tag? [14:48:23] McCormack, not if it's cached. If you miss the cache, then every extra byte of text that must be parsed will take a little extra time at least. There are limitations in place to keep this within reasonable bounds. If you notice a problem with a particular page, simplifications may speed it up. [14:48:40] That is, simplifications on your part to the markup used. [14:51:42] Good morning. Is there a decent way to indent the first line of a paragraph other than using 5 times? I'm not much of a CSS person, so if there's a way to do it with that, I'm not familiar with it. [14:52:13] jcw, p { text-indent: 3em; } [14:52:35] jcw, you may also want to kill the margins in that case, like p { text-indent: 3em; margin: 0; } [14:53:13] <^demon> jcw: http://www.w3.org/Style/Examples/007/indent - has a good overview of the subject, too. [14:54:59] Shouldn't it be possible to use ReCaptcha and ConfirmEdit together? [14:55:10] is it possible to grab revisions made by anons based on a network/mask in sub-linerar time? [14:55:20] seb_zemann: We should really integrate ReCaptcha's audio captchas into ConfirmEdit [14:55:35] AzaTht: No, not without a schema change storing the IP in binary form [14:55:40] RoanKattouw, it's closed-source. [14:55:43] *snyft* [14:55:58] Schema change!!!! [14:56:02] RoanKattouw: So it's not possible yet? [14:56:07] Simetrical: I know, which is why we need our own implementation [14:56:13] AzaTht, well, you'd have to expand it to the nearest octet. It would only be linear in the number of entries beginning with those octets. [14:56:15] No, and it probably never will be [14:56:20] RoanKattouw, or find another OSS implementation. [14:56:28] AzaTht: ucuserprefix [14:56:29] PostgreSQL has an IP address storage type! [14:56:31] :D [14:56:34] Not linear time, I know [14:56:42] RoanKattouw: Ah, okay. And what does mediawiki.org use? ConfirmEdit? [14:56:47] Simetrical: INT(4) will do too [14:56:51] Yep [14:56:59] RoanKattouw: Thank you! [14:57:01] ConfirmEdit (since ReCaptcha isn't open source, MW can't use it) [14:57:17] RoanKattouw, INT(4) means an INT that displays as four characters long when MySQL does pretty-printing. [14:57:30] The number doesn't affect the storage for integer types. [14:57:38] INT is always 32-bit, BIGINT is 64-bit. [14:57:41] Oh [14:57:46] Well anything 32-bit will do [14:57:51] So INT [14:57:57] For IPv4 it will. :) [14:57:58] Can you AND INTs in MySQL? [14:58:04] Simetrical: in a good layout, you should be able to be able to get information if ip has edited in O(M log N), where M is the number of IP's [14:58:21] For IPv6 you need BINARY(16) or something. [14:58:23] Simetrical: thanks! [14:58:23] AzaTht: IF an IP has edited? i.e. LIMIT 1? [14:58:29] s/good/ok/ [14:58:35] RoanKattouw, bitwise operations don't use indexes. [14:58:41] B-trees can't really support them. [14:58:44] 03(NEW) Diff error in it.wikisource - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13704 normal; normal; MediaWiki: History/Diffs; (powerpdn) [14:58:50] At least not ones that are meant to index integers for usual purposes. [14:58:55] RoanKattouw: more ips in a network might have been editing [14:59:03] but not all [14:59:04] You have to design the index specially to allow bitwise operations to be efficient. [14:59:21] Simetrical: For a field that stores binary data, that's what you'd do, right? [14:59:27] a network index could be useful [15:00:17] Hey maybe store the octets in separate fields and INDEX(ip_1, ip_2, ip_3, ip_4) [15:00:27] So you can WHERE individual octets [15:00:46] there is lots of spamming happening on MediaWiki.org !! [15:00:58] 67.52.164.132 [15:01:25] Been blocked [15:01:29] ok [15:01:37] By Spacebirdy, see Special:Log/block [15:02:23] ok [15:04:07] RoanKattouw, what's what you'd do? [15:04:16] There's no reason to WHERE individual octets. [15:04:25] The second octet is worthless without knowing the first. [15:04:32] At least in IPv4. [15:04:38] Simetrical: I meant WHEREing 1, or 1&2, or 1&2&3 or 1&2&3&4 [15:04:54] You can do that with an index on the whole field. [15:05:00] B-trees will work for operations on prefixes. [15:05:12] So that's back to the old ucuserprefix stuff [15:05:18] <^demon> RoanKattouw: As opposed to WHERE 1 OR 2 or 3, etc....why not do a WHERE IN (1,2,3...)? [15:05:42] ^demon: I didn't mean AND, I meant WHERE ip_1=123 AND ip_2=45 [15:06:00] RoanKattouw, you can do WHERE ip LIKE '123.45.%' as well. [15:06:05] It will work as efficiently. [15:06:11] Provided it's properly indexed for that? [15:06:20] An index on (ip) is fine for that. [15:06:23] A B-tree. [15:06:24] Right [15:06:46] So rev_user_text LIKE '123.45.%' should work reasonably well, right? [15:06:49] Same as if you have multiple fields, you still need an index on (ip1, ip2, ip3, ip4) for it to work quickly. [15:06:49] ^demon, Simetrical, thanks. Now can I put the CSS strictly in the source of one page, or do I have to create a skin for that page? [15:06:49] Anyone know how to write colored text? (without having to do so in a table?) [15:06:54] RoanKattouw, assuming that's indexed. [15:07:16] Simetrical: There's an index on (rev_user_text,rev_timestamp), and I'm sorting by timestamp [15:07:25] jcw, you can use the style attribute to put it on individual elements. To make it hold for all paragraphs (or whatever) on your site, add it to MediaWiki:Common.css, or MediaWiki:Monobook.css. [15:07:32] RoanKattouw, it will break sorting. [15:07:34] And force a filesort. [15:07:42] That's unavoidable. [15:07:53] There's no indexing scheme that will solve it, at least not in MySQL. [15:08:05] Simetrical: Not if I do ORDER BY rev_user_text DESC, rev_timestamp DESC [15:08:19] RoanKattouw, in which case it's not sorted by rev_timestamp. [15:08:24] Exactly [15:08:27] Right. [15:08:28] Then it's fine. [15:08:35] !style [15:08:35] --mwbot-- To change styles for your wiki, go to one of the MediaWiki:xxx.css wiki page and put your custom styles there (sysop/admin rights required). MediaWiki:common.css is for all skins and should be used for content styles. MediaWiki:monobook.css is for the MonoBook skin (default), etc. For more information, see !skins and [15:08:40] And a lot less useful. [15:08:41] RoanKattouw: if the ip is stored binary, it's simple just to make WHERE ip => low AND ip <= high [15:08:43] Simetrical: That sucks though [15:09:05] Which is why I filed a bug to implement a (rev_timestamp, rev_user_text) index, but Brion said that wouldn't help me [15:09:11] Hm, I want the Captcha thing on every edit! It doesn't work. [15:09:12] I added: require_once( "$IP/extensions/ConfirmEdit/ConfirmEdit.php" ); on the LocalSettings.php file [15:09:14] And uploaded all three php files in the /extensions/ConfirmEdit/ folder... [15:09:15] There is no Captcha at all... [15:09:16] Adding WHERE rev_user=0 to cut down the number of rows scanned was reverted [15:09:19] RoanKattouw, it wouldn't. [15:09:25] Apparently I'm not yet smart enough to see how to set the style on a single page. [15:09:42] seb_zemann: See the settings in ConfirmEdit.php [15:09:52] RoanKattouw, basically, whenever you retrieve from a B-tree index on (a,b,c,...), you *always* retrieve in the order a ASC, b ASC, c ASC, ... or a DESC, b DESC, c DESC, ... [15:09:57] <^demon> jcw: What did you put in that individual page? [15:10:04] RoanKattouw, there is absolutely no way to ever retrieve in any other order no matter what you do. [15:10:05] CAPTCHAs are disabled for sysops and usually only shown when adding external links [15:10:17] ^demon, nothing yet :) I'm not sure how I need to do it. [15:10:38] Simetrical: That's why I wanted to flip the index around, so I could flip the ORDER BY too [15:10:38] RoanKattouw, if you have something like WHERE a = 1 ORDER BY b, then it can silently rewrite that to ORDER BY a,b, since ordering by a does nothing, and a filesort can be avoided. [15:10:47] RoanKattouw, but then you can't use it for retrieval. [15:10:53] Simetrical: nah, just stop use B-trees ツ you'll probably have enough memory anyway ツ [15:10:58] RoanKattouw, only prefixes of B-trees can be used for retrieval. [15:11:04] Argh [15:11:14] If there are no conditions on the first part, you're scanning the whole index. [15:11:16] As I said, it's impossible. [15:11:19] I meant it. :) [15:11:38] It might be possible if we could make up some custom type of index, but you can't, at least not in MySQL. [15:11:39] So you can't sort by rev_timestamp if it's not the first field, and you can't do a prefix search on rev_user_text if it's not the first field [15:11:43] In PostgreSQL, incidentally, you can. [15:11:47] RoanKattouw, right. [15:11:53] <^demon> jcw: Are you wanting it on all pages, just monobook pages, or just one page, or just for your user? [15:11:58] RoanKattouw: you'll have to poke domas [15:12:03] ^demon, just a single page. [15:12:10] I did that a few times a while ago [15:12:13] hehe [15:12:23] When I wrote that query [15:12:34] AzaTht, no he won't, I know what I'm talking about. :P [15:12:35] 03siebrand * r33130 10/trunk/extensions/AuthorProtect/ (AuthorProtect.i18n.php AuthorProtect.php): Fix indentation and EOL whitespace [15:12:50] Simetrical: talking about rewriting mysql [15:12:56] Oh. :P [15:12:58] <^demon> jcw: Try wrapping your content in a

tag with the styles applied (ie:

CONTENT CONTENT

) [15:13:07] Apparently adding WHERE rev_user=0, while reducing the amount of scanned rows, uses the (rev_user,rev_timestamp) index instead which hurts prefix search performance [15:13:11] No, just let's switch to PostgreSQL, and then we can do all sorts of magic! [15:13:20] Simetrical: though could it be possible to have an ip-index? [15:13:33] RoanKattouw, it means that it just scans all anonymous users' rows. [15:13:37] <^demon> Simetrical: I'll write a flat-file extension :-P [15:13:43] haha [15:13:44] RoanKattouw, for a small prefix, that's probably okay, but not for a long prefix. [15:13:48] Simetrical: Instead of all users' rows [15:14:16] RoanKattouw, using (rev_user_text, rev_timestamp) doesn't scan all users' rows, it scans only the subset that match the prefix. But then it sorts them. [15:14:23] And it has to scan *all* of them, regardless of the LIMIT. [15:14:30] "If you use ... LIKE '%string%' and string is longer than three characters, MySQL uses the Turbo Boyer-Moore algorithm to initialize the pattern for the string and then uses this pattern to perform the search more quickly." [15:14:32] Retrieving in sorted order means you can retrieve only up to the limit. [15:14:48] AzaTht: I'm doing LIKE 'string%' which is even better [15:15:00] AzaTht, yes, but it's still an index scan. Which is horrible. [15:15:09] It can't use that to cut down the number of rows retrieved. [15:15:14] AzaTht, Simetrical : User generators should solve this though [15:15:23] RoanKattouw, eh? [15:15:39] First do a query on the revision table with rev_user_text LIKE '123.45.%' [15:15:45] Oh wait [15:15:49] Argh [15:15:53] You'd have to do DISTINCT, which sucks [15:15:55] ^demon, I can do that. Although there's a number of paragraphs on the page that I'd like to apply it do. Is there no way to embed the CSS in the page itself to apply it automagically? [15:15:59] Right? [15:16:05] 03siebrand * r33131 10/trunk/extensions/Translate/ (MessageGroups.php Translate.php): [15:16:05] * add support for AuthorProtect [15:16:05] * fix sorting of supported extensions [15:16:07] Why would you want DISTINCT? [15:16:16] How bad would that be, rev_user_text LIKE '123.45.%' DISTINCT with no other conditions? [15:16:29] Because I can then do rev_user_text IN ('123.45.1.1', ...) [15:16:33] <^demon> jcw: Not sure offhand. [15:16:35] . . . what? [15:16:39] What fields are you retrieving? [15:16:50] First all IPs that match the prefix [15:16:52] Then their contribs [15:16:55] Two separate queries [15:17:01] You're still going to have to filesort. [15:17:08] 03siebrand * r33132 10/trunk/extensions/Translate/Translate.php: Where did that come from? [15:17:09] But the part that sucks is that we don't store those IPs in the user table [15:17:09] It's still a range scan on rev_user_text. [15:17:24] So subsequent parts can't be used for sorting. [15:17:27] Simetrical: Sorted by .... rev_user_text, rev_timestamp which is indexed [15:17:37] Or does the DISTINCT kill the index? [15:17:44] No, DISTINCT is irrelevant here. [15:18:00] If you sort by rev_user_text, rev_timestamp, that will never be a problem. [15:18:03] Oh you mean the IN part is bad [15:18:06] Yes. [15:18:16] *sigh* [15:18:21] The point is you cannot sort by rev_timestamp alone if your retrieval condition is a range scan. [15:18:30] That includes LIKE, IN, whatever. [15:18:55] Because of the silent rewrite of WHERE a = const ORDER BY b to ORDER BY a,b [15:19:03] Which you said earlier [15:19:06] Yes, which is necessary to avoid a filesort. [15:19:23] The reason you can avoid it in the first place is because you're retrieving in the same order as it's physically stored in the index. [15:19:26] Which is (a,b,c,...). [15:19:57] Shouldn't "$wgCaptchaTriggers['edit'] = true; " do the trick that every edit needs a captcha? [15:19:58] It dosn't work [15:20:58] 03catrope * r33133 10/trunk/phase3/ (12 files in 3 dirs): Committing patch for bug 10931, which also fixes bug 13651. For a detailed explanation of what this commit does, see bug 10931 comment #24 [15:21:06] 03(FIXED) Wikimedia redirect tables are missing many entries - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=10931 +comment (10roan.kattouw) [15:21:27] 03(FIXED) "Automatically resolve redirects"-feature causes error at fi. wikipedia.org. - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13651 +comment (10roan.kattouw) [15:24:38] <[exa]> hi..one more question [15:25:19] <[exa]> I made my own wiki skin (based off monobook), but the editbox buttons don't appear. No idea why, javascript code is there. [15:25:39] [exa]: Did you clear your cache? [15:26:18] <[exa]> If you have any hints what could be missing or where to start searching... [15:26:39] <[exa]> I refreshed cache several thousand times before I noticed it's missing. [15:27:05] [exa]: You did purge it with Shift+Refresh, didn't you? [15:27:13] Also, if you have Firefox, check your error console [15:29:04] A link would help, if someone is willing to look (I'm not). [15:29:40] <[exa]> if someone would be willing to help, tell me then, I'll upload the site somewhere... [15:29:45] <[exa]> ;D [15:29:54] <[exa]> ok, error console's clear, shift refresh tried. [15:30:46] seb_zemann: It should. Make sure it's in LocalSettings.php, AFTER the require_once line that includes ConfirmEdit.php [15:31:20] seb_zemann: Try logging out, maybe you're a sysop so you don't get CAPTCHAs [15:34:51] RoanKattouw: I tried logged out... [15:35:03] RoanKattouw: it could be a good idea to type in the commit message at least something about what has been fixed ツ [15:35:29] as you didn't had an explenation at all :-P [15:35:30] AzaTht: That was a 10-line list which is also at bug 10931 comment #24, people who wanna know what's been fixed can go there [15:35:58] hello [15:36:20] RoanKattouw: This line ($wgCaptchaTriggers['edit'] = true; ) should be in the LocalSettings.php as well ? [15:36:32] Yes, BELOW the line that includes ConfirmEdit.php [15:36:51] The lines in ConfirmEdit.php are *defaults*, do *not* edit them, but copy them to LocalSettings.php, then change them [15:37:10] *AzaTht want an O(log log N) search algorithm [15:37:10] If you don't do that, any upgrade of ConfirmEdit will overwrite your settings [15:37:17] lol [15:38:33] AzaTht: You know that systems claimed to be log (N) are really log(N) / log(2) = 3.32 * log (N) [15:38:46] Because it's usually 2log(N) [15:38:54] (Binary trees and all that stuff) [15:39:11] Oh hey wait that doesn't matter because 3.32 is a constant [15:39:16] Never mind xD [15:43:33] ^demon, the

worked. I didn't realize that I didn't have to apply it on a per-paragraph basis, but could

[all paragraphs]

. Thanks! [15:46:00] RoanKattow: Nothing... Hm. I loaded the old ConfirmEdit up, and edited the LocalSettings. [15:49:38] 04(REOPENED) Recentchanges feed doesn't show the editor's name/IP address - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13045 +comment (10danny_b) [15:56:26] 03ialex * r33134 10/trunk/phase3/ (9 files in 3 dirs): [15:56:26] * Ported tests from test/ branch to t/ branch [15:56:26] * Added maint option to Makefile, only does maintenance tests [15:56:26] * Added .htaccess file in t/ to disallow acces from the web [16:10:35] ialex: hmm, a user wants logs grouped by page again [16:10:39] I do to [16:10:46] was the only problem the creation log spam? [16:11:46] AaronSchulz: and another was requesting by log :) [16:12:03] requesting by log? [16:12:21] like now [16:12:31] (groupped by log) [16:13:30] anyway to just make it the newuserlog? [16:13:44] I mean the event *is* to the page, why not group it there? [16:14:47] but also when you delete and protect a page, it appears like there was two edit one same page, with a blue link for the page [16:15:03] *on the [16:17:10] imho only user creation and blocking should be grouped together [16:17:21] 03siebrand * r33135 10/trunk/extensions/Interwiki/SpecialInterwiki.i18n.php: [16:17:21] Some spelling errors and newlines. [16:17:21] No localisation updates needed. [16:19:01] yeah, for some it might be better to be groupped by page and others by log [16:19:06] *sigh* [16:19:39] a cute little global will do [16:20:40] hehe [16:23:17] 03(mod) Advertise locked wikis in Special:SiteMatrix - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13703 +comment (10cbrown1023) [16:40:36] 03ialex * r33136 10/trunk/phase3/includes/EditPage.php: Show edit comment as it will appear in history when using preview and section=new [16:48:35] hey domas [16:49:03] <[KrusheR]> gñap [16:49:11] <[KrusheR]> hmm wrong chanel [16:50:27] 03aaron * r33137 10/trunk/extensions/FlaggedRevs/ (FlaggedRevs.php FlaggedRevsPage.php): [16:50:27] *Small fix [16:50:27] *More performance tweaks [16:57:42] hiii [16:58:17] domas: cheerio [17:01:52] 03aaron * r33138 10/trunk/phase3/includes/SpecialContributions.php: Don't link to new page diffs [17:19:34] 03aaron * r33139 10/trunk/extensions/FlaggedRevs/FlaggedArticle.php: Better to just check these [17:21:29] <^demon> Has anyone ever had experience with using Metastorm for their business processes at work? [17:21:56] <^demon> s/experience with/the misfortune of/ [17:25:08] 03huji * r33140 10/trunk/phase3/languages/messages/MessagesFa.php: * Adding/updating Persian translations [17:25:21] <[KrusheR]> is there any way to resize the pictures? [17:25:35] <[KrusheR]> not making thumbnails but resizing the uploaded imagenes themselves [17:28:17] hey brion :-) [17:28:23] oy [17:29:19] oh, brion... [17:29:49] [KrusheR], download, resize on your computer, reupload. [17:29:50] All power to brion, our liberator! [17:30:23] <[KrusheR]> i'm fed up with users uploading pictures of their cats at 30000000x80000000 [17:31:03] that's pretty big [17:31:22] <[KrusheR]> yes, you can even see the cells of the animal [17:31:26] need to have an really big camera for that [17:31:47] <[KrusheR]> i readed the serial number on a snake once [17:32:15] those images must then have been terabytes in size [17:32:23] if not more [17:32:37] <[KrusheR]> I have a good raid [17:32:42] hehe [17:33:01] jpg? [17:33:13] <[KrusheR]> worse, bmp renamed to jpg [17:33:18] hahaha [17:33:22] That should be rejected by file type checking. [17:33:40] <[KrusheR]> bmp was allowed some time ago [17:34:41] <[KrusheR]> or i'm wrong and "MIME: image/x-ms-bmp" doesn't mean that? [17:36:39] <[KrusheR]> can it be possible that mediawiki isn't checking mime type on upload? [17:37:28] <[KrusheR]> i wish mediawiki had a batch image remover [17:40:15] brion: I'm thinking about r32267 [17:40:31] *AaronSchulz wants it back ;) [17:40:42] I restrict access for the "user" group to just read only and no other rights, but when they append "&action=edit" to the URL they can get access to the wiki code of the page! [17:40:52] how I can prevent this? [17:41:34] <[KrusheR]> did you use $wgGroupPermissions['*']['edit'] = false; in LocalSettings.php? [17:41:36] Hi, I'm