Sunday, 2009-07-05

../irclogs/#mantishelp.2009-07-05.log
--- scribe started ---00:00
* paul_ slaps vb123 00:48
vb123paul_:  why?00:49
paul_18:52 < CIA-17> Mantisbt: vboctor master-1.2.x * rbc945ed9eb4c /file_download.php: Fixes #10135: Downloading large attachments results in Internal Server Error - 500.00:52
paul_you miss the point00:52
paul_and potentially break stuff with that commit ;/00:52
paul_the problem is an out of memory error00:53
paul_so isn't affected by ob_end_flush00:53
vb123why would it break stuff?00:53
paul_calling ob_end_flush at that point potentially breaks stuff as we start >1 ob handler00:53
paul_one of which handles gzip00:53
paul_anyway, gonna try and  fix completely00:54
vb123thanks.01:04
CIA-17Mantisbt: vboctor * rf94f4df5f621 / (10 files in 3 dirs): Fixes #10681: Print issue page is not consistent with View Issue page relating to version handling.02:28
vb123dhx_m: I just checked in bug #10681 which has some cleanup for versions display across several page + refactoring on show dates logic.  Have a look at the changes when you get a chance.02:37
mantisbotBug 10681 - vboctor - fixed - resolved02:37
mantisbotPrint issue page is not consistent with View Issue page relating to version handling - http://www.mantisbt.org/bugs/view.php?id=1068102:38
* vb123 wondering what inherit_global flag on the projects is used for.04:03
dhx_mvb123: thanks, your version cleanup is good :)07:21
vb123dhx_m, I did some more changes to use ids instead.07:26
dhx_mwhat would be nice is converting the version rows array into an associative array07:33
dhx_mso it becomes O(n) lookup for a version07:33
dhx_malthough I imagine it'd have to be in the format:07:34
dhx_mversions[project_id][version_string]07:34
dhx_mbut I guess that could come later as part of a change to store versions as IDs in the database as opposed to strings07:35
dhx_mor is that what you also fixed?07:35
dhx_mnooo, bad time for mantisbt.org to die07:44
paul_moo10:21
paul_dhx_m: ?10:22
dhx_mpaul_: hey10:54
paul_lo10:54
paul_you know your previous commit10:55
paul_could you try just using rawurlencode insted10:55
dhx_mwell not entirely, no10:55
dhx_mI put a URL in the comments in the code that explains it better10:56
paul_well, other projects just use rawurlencode for IE10:56
dhx_mI gather it is incorrect for browsers to use urlencoded filename attributes?10:59
dhx_msee http://greenbytes.de/tech/tc2231/#attwithfnrawpctenclong11:01
dhx_mand the one above that, http://greenbytes.de/tech/tc2231/#attwithfnrawpctenca11:01
paul_hmmmm11:26
paul_dhx_m: I've found something to break :)11:28
dhx_mpaul_: woohoo! :)11:29
dhx_mwhat's breaking then :p11:29
paul_well11:30
paul_compressapi11:30
paul_:)11:30
paul_although it's confusing11:35
paul_in theory11:35
paul_we can replace compress_api with 1 line11:35
paul_aka11:35
paul_ini_set('zlib.output_compression', true);11:35
dhx_mlol11:35
dhx_mI haven't looked at compress_api yet but by the sounds of things it doesn't appear to be overly useful11:36
dhx_mIMO compression is the job of the web server or disk filesystem11:36
paul_now to read php source code ;(11:36
paul_MF53: - Fixed bug #45202 (zlib.output_compression can not be set with ini_set()11:38
paul_# This is fixed in HEAD already. :)11:38
paul_ahh11:38
paul_so from http://cvs.php.net/viewvc.cgi/php-src/ext/zlib/zlib.c?r1=1.183.2.6.2.11&r2=1.183.2.6.2.1211:38
paul_if php >= 5.2.10 || php >= 5.3.011:39
paul_if php >= 5.2.10 || php >= 5.3.0 || php >= 6.011:39
paul_we can call ini_set11:39
paul_else11:39
paul_we fallback to the old (slower) behaviour11:39
dhx_mah ha :)11:42
dhx_mdoes safe mode prevent people from using ini_set though?11:42
paul_safemode is removed in php611:43
dhx_mhooray! :)11:48
paul_but yea11:52
paul_by supporting old php versions11:52
paul_we end up with multiple code paths ;/11:52
dhx_myeah :(11:52
dhx_many thoughts on when PHP 6 may be expected>?11:52
paul_never? :)11:52
dhx_mare we talking this year... next year...?11:52
dhx_mlol ok my 3rd option was "when Perl 6 is released" :p11:53
paul_php6 was unicode11:53
paul_unicode status of php has been 70% for about 2 years11:53
paul_and stuck around there11:53
dhx_mhmmm11:53
dhx_mpeople still develop PHP6?11:53
