]] a link, and the second Image: left verbatim [08:59:11] you have to do a lot of research if you want to change the parser [08:59:20] the research probably takes longer than the coding [08:59:42] it's not something I can just do for you on IRC [08:59:48] mhmm [08:59:58] well, I could, but it would take a long time [09:00:04] heh [09:00:14] yknow, Tim, when I helped ya debug the preprocessor rewrite I had insomnia for a month after? [09:00:20] and I never even touched the php, just the wikicode [09:00:26] Dan: avoid the parser, go do something safer [09:00:26] What are the pieces of things to research? [09:00:30] heh [09:01:46] Eeeekkkk... [09:02:01] http://dev.wiki-tools.com/wiki/LinkHook#Old_Tests 2nd to last... [09:02:02] Research means looking at how things are used. [09:02:07] before you can change the parser, you have to know what the syntax is [09:02:11] Right... I believe that was a bug report [09:02:12] so that you know whether you're changing it or not [09:02:41] you have to know how the current parser works and why it's written that way [09:03:14] has anybody else noticed quite how odd the inheritance of Skins is? [09:03:59] ^_^ Which is why I'm asking... is the screwed up handling of nested embeds on purpose, or due to technical limitations? [09:04:11] Werdna: you mean SkinTemplate? [09:04:24] Dantman: when would you like an answer by? [09:04:35] heh... dunno [09:04:39] bah... i'm afraid my skin revamp has grown stale... need to redo that :/ [09:04:59] Time to track down link related bugzilla reports [09:05:27] I recal there was one directly on this screwed up image syntax... ^_^ I believe they wanted it fixed... rotlf [09:06:14] TimStarling: well, it looks like Skin inherits from SkinTemplate, which inherits from Linker [09:06:46] no, SkinTemplate inherits from Skin [09:07:00] class SkinTemplate extends Skin { [09:07:22] ah, so skins themselves inherit from SkinTemplate, which inherits from Skin, which inherits from Linker? [09:08:37] SkinTemplate should actually be called TemplateSkin [09:08:51] the misnomer has been bugging me for years now [09:09:13] the odd thing is that SkinTemplate doesn't even use a template [09:09:34] well, ideally we'd have a separate Linker class, loaded on demand from $skin->getLinker(), and a single parent class to inherit from to define a skin. [09:09:41] so I think SkinTemplate should just not exist at all [09:10:05] right, a single skin class which you inherit from [09:10:11] I'm not sure why we have Linker [09:10:33] domas introduced that one, iirc, he never really explained it to me [09:10:56] btw tim, somebody just asked me about disabling Special:CentralAuth / Special:GlobalGroupPermissions / Special:GlobalGroupMembership everywhere except meta, to make sure logging is in the same place. [09:11:20] not right now [09:11:40] have to fix the parser [09:14:35] 03mkroetzsch * r39400 10/trunk/extensions/SemanticMediaWiki/ (README includes/SMW_GlobalFunctions.php): Modified parser function registration as per DanTMan's suggestion + added Daniel to the contributors' list [09:15:42] http://xkcd.com/463/ <--- and it's *true*! [09:20:03] what's that function I use to clean up a comment? [09:20:38] skin->commentBlock( [09:20:39] that's the one [09:29:48] hello all [09:36:29] 03(mod) maintenance/updateSpecialPages. php tries to INSERT invalid values - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=14414 summary; +comment (10overlordq) [09:49:06] !bugzilla [09:49:06] --mwbot-- All bugs in MediaWiki should be reported at https://bugzilla.wikimedia.org. This is also the place to request site configuration changes, new features or enhancements to existing features, although bear the following points in mind before making a feature/enhancement request: 1) If the request is specific to a Wikimedia wiki, please discuss the issue on that wiki first. 2) Consider whether a custom extension would be more appropriate. [10:02:45] TimStarling: wasn't me [10:03:00] webresponse is me [10:03:01] webrequest is me [10:03:10] sorry domas [10:03:17] linker isn't me! :) [10:03:18] thats ok! [10:03:24] people blame me everywhere! [10:04:22] with great ego comes great culpability [10:07:58] *Werdna throws rocks at Splarka [10:08:42] I am a billion times more modest than you! haha [10:08:48] [10:09:02] In fact, I'm the LEAST competitive, so I WIN. [10:18:34] *Splarka pouts, moves to Penultimateville. Population: 1 [10:29:13] > $t = microtime(true); foreach ( explode( "\n", $test ) as $elt ); print microtime(true)-$t [10:29:13] 1.2810978889465 [10:29:13] > $t = microtime(true); foreach ( new ExplodeIterator( "\n", $test ) as $elt ); print microtime(true)-$t [10:29:14] 9.4570820331573 [10:29:26] is that cool or what? [10:29:56] eh? [10:30:04] PHP only 7.4 times slower than C, and uses 200 times less memory [10:30:41] they both do the same thing, but the ExplodeIterator class (which I just wrote), doesn't store elements as a hashtable [10:30:47] it just iterates through the string using offsets [10:33:31] > $t = microtime(true); substr_count($test, "\n"); print microtime(true)-$t [10:33:31] 0.031369924545288 [10:34:03] see, I can use substr_count(), which is insanely fast, to determine whether to use explode() or ExplodeIterator [10:34:18] and so choose the optimal trade-off between size and speed [10:34:29] awww [10:34:31] that's cute [10:34:38] foreach ( StringUtils::explode("\n",$lines) as $line ) { [10:34:59] hm [10:35:03] why there isn't some extension that adds memory efficient algorithms and datatypes to php? [10:35:03] I can use this code [10:36:18] TimStarling: ^_^ A third preprocessor? Could you also come up with a method of editing the internals which suffer from utter code duplication... [10:36:45] Don't you wish PHP had multiple inheritance [10:36:45] who's writing preprocessors? [10:37:07] what's that? what's this little project for then? [10:37:22] this is reducing memory usage of existing parser code [10:37:35]