running Mediawiki 1.12.0 and I can't get $wgUsersNotifiedOnAllChanges (http://www.mediawiki.org/wiki/Manual:%24wgUsersNotifedOnAllChanges) to work. After trying many things, I decided I'd look at the source. However, I couldn't find it: [17:41:37] wiki2# grep -ir "wgUsersNotified" . [17:41:37] ./LocalSettings.php:$wgUsersNotifiedOnAllChanges = array('cemsadmin'); [17:41:49] am I doing something stupid, or should I file a bug report? [17:41:52] [KrusheR]: yes, of course [17:42:13] <[KrusheR]> maybe they are inheriting permissons from another group? [17:42:33] <[KrusheR]> it should show the source code and not to allow editing [17:42:51] AaronSchulz: can you describe it in a sentence? [17:43:13] *Jack_Phoenix prods brion about svn accounts again ;) [17:43:13] 'Check extension of key too, for the off chance of collisions' [17:43:18] I think I found the problem [17:43:27] It is misspelled $wgUsersNotifed [17:43:32] OnAllChanges [17:43:37] in version 1.12.0 [17:43:38] I [17:43:41] [KrusheR]: I'm not using inheritance; I don't want it to show the source code [17:43:42] 'll fill out a bug report [17:43:53] AaronSchulz: well at minimum you need to normalize the extension [17:43:59] <[KrusheR]> ah, i understand now [17:44:12] brion: I'm more worried about integrity rather than catching all dups [17:44:32] <[KrusheR]> what's wrong with displaying the source code anyway? [17:44:38] <[KrusheR]> its just wikicode [17:44:40] two keys with the same sha1 are more likely to actually be the same if the ext is matched [17:45:12] when are two keys ever going to have the same sha1? [17:45:15] having a few redundant copies won't hurt [17:45:22] that seams a bit unlikely to be worrying about... [17:45:34] [KrusheR]: it's besides the point; different people have different requirements [17:45:46] <[KrusheR]> ok, sorry [17:46:05] <[KrusheR]> i think mediawiki has no permisson category to view source code... [17:46:06] [KrusheR]: sorry I don't mean to sound rude [17:46:07] brion: so, if I normalize it? [17:46:08] 03(mod) Enable FlaggedRevs on Russian Wikipedia - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13659 (10vladimir.medeyko) [17:46:17] <[KrusheR]> np [17:46:45] [KrusheR]: yeah, that's an issue for me; [17:46:48] <[KrusheR]> maybe you can alter the skin to not show the code when the user is anonymous? [17:47:11] I may have to modify the default core mediawiki code [17:47:27] hmm [17:48:05] brion: is it ok to re-add? [17:48:08] Whats the edit summary character limit anyone? [17:48:43] Reedy: 255 [17:48:47] [KrusheR]: that's a possibility; thanks [17:48:51] Wondered if it was around that, cheers [17:49:02] Reedy: well, actually iirc less than that for byte reasons [17:49:04] sorry [17:49:20] AaronSchulz: remind me what it does with the lookup? [17:49:31] <[KrusheR]> ping me if you need help modifying the skin [17:49:59] <[KrusheR]> i don't know if it's a good solution tough [17:50:01] when restoring a file, if other deleted rows use that file, it knows to leave the file [17:50:15] otherwise, it deletes it after copying to public dir [17:50:40] well, not "deleted rows", but filearchive/oldimage rows [17:51:02] [KrusheR]: it's a hack, but it doesn't touch the core code [17:51:58] brion: also, if there is a bad collison, at least the file is *likely* to be of the same type ;) [17:52:16] rather than going to a .jpg image and getting an .ogg [17:54:42] hmm [17:54:45] well use the whole key i guess [17:56:36] AaronSchulz, there's no point in worrying about SHA1 collisions. You may as well worry about meteors striking the data center. [17:57:03] It's a waste of time and LOC. [17:57:09] <^demon> Simetrical: Could that happen?!? Do we have meteor insurance?!? [17:57:22] ^demon, well, someone will probably have a DB dump!! [17:59:14] "For a detailed explanation of what this commit does, see bug 10931 comment #24" [17:59:17] bah [17:59:34] cut-n-paste doesn't work, catrope? :) [17:59:51] brion: told him that before [18:00:05] [17:35:30] AzaTht: That was a 10-line list which is also at bug 10931 comment #24, people who wanna know what's been fixed can go there [18:00:10] thx AzaTht :D [18:00:24] 1) cut-n-paste is easiest [18:00:29] 2) URL is second-best [18:00:44] 3) vague description of where to find the details -> bad [18:00:47] brion: it's ktnhxbye ツ [18:00:54] 10-line commit messages are fine. [18:01:13] if i'm some random guy trying to figure out why this commit broke something, i want as much information as possible [18:01:28] if i have to do another google search to find the bug tracker, that's kinda a pain :) [18:02:21] how long should i wait for a patch in r33133 to take effect on en.wp? [18:02:27] it was committed about three hours ago [18:02:43] brick_ijigg: up to a week [18:02:46] is there a specific place in mediawiki where i place my js? [18:02:52] depending on how much poke is given [18:02:55] AzaTht, whoa, OK [18:03:09] any guidelines for determining/monitoring the status? [18:03:12] and how many waffles is donated [18:03:14] brion: hmm, there are only like 10 extensions allowed or so, so it may just go from something funny like .0000000000000000001% to .00000000000000000001% ;) [18:04:39] brick_ijigg: perhaps you get better response from them in #wikimedia-tech ツ [18:04:39] AzaTht, i'll offer a dozen waffles if it helps :) [18:04:41] when I think of it like that, bah skrew it [18:04:41] ahh, OK [18:07:45] brion: yeah, it would better to just catch more redundant files that get that small of a gain in bad collision reduction [18:08:25] well, just be consistent in the key usage with the regular deletion [18:08:46] well that would mean committing the change [18:08:49] brion: what should I do? [18:08:59] I need to have waffles. [18:09:03] I haven't had any waffles in a really long time. [18:09:47] brion: deletion currently checks the whole key [18:10:11] sure, then do it :) [18:10:23] just use the extension normalization func to make sure it's right [18:10:32] the key is hash + normalized ext [18:11:04] (so Flipper_baby.JPEG -> 842798374987234.jpg) [18:11:55] hi [18:11:59] I'll use File::normalizeExtension [18:12:34] hi [18:13:04] Simetrical: have you considered some way to update the category table count, for cases where it gets inaccurate? [18:13:04] MediaWiki:Sidebar is the bit above the search bax [18:13:06] *box [18:13:19] isn't there a way to edit the bit between the search box and the toolbox [18:13:26] like, when it says there are fewer than 0 pages [18:13:39] MrZ-man, I don't want to potentially trigger five-second select queries to recalculate huge categories' values, on ordinary page views. [18:13:58] well, if it thinks it's <0 it's _probably_ a small value [18:13:58] It might be tolerable, though. [18:14:03] Especially