paul_and new features have been put into 5.3 etc :)11:53
dhx_mor is it looking rather dead?11:53
paul_well, the zlib ini_Set thing was a backport from 6...11:54
dhx_mdo they use git?11:55
paul_no11:55
paul_cvs11:55
dhx_mAHHHHHHHHHHH CVS?!?!?11:55
paul_they are planning on moving from cvs to svn11:55
dhx_mPHP 6 won't be out until 2020 at that rate11:55
paul_however11:55
dhx_mlol11:55
dhx_m... to SVN?11:55
paul_the php5.3 release manager11:55
paul_sync's a git repo against cvs11:55
dhx_mstill... what sort of programmers would want to help a project that still uses CVS11:58
paul_any?11:58
dhx_mfairly frustrating to use if you ask me11:58
dhx_mIMO it is usually a sign of an unhealthy project11:58
paul_remember11:59
dhx_msame for projects that don't have an issue tracker, etc11:59
paul_they've got a fair chunk of stuff using it11:59
paul_i.e. converting from cvs->svn is apparently a 3day task for the conversion iirc11:59
dhx_myeah that's not too bad11:59
dhx_mfuture investment :)11:59
dhx_malthough I have no idea why they'd pick SVN12:00
dhx_munless their development community all has commit access12:00
paul_http://git.chinstrap.eu/12:00
dhx_mand they all commit in real time12:00
dhx_mI like one commit I see there with the description: "This is PHP 6 - if this ... ever gets out12:04
dhx_m" :p12:04
dhx_mI'm not seeing much activity in their repo?12:08
dhx_munless php-src is not where I should be looking12:08
paul_dhx_m: http://git.chinstrap.eu/php/php-tsrm-ze2.git/12:09
Chris1-I want mantis to send a email if there is a new entrey.12:09
paul_you want that one12:09
dhx_mhmm that's easier to read than viewcvs :)12:11
dhx_mnow if only comments weren't "Update" or "Fix"12:11
dhx_m(one thing that really gets to me is non-descriptive titles for emails, bugs or commits :p)12:12
paul_right12:14
paul_wanna review a patch? ;p12:14
dhx_mhell yeah! :p12:14
paul_http://mantisbt.pastebin.com/m5f15117012:16
paul_from core.php/compress_api.php12:17
dhx_mok12:19
dhx_mmalformed patch?12:22
dhx_mwhat command do I use to apply it?12:23
paul_I expected you to just read it12:23
dhx_moh ok12:23
dhx_mlooks good then :)12:23
* paul_ is making a mess though ;(12:23
dhx_mat least there isn't 400 expressions on the same line in your patch :p12:24
dhx_mmy only comment is that you've used // to comment on some lines and # on others12:24
dhx_mmy other question is more related to the entire Mantis codebase... and why we seem to invert expressions12:25
dhx_mso... if (OFF = ... ) instead of if ( ... = OFF )12:25
dhx_mof course, == :)12:25
dhx_malso core.php:145 calls ob_start( 'compress_handler' ) so that'd need to change too12:27
dhx_moops12:27
dhx_mI should read better, you already fixed that :p12:27
paul_seems i need to define a COMPRESSION_DSIABLED varaible :(12:31
dhx_mwhy?12:32
paul_or a global12:32
paul_so I can turn it off ;/12:32
dhx_mg_compress_html?12:33
paul_constant would be better then a global?12:33
dhx_mnot in this case, because you'd want the user to customize it?12:33
paul_i'm trying to fix my version of download12:34
dhx_mbut use config_get_global12:34
paul_which needs it disabled :)12:34
dhx_mnot config_get?12:34
paul_no12:34
dhx_mwell this doesn't make sense to be on a per-project basis?12:34
paul_different issue :)12:34
paul_i'm trying to do chunked downloads12:34
paul_to fix memory limit with large downloads12:35
paul_but you can't call ob_flush if zlib.output_compression is on12:35
paul_in addition, other projects turn that off for IE as a 'bugfix'12:35
paul_and it seems you can't do12:35
paul_ini_set(on);12:35
paul_ini_set(off);12:35
paul_header();12:35
paul_as when you turn it on it sets itself up permanently12:35
dhx_mah12:37
dhx_myou could ini_set a new script timeout?12:37
paul_spot something strange with http://mantisbt.pastebin.com/m119787cf12:38
dhx_mand instead of sending a new chunk... increase the timeout as required?12:38
dhx_mstrlen() on binary data... = bad?12:39
dhx_mthat's the first thing I notice12:39
dhx_mwho is to say that binary files don't include string termination chars (0x00) :p12:39
dhx_manother idea I had was to allow the user to specify a writeable public folder where Mantis can dump files for the user to read?12:40
