Saturday, 2009-06-06

../irclogs/#mantishelp.2009-06-06.log
--- scribe started ---00:00
paul_hmm00:02
paul_apc?00:02
nuclear_eclipsego back to sleep00:02
nuclear_eclipsejreese@mach[~/workspace/mantisbt] git bisect good                                                                           < 20:0200:03
nuclear_eclipse1096c049774b5f6e85813c8f7d7bb37eea33dd1f is first bad commit00:03
nuclear_eclipsecommit 1096c049774b5f6e85813c8f7d7bb37eea33dd1f00:03
nuclear_eclipseAuthor: Siebrand Mazeland <s.mazeland@xs4all.nl>00:03
nuclear_eclipseDate:   Sun May 3 17:57:23 2009 +020000:03
nuclear_eclipse    * remove some unneeded strtolower()00:03
nuclear_eclipse:100644 100644 823f8f347e0ec7fae84038cffda5eda6b69fb88d 636c439d9e79b35c68cfa9872389963f5077fb91 Mchangelog_page.php00:03
nuclear_eclipse:100644 100644 21b405f1b3fc11d20203076bf3bac3ba6f9874f8 e599aedf198bfc8ed28c7e806aad9adc0abe123e Mroadmap_page.php00:03
nuclear_eclipseit's quite obvious now :P00:04
nuclear_eclipseor rather, it's obvious that you broke something00:04
nuclear_eclipseI still haven't figured out what the actual bug is....00:04
siebrandnow I'm more confused.00:05
nuclear_eclipselol00:05
siebrandWe have established that the 'live' version is broken, but master is not.00:05
siebrandIs the right?00:05
nuclear_eclipseright, that's the commit that "fixed" it :)00:05
siebrandOK. I do not see what's wrong with that commit. Please teach me, oh master :)00:06
nuclear_eclipsein this context, `bad` means it exhibits the current good behavior (ie, I'm working in the context of git-bisect)00:06
nuclear_eclipsegot it now00:06
siebrand(except that I forgot to note the adding of the space in the commit msg)00:06
nuclear_eclipsethere *was* a bug, and you adding the word_separator actually fixed it ;)00:06
siebrandah, less confused.00:06
siebrandConclusion: much ado 'bout nuttin.00:06
nuclear_eclipseor rather, you apparently fixed it by accident :P00:07
siebrandnope, I didn't. That was done on purpose, because the lack of space annoyed me.00:07
siebrandI just forgot to note it on the commit.00:07
nuclear_eclipsethe confusion is that the live install is officially stale :D00:07
siebrand*past* bed time for me.00:07
siebrandCheers!00:07
nuclear_eclipsecheers siebrand00:07
CIA-61Mantisbt: s.mazeland live * rc53b7f80f958 / (changelog_page.php roadmap_page.php): * remove some unneeded strtolower()00:09
nuclear_eclipsequit00:11
paul_hmm00:14
paul_why are we patching live?00:14
paul_in fact, why does live exist? ;/00:15
nuclear_eclipsebecause you broke dates and still havent fixed them...00:18
paul_hmm00:26
paul_I made a change a while back00:26
paul_to let you change globalconfig settings within pages00:26
paul_that's not a great idea with plugins now available or is it?00:26
nuclear_eclipsewhat would be the point of it?00:27
paul_http://git.mantisforge.org/w/mantisbt.git?a=commitdiff;h=d26d7cf6e8d7050cf7eba4793fcc0296a93a8e9500:27
nuclear_eclipsecant check that atm... =/00:28
* paul_ tries to recall00:33
paul_where/when do we auth a user00:33
paul_public function __get( $t_string ) {00:37
paul_if( is_null( $this->{$t_string} ) ) {00:37
paul_$this->{$t_string} = config_get( $g_default_mapping[$t_string], null, $this->pref_user_id, $this->pref_project_id );00:37
paul_should that also check for empty? ;/00:37
paul_i.e. ""00:37
paul_ahh right00:50
paul_we authenticate people for the first time by calling collapse_cache_token!00:51
paul_nn01:00
* paul_ sighs loudly11:12
paul_.17:35
nuclear_eclipsehi paul_18:11
setuidQuestion: I'm trying to figure out where in the Mantis config, I'd supply our proxy configuration. We run Squid behind everything here to accelerate Apache and other apps. Where would I plug that into Mantis, so it can cache its pages in Squid as well?18:50
nuclear_eclipsesetuid: you generally wouldn't want your mantis pages to be cached by squid18:51
setuidIn other words, a request for a page from Mantis, goes to Squid before MySQL18:51
setuidnuclear_eclipse, We most-definitely do18:51
setuidnuclear_eclipse, Squid is configured appropriately, and there's no reason why they can't be cached. Our benchmarks show a 700% performance improvement doing so with other php-base apps.18:51
nuclear_eclipsethat would be serving potentially stale informaton then, esp after updating an issue18:52
setuidServer disconnected me...18:53
setuidI missed anything you might have said after I said:18:53
setuidnuclear_eclipse, Squid is configured appropriately, and there's no reason why they can't be cached. Our benchmarks show a 700% performance improvement doing so with other php-base apps.18:53
nuclear_eclipsethat would be serving potentially stale informaton then, esp after updating an issue18:54
setuidNope, never serves stale information.18:54
setuidThat's precisely the point18:54
setuidSquid knows when a page has changed, and serves up the appropriate page as needed.. either from cache, or from MySQL (and caches that)18:55
nuclear_eclipsethen let's just say Mantis doesn't yet play nice with caching headers18:55
nuclear_eclipseif you really are after performance, you could try setting up an opcode cache for PHP, which will give you comparable performance gains18:56
setuidWow, Freenode is flaky this morning18:56
kenguestmorning? where?18:56
setuidWe don't have a large system... maybe 3,000 issues filed, 1,000 or so active users.18:57
nuclear_eclipse14:55 < nuclear_eclipse> then let's just say Mantis doesn't yet play nice with caching headers18:58
nuclear_eclipse14:56 < nuclear_eclipse> if you really are after performance, you could try setting up an opcode cache for PHP, which18:58
nuclear_eclipse      will give you comparable performance gains18:59
paul_setuid: you run apc?19:00
setuidpaul_, yes19:00
paul_nuclear_eclipse: I hate mantis19:00
nuclear_eclipsegood for you paul_19:00
paul_remind me where you run mantis again19:00
setuidmoi?19:00
paul_ya19:00
* paul_ assumes it's something public19:00
setuidbugs.plkr.org, bugs.pilot-link.