if it thinks it's negative, yeah. [18:14:18] So should I add a check for that, and refresh if it's negative? It would be easy enough. [18:14:50] It kind of sucks for fast-changing but small categories like prod or whatever. [18:16:20] well, they shouldn't go negative to begin with :) [18:17:32] anyone able to help me? [18:18:45] 03aaron * r33141 10/trunk/phase3/includes/filerepo/LocalRepo.php: Be consistent with filearchive system and use whole key. This is a modified version of 32267. (talked to brion about this, who reverted it earlier) [18:20:27] 03(mod) Templates used inside tags aren' t listed at the bottom of the page on preview - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13672 (10huji.huji) [18:23:01] AaronSchulz: re r33138, the contribs diff link for new pages [18:23:20] the problem i see is that the diff link is the only link we have in there that goes directly to the particular revision [18:23:33] so for a new page, we no longer have any link to the revision [18:24:17] a diff view of the first rev will show us that revision, plus navigation links we can use to go rev-by-rev watching the next revisions [18:24:25] so i'd tend to consider we should keep it [18:24:50] yeah, I thought about that some, but wasn't sure how likely it was to have changed since [18:24:59] maybe better to keep them [18:25:12] i'd say keep em, yeah [18:25:37] brion: maybe they should have a different message? [18:25:45] 'diff' doesn't make much sense [18:25:53] well, it's the diff view [18:25:55] with the diff navigation [18:26:02] so i don't mind it saying 'diff' :) [18:26:23] I don't either, as a dev ;) [18:26:34] as a user, it is kind of wierd [18:27:21] where do u put your javascript file on mediawiki? [18:27:35] 03aaron * r33142 10/trunk/phase3/includes/SpecialContributions.php: Revert self - diff links gives nice perma-link [18:28:09] ok, lunch time [18:30:05] hello [18:30:14] !js | jtong [18:30:14] --mwbot-- jtong: I don't know anything about "js". You might try: !javascript [18:30:20] !javascript | jtong [18:30:20] --mwbot-- jtong: To set up custom javascript on a wiki, edit [[MediaWiki:Common.js]] as a sysop. If $wgAllowUserJs is true in LocalSettings.php, you may also edit the /monobook.js (assuming your skin is monobook) subpage of your userpage. [18:30:26] !js alias javascript [18:30:26] --mwbot-- Successfully added alias: js [18:31:29] Simetrical: thanks, this information seems to be very hard to find on google [18:32:33] is there a MediaWiki: page to insert arbitary html somewhere in the sideabar? [18:35:45] !sidebar [18:35:45] --mwbot-- To edit the navigation menu on the left, edit [[MediaWiki:Sidebar]] using its special syntax. For more details, see . [18:35:57] Not arbitrary HTML. [18:36:06] We try to avoid that in messages (although it's not totally eradicated). [18:36:14] Simetrical: hehe [18:37:27] Simetrical: right [18:37:35] i know how to edit the standard links [18:37:39] Simetrical: ok i found out i dont have admin access [18:38:00] http://silver0708.scribblewiki.com/Main_Page -> how has that .co.nr link been done? [18:38:05] Simetrical: is there a way i can set up my account to be an admin account? the guy who set up our wiki is gone now [18:38:13] jtong: for user js, see User:Foo/skin.js [18:38:22] jtong, do you have database access? [18:38:32] If so, you can add a row to user_groups. [18:38:33] Simetrical: yes i do [18:38:58] jtong, something like this should work: INSERT INTO user_groups( ug_user, ug_group ) VALUES ( 12345, 'sysop' ); [18:39:03] Where 12345 is your user id. [18:39:12] Which you can get with SELECT user_id FROM user WHERE user_name='My name'; [18:40:45] Simetrical: btw, i have to copy and paste all the js into Common.js? there must be a more elegant way of orgainizing multiple js [18:41:23] jtong, you can use document.write() to include extra JS. [18:41:43] document.write( '' ); [18:46:44] anyone here able to help with the addhtml extension? [18:48:00] addhtml? deprecated: I've got a better one: SecureHTML [18:48:35] well, ive fixed it anyway [18:48:50] fixed it? [18:48:51] and its on a wiki-farm, so i can't decide that [18:49:01] jldupont: got it working [18:49:21] 03(mod) Add non-breaking spaces in additional places automatically - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13619 (10brion) [18:49:52] 03(mod) Use of templates in intersections - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13692 (10vacuum) [18:50:31] 03(mod) #iferror should catch imagemap errors - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13620 (10brion) [18:52:30] 03(mod) Templates used inside tags aren' t listed at the bottom of the page on preview - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13672 (10brion) [18:52:40] 03(mod) Add non-breaking spaces in additional places automatically - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13619 (10Simetrical+wikibugs) [18:53:22] 03(mod) DPL2's include parameter broken with new parser - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13677 summary (10brion) [18:54:25] 03(mod) block message in centralauth extension - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13697 04BLOCKER->04CRIT (10Simetrical+wikibugs) [18:59:42] brion: eeeek [19:00:21] <^demon> Question for anyone. In regards to bug 1037 (https://bugzilla.wikimedia.org/show_bug.cgi?id=1037)...would a patch changing all to and all to be a step in the right direction? [19:02:59] ^demon: no [19:03:08] carries semantic information which does not [19:03:11] same with vs [19:03:14] at least, that's the theory ;) [19:03:22] which is why xhtml strict is stupid [19:03:30] it forces you to either dilute the semantic information [19:03:36] or fill your code with [19:03:39] ugh [19:03:51] 03simetrical * r33143 10/trunk/phase3/ (RELEASE-NOTES includes/WebRequest.php): (bug 13690) Fix PHP notice on accessing some URLs. parse_url() in some versions of PHP doesn't set the path element if it's empty, rather than actually returning it as an empty string. [19:04:03] brion, it's been pointed out to me that XHTML 2.0 keeps and , although not [19:04:03] . [19:04:13] well, xhtml 2.0 is probably a dead-end [19:04:21] the buzz is about html 5 working group these days :) [19:04:26] 03(FIXED) PHP Notice: Undefined index: path in /var/www/w/includes/ WebRequest.php on line 82 - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13690 +comment (10Simetrical+wikibugs) [19:04:30] <^demon> brion: Then why were the words 'd and 'd to begin with if not to emphasize or bolden a particular concept? [19:04:54] ^demon, read a manual of style for the wide variety of ways they're used. [19:04:54] <^demon> I understand the semantic difference between the tags, of course. [19:04:56] There are *lots*. [19:05:07] The