dhx_mbut I guess if you wanted that, you'd just set that up in Mantis anyway (instead of storing attachments in the database)12:41
paul_cool12:52
paul_I just made a file that hangs my av12:52
dhx_mhaha12:53
paul_or not12:53
paul_hmm12:53
paul_do we want to support download accelerators?13:01
dhx_mhmmm13:34
dhx_mwe'd want to support resuming13:34
dhx_mmulti-threaded downloads less so13:34
dhx_malthough I guess it'd be nice to support that too :)13:34
paul_;P13:37
dhx_mI just think the ideas floating around about using the new zip extension to PHP to compress files as they're uploaded to the server is stupid13:45
dhx_m(no offence)13:45
dhx_mmakes no sense to me13:46
paul_hmm?13:53
dhx_mwell there was a ticket/mailing list post about the possibility of zipping all attachments as they arrive on the server14:01
paul_when?14:07
paul_a while ago?14:09
dhx_ma week or two ago14:09
dhx_mI'd ignore it :p14:09
dhx_myour idea is much better :p14:10
dhx_mthe limiting factor is always going to be CPU cycles or bandwidth rather than disk space14:10
[KK]Kirillhi all14:20
[KK]Kirillgit.mantisbt.org not work again14:21
dhx_myeah I wouldn't say mantisbt.org is the most stable server I'd seen :p14:22
[KK]Kirill^)14:23
[KK]Kirill:)14:23
[KK]KirillJohn not exists?14:23
paul_well14:25
paul_dhx_m: people keep rebooting the box when it dies14:25
paul_making it hard to diagnose14:25
dhx_mpaul_: is it some runaway processes using 100% CPU?14:26
paul_apache ;p14:26
dhx_mpaul_: it sounds to me like it could be a problem with memory usage?14:26
dhx_mmaybe some bot is coming along and hitting the server too hard at once?14:26
paul_yea14:27
paul_it's called msn14:27
dhx_mbing? :)14:27
paul_ping14:27
[KK]Kirill^)14:27
paul_pong14:27
[KK]Kirill:)14:27
dhx_mmsn => live search => bing => failure? :p14:27
paul_Content-Description: Download Data'14:42
paul_whats that?14:42
dhx_mthat doesn't look good14:44
dhx_mwhere did that come from?14:44
paul_i mean14:44
paul_whats that header14:44
dhx_mdoesn't look legit to me?14:44
dhx_moh14:44
dhx_mI'm thinking of Content-Disposition14:44
dhx_m   The ability to associate some descriptive information with a given14:46
dhx_m   body is often desirable.  For example, it may be useful to mark an14:46
dhx_m   "image" body as "a picture of the Space Shuttle Endeavor."  Such text14:46
dhx_m   may be placed in the Content-Description header field.  This header14:46
dhx_m   field is always optional.14:46
dhx_m     description := "Content-Description" ":" *text14:46
dhx_m   The description is presumed to be given in the US-ASCII character14:46
dhx_m   set, although the mechanism specified in RFC 2047 may be used for14:46
dhx_m   non-US-ASCII Content-Description values.14:46
dhx_mmost likely it only supports ASCII14:46
dhx_mgiven that most browsers don't support the RFC extensions for Unicode text properly14:47
[KK]KirillI talk John, that I want events for work with attachments14:48
[KK]Kirill(rename, revoke to story, etc)14:49
[KK]Kirillplugin's events14:49
paul_dhx_m: so drop that header?15:18
dhx_mpaul_: it's not bad15:20
dhx_mpaul_: but if all we're doing is saying it is "Download Data" we're not really making any use of it15:20
paul_testing time :)15:21
dhx_mactually I think it can be dropped15:21
dhx_mwe don't have attachment descriptions anyway15:21
paul_if ( in_array( utf8_strtolower( file_get_extension($t_filename) ), $t_inline_files ) ) {15:21
paul_$t_disposition = ''; //'inline;';15:21
paul_whats with that15:21
paul_} else {15:22
paul_inline commented out?15:22
paul_also15:22
dhx_mno need for inline, it's the default15:22
dhx_mused for display images?15:22
dhx_mie. inline display of attachments from the database15:22
paul_dhx_m: you have some download accelerators?17:43
* paul_ prods dhx_m 19:59
CIA-17Mantisbt: vboctor * re68de4159cae /core/bug_api.php: Fixes #10685: Undefined variable: t_bug_id on bug creation.21:00
CIA-17Mantisbt: vboctor master-1.2.x * rd95b85d139be /core/bug_api.php: Fixes #10685: Undefined variable: t_bug_id on bug creation.21:00
TuxmanHi can any provide help ?23:59
TuxmanSYSTEM WARNING: session_start() [function.session-start]: Cannot send session cache limiter - headers already sent (output started at /var/www/vhosts/.../config_inc.php:8)23:59

Generated by irclog2html.py