names of creative works are often italicized, for instance. [19:05:22] We originally had '' and ''' as and , but users of screen readers complained. :) [19:05:32] There's an old bug about it. [19:05:34] weee, dialog tags [19:05:42] Yeah, HTML 5 is pretty wacky. [19:05:45] <^demon> Simetrical: I see, I see. [19:05:52] there's all kinds of interesting semantic elements in html [19:05:56] It's *huge*. I'm not sure if that's a good thing. [19:05:59] and and :) [19:06:07] http://www.w3.org/TR/html5-diff/#open-issues [19:06:15] Certainly a lot more semantic info, but it's sometimes hard enough to figure out what tags to use as-is. [19:06:19] *^demon reverts the really really big patch he just made [19:06:27] hmm, 'progress' tags [19:06:32] sounds interesting [19:06:53] autofocus, sweet [19:06:58] no more JS hacks [19:07:08] AaronSchulz: does wgFlaggedRevsAutopromote work basically the same as regular Autopromote? [19:07:16] not at all ready ;) [19:07:53] if a user gets the rights through the autopromote, can they be manually removed? [19:08:06] yes [19:08:13] ooh, nice [19:08:34] you should fix normal autopromote to work like that [19:08:37] AaronSchulz, a ton of stuff in HTML 5, WebForms, etc. obsoletes ubiquitous JavaScript hacks. [19:08:45] It's fairly awesome. [19:08:56] of course, b/c will demand there use for a while [19:09:09] and in 2018 when ie 8 is dead, we'll be able to use these features [19:09:10] one think I like is autoupdate stuff is in browsers more [19:09:14] so the web updates faster [19:09:23] hehe, 2018 [19:09:28] that's a long way [19:09:37] Breaking away from SGML and especially XML was a great decision too. [19:09:52] "datagrid represents an interactive representation of a tree list or tabular data. " [19:09:56] html 5, i love you [19:10:18] I thought people said there are problems with html5 [19:10:48] people say a lot of things [19:10:58] "# [19:10:58] details represents additional information or controls which the user can obtain on demand. " [19:10:58] Awesome. [19:11:07] 03siebrand * r33144 10/trunk/phase3/languages/messages/MessagesEn.php: Oops. Forgot to put the namespace "Help:" back in. [19:11:07] there must be pros and cons? [19:11:17] it's rare to find something with all pros [19:11:29] blue_asterisk, it adds a ton of new features and will take forever to be implemented. [19:11:36] It's also much, much more complicated to understand and implement. [19:12:10] we'd have to support darn ie6 for a long while still [19:12:11] Plus, it's controversial because it's basically a rogue consortium breaking away from the direction the W3C had been taking, which a lot of standards fanatics defended reflexively. [19:12:21] <^demon> Simetrical: Will such a confusing system keep people from creating crappy websites? [19:12:26] AFAIK. [19:12:28] ^demon, no. [19:12:31] <^demon> Then it accomplishes nothing. [19:12:40] It will allow those of us who create good websites to do so more effectively. :) [19:13:31] "The value attribute for the li element is no longer deprecated as it is not presentational. The same goes for the start attribute of the ol element. " [19:13:35] WTF, those were deprecated? [19:13:41] *Simetrical stabs the XHTML Working Group [19:14:02] <^demon> Until I see browsers refuse to display the craptastic websites of the 90s (which still get made, oddly enough), I have seen no progress in HTML. Albeit granted, it's partially incompetent teachers who "educate" their students on making a website, thinking they know how themselves. [19:14:11] 03(mod) multiple CSS3 multi-column errors across multiple browsers - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13686 (10brion) [19:14:27] ^demon, don't be so elitist. A low barrier to entry is a good thing overall. [19:14:44] maybe an "html5 to html4.01 transitional compiler" would work, heh; ie6 doesn't have html5 support so we're stuck with html4 a long while still [19:14:58] blue_asterisk, you're thinking way, way too short-term. [19:14:58] <^demon> Simetrical: It's not about barrier to entry. I'm more than happy to have people learning HTML, PHP, what have you. As long as they learn it right. [19:15:20] blue_asterisk, HTML5 features will be usable without backward compatibility maybe ten years from now, if all goes well. [19:15:21] "right", hehe [19:15:30] ^demon, that's a very high barrier to entry. Learning it wrong is much, much easier. [19:15:42] blue_asterisk, maybe fifteen. IE6 is irrelevant. [19:15:48] No other browser supports it either, after all. [19:16:05] *blue_asterisk nods [19:16:27] Some bits of it are implemented now, it's true. [19:16:30] Like contenteditable. [19:16:55] "The draggable attribute can be used together with the new drag & drop API. " [19:17:01] "The irrelevant attribute indicates that an element is not yet, or is no longer, relevant. " [19:17:04] <^demon> Simetrical: It's just the little things that irk me...such as telling your students "It's ok to use uppercase HTML tags." Sure, it'll show up right... [19:17:05] o_O @ irrelevant [19:17:21] ^demon, HTML 5 allows uppercase HTML tags, I'm pretty sure. :P [19:17:31] It's only XHTML that required lowercase. [19:17:36] HTML 4 is totally valid with uppercase tags. [19:17:45] And it has practically all the features of XHTML, so there's no good reason to avoid it. [19:17:54] Some W3C pages are written in HTML 4 with uppercase tags. [19:18:24] Whoa, repeat tags? Freaky. I wonder how that works. [19:18:26] brion: will you feel like adding ar_parent_id anytime soon? [19:18:34] <^demon> Uppercase tags are harder to read :'( [19:18:43] *AaronSchulz will either add it to SVN or go to sleep now ;) [19:18:47] AaronSchulz, if we used PostgreSQL we could do live alter tables! [19:19:02] Simetrical: really? [19:19:12] AaronSchulz, yes. Well, at least it supports them. [19:19:44] There are some caveats, of course, but none seemed prohibitive when I looked it over. [19:19:48] Unlike some stuff in MySQL. [19:20:01] Like stored procedures not using the query cache, WTF. [19:20:04] *Simetrical trolls domas [19:20:07] anyone know of a basic wiki editing page i can freely copy and use for my help section? [19:20:20] AaronSchulz: i'd like to avoid more surprise schema changes this week :) [19:20:22] alaya, the entire "Help:" namespace of www.mediawiki.org is released into the public domain for this purpose. [19:20:34] Simetrical: awesome. thanks. [19:20:55] brion: ok, I'll sleep first add only add the sql stuff ;) [19:21:05] so why doesn't the querycache work? [19:21:15] Simetrical: so i can just cut and paste what i want out of it. do i need to do attribution? [19:21:20] alaya, no. [19:21:27] alaya, do whatever you want. There are no restrictions at all. [19:21:27] simetrical: lots of things don't use query cache [19:21:31] domas, :) [19:21:35] AaronSchulz: on wgFlaggedRevsAutopromote, what are "spacing" and "benchmarks"? [19:21:40] simetrical: on the other hand, how would SPs use query cache? [19:21:47] did you read the documentation page? [19:21:58] simetrical: query cache has similar problem with dependancies tracking as our parser cache [19:22:13] i have another question: is wiki editing and management sort of become a field in its own right now? [19:22:15] Er, I didn't mean stored procedures, I don't think. I meant prepared statements. [19:22:17] 03rotem * r33145 10/trunk/phase3/languages/messages/MessagesHe.php: Fix. [19:22:23] ah, PS, right [19:22:31] I've never used either, so I get them confused. :P [19:22:32] those should be cached in some new revisions [19:22:37] MySQL 5.1.something does, yeah. [19:22:45] But it's a fairly huge gotcha. [19:22:50] Lots of apps rely on query cache quite a bit. [19:23:35] Hmm, HTML 5 also removes . [19:23:37] Interesting. [19:24:00] <^demon> alaya: For what it's worth, a company wanted to hire me as a "wiki-based medias consultant" or some silly title. Don't know if that answers your question though. [19:24:54] ^demon: i guess maybe the answer would be that documentation maintenance in a broader sense probably is a field [19:25:01] hi. I would like to enable, if possible, automatic redirects toward wikipedia. That is, for certain titile page I'd choose, the user wouldn't need to type [[wp:Title]] but just [[Title]], still it will pop wikipedia, not an internal page :)]] [19:25:04] I wonder, does our getElementsByClassName() interfere with the HTML 5 one in Fx3 et al.? [19:25:27] <^demon> alaya: Yes, it would be. We hired someone recently at my employment solely to work on documentation management. [19:25:29] jd_, there may be a config option to do that, but it's kind of scary/disorienting. [19:25:57] yeah I know, but it would be great in my case ;) [19:26:03] Hurrah, innerHTML becomes standard too. [19:26:12] jd_, check the docs for a config setting for this. [19:26:16] I vaguely recall one exists, but not sure. [19:27:01] I did check on mediawiki.org and meta, but didn't find much :) [19:27:10] 03(mod) Support MySQL >= 4.x boolean logic in search - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=5711 (10brion) [19:27:43] !wg DisableHardRedirects [19:27:43] --mwbot-- http://www.mediawiki.org/wiki/Manual:%24wgDisableHardRedirects [19:27:46] jd_, ^^ [19:27:58] Hmm. [19:28:01] cool :) [19:28:05] The docs say hard redirects should work by default. [19:28:17] yep [19:28:25] 03(mod) importdump.php stops with no error - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13687 (10brion) [19:31:58] 03(mod) Add a new parameter to Article::purge() to prevent output - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13689 (10brion) [19:36:52] 14(INVALID) (top) in watchlists - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13698 +comment (10brion) [19:39:38] 03(mod) multiple CSS3 multi-column errors across multiple browsers - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13686 +comment (10Simetrical+wikibugs) [19:46:12] 14(WFM) Templates used inside tags aren' t listed at the bottom of the page on preview - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13672 +comment (10huji.huji) [19:55:48] 03(mod) multiple CSS3 multi-column errors across multiple browsers - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13686 (10superluser) [20:05:41] 03nikerabbit * r33146 10/trunk/phase3/languages/messages/MessagesVo.php: * Alternative time format for vo [20:07:05] 03(mod) multiple CSS3 multi-column errors across multiple browsers - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13686 (10danny_b) [20:07:09] 03(mod) CSS (tracking) - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=12788 (10danny_b) [20:23:10] Hi there. Does anyone know of an extension to force an automatic signature to talk pages? [20:24:36] Or, if there's any way to force it in a default installation, without the need for an extension. [20:25:30] Verilium: there is a message to customize it in latest version [20:26:35] 14(INVALID) multiple CSS3 multi-column errors across multiple browsers - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13686 +comment (10brion) [20:38:31] 03(NEW) Rollback link should not appear for earliest revisions in page history - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13705 15enhancement; normal; MediaWiki: History/Diffs; (meno25wiki) [20:40:41] ialex: Hmm... [20:41:38] ialex: mmmmm [20:42:10] MediaWiki:Signature and MediaWiki:Signature-anon [20:42:17] Nikerabbit: yes [20:42:27] ialex: was that a question? [20:43:13] Hi there. Does anyone know of an extension to force an automatic signature to talk pages? [20:44:01] 14(DUP) Diff error in it.wikisource - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13704 +comment (10brion) [20:44:06] 03(mod) Erratic corrupted diff (repeated lines, etc.) - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=9533 +comment (10brion) [20:44:15] 14(DUP) Diff in en.wikipedia displaying completely incorrectly - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13575 +comment (10brion) [20:44:23] 03(mod) Erratic corrupted diff (repeated lines, etc.) - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=9533 +comment (10brion) [20:45:39] bug 9533 has a lot of dupes... [20:51:10] ialex: Even if I customize it though, will this new method make it so anyone that leaves a comment, automatically signs it? [20:51:28] Verilium: no [20:51:42] Or maybe I just don't understand how to use this MediaWiki:Signature. :) [20:52:05] it will be used when a user use ~~~~ [20:53:00] Yeah. Darn, hmm.. [20:53:31] I've only managed to find 1 extension that can automatically add the signature when someone adds a comment. DiscussionThread. [20:53:37] Which doesn't seem that bad.. [20:57:48] <^demon> Verilium: The only way I could see automatically adding the ~~~~ after posting would be to take the old contents, take the new contents, compare them, and then put the ~~~~ after the addition. How else would you know where to put it otherwise? [21:03:00] <[exa]> hi...anyone could help me? where should I search for errors when any File:xxx page redirects to main mage? [21:03:18] <[exa]> it's pretty annoying [21:12:00] Adam____: Thanks for the help: When in doubt, back up! [21:12:18] Oops, wrong window, sorry [21:14:23] [exa]: That would be annoying. I thought any page that's attempted to be opened or edited will automatically try to let the user start that page [21:16:12] <[exa]> RickZilla: guh, got it. It's probably about bad redirect engine. gonna disable it. [21:16:37] <[exa]> index.php/File:... doesnt work, index.php?title=File:... works OK [21:17:01] <[exa]> I'm on strange hosting, doesn't seem like wiki error. [21:19:03] Hm. [21:19:29] Is there anyway to avoid a template being in the category like with , but when including itself in the template for documentation example? [21:19:41] 03siebrand * r33147 10/trunk/extensions/SocialProfile/UserBoard/UserBoard.i18n.php: Fix typo [22:06:51] I followed the instructions here: http://www.mediawiki.org/wiki/Extension:LDAP_Authentication/Non-AD_Configuration_Examples to get LDAP authentication working [22:07:11] But when I try and log in via LDAP my web browser just tries to download index.php [22:14:22] I'm trying to look for an extension that automatically redirects when a page contains a wikitext redirect tag; I thought I saw it somewhere before [22:14:48] anybody happen to know how to accomplish this off of the top of their head? [22:17:10] 03(NEW) Flagged bots have exeption from spamlist - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13706 15enhancement; normal; Wikimedia: General/Unknown; (wikipedia.kawaii.neko) [22:18:37] 03(mod) Flagged bots to have exeption from spamlist - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13706 summary (10wikipedia.kawaii.neko) [22:18:50] you lag as hell wikibugs [22:48:33] 03siebrand * r33148 10/trunk/phase3/languages/messages/MessagesEn.php: Fix typo [22:49:47] brion: hmm, rev_parent_id would be nice for bug 13705 ;) [22:57:13] 03aaron * r33149 10/trunk/phase3/maintenance/ (archives/patch-ar_parent_id.sql tables.sql updaters.inc): *Add ar_parent_id [22:58:20] 03(NEW) undocumented ? {blank} for first column in inline queries has some inconsistencies - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13707 minor; normal; MediaWiki extensions: Semantic MediaWiki; (info) [22:58:27] rc_timestamp is not a unix timestamp? [22:59:16] maybe, maybe not [22:59:22] don't assume, it depends on the DB [23:01:34] AaronSchulz: is it a unix timestamp on mysql? looking through the source it looks like it uses php time() [23:01:58] i think with mysql you get it stored as a string while with PG you get a proper time data type ;) [23:02:31] 03aaron * r33150 10/trunk/phase3/languages/messages/MessagesEn.php: Fix link [23:02:38] yeah, hehe, but I for sure know this particular instance of mw won't be using postgres [23:02:51] *blue_asterisk did consider postgres [23:04:14] AaronSchulz: what did r32684 change? re line ~449 changing $title to $target? [23:05:02] !r 32684 [23:05:02] --mwbot-- http://svn.wikimedia.org/viewvc/mediawiki?view=rev&revision=32684 [23:05:50] OverlordQ: RC entries for logs have the proper title in the title field [23:06:23] ah, was tracking down what was causing bug 13679 and think I narrowed it down to that. [23:08:51] is that PG? [23:09:08] yea [23:10:15] 03(mod) Flagged bots to have exeption from spamlist - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13706 +comment (10pathoschild) [23:11:56] the recentchanges has a foreign key on page_id from page, but when you're deleting, it gets removed from page breaking that [23:12:00] oh well, if anybody is interested, if you are using mysql, you can grab rev_timestamp from the revision table and just use substr to extract the year, month, date, etc [23:12:56] it's a hack I know [23:13:13] sometimes hacks are sufficient :) [23:14:17] before it was null since if you're deleting the page there isn't going to be a current rev id, but when you changed it from $title to $target, (going on a limb here) it's using the old revision id of the thing you just deleted [23:14:58] $title used to be log:special [23:15:15] the ID probably didn't matter there [23:16:09] did PG allow -1 ? [23:18:09] 03erik * r33151 10/trunk/tools/planet/en/config.ini: change name per request [23:18:34] did? doubt it, -1 would give you no such foreign key error also since no page -1 [23:19:16] how did it work before? [23:19:23] because that is what the title would be [23:19:48] let me revert and log the query it made [23:22:31] yeah, it would have been -1 [23:24:10] take a sec for a revert [23:28:09] mysql: rc_cur_id int unsigned NOT NULL default '0', [23:28:17] pg: rc_cur_id INTEGER NULL REFERENCES page(page_id) ON DELETE SET NULL, [23:28:19] *AaronSchulz sighs [23:28:32] OverlordQ: that should be ON DELETE SET 0 [23:30:39] *AaronSchulz thinks up some hack around [23:31:00] just a SET NULL and SET DEFAULT [23:32:02] actually [23:33:21] can you make a fix just to the PG table? [23:33:21] before r32684 rc_cur_id wasn't used on the INSERT into recentchanges [23:33:55] 'rc_cur_id' => $title->getArticleID(), [23:34:02] still was there [23:34:15] Yea but it wasn't included in what actually reached the DB [23:34:39] http://pastebin.ca/981684 [23:35:51] save() [23:35:53] if ( $dbw->cascadingDeletes() and $this->mAttribs['rc_cur_id']==0 ) { [23:35:54] unset ( $this->mAttribs['rc_cur_id'] ); [23:35:56] } [23:36:03] ok, that makes a fix easier [23:36:25] that's why it wasn't getting to the DB [23:38:47] still, not sure why it wasn't -1 [23:39:23] ok, I see, linkcache [23:39:24] made it 0 [23:39:31] ok, time to make fix [23:42:21] 03aaron * r33152 10/trunk/phase3/includes/Article.php: Reset ID before adding to log. That way it goes in as 0 for deletions. RC save() stuff then makes values of zero become NULL. This should fix PG error. [23:44:07] works good now :) only problem left with deletes is CheckUser :D [23:44:42] O.O [23:44:46] *Mike_lifeguard was pinged :\ [23:45:08] *Mike_lifeguard removes that from his stalklist [23:45:20] which is probably related to the exact same thing: Undefined variable: rc_cur_id in /home/wiki/extensions/CheckUser/CheckUser.php on line 76 [23:50:42] 03brion * r33153 10/trunk/extensions/StaticWiki/StaticWiki.php: [23:50:42] * (bug 13688) Use Http::get() in place of file_get_contents() [23:50:42] Patch by ^demon - https://bugzilla.wikimedia.org/attachment.cgi?id=4801 [23:50:47] 03(FIXED) Remove file_get_contents() from StaticWiki - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13688 +comment (10brion) [23:53:38] 03aaron * r33154 10/trunk/extensions/CheckUser/CheckUser.php: On PG, this may not be defined [23:53:42] OverlordQ: try now [23:54:44] 03(FIXED) Deleting images and RecentChanges broken - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13679 +comment (10JSchulz_4587) [23:55:11] https://bugzilla.wikimedia.org/show_bug.cgi?id=13345 [23:55:18] Yaaay :) [23:55:19] is that the only known PG breakage left? [23:55:46] replacing images wfm [23:56:21] 1.11.2 ? [23:56:34] maybe it was fixed since then [23:56:35] oh 1.11.2? no, svn [23:59:27] 14(WFM) Linking to anchor reloads page if given custom title - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=13685 +comment (10brion)

xxx

HTML headings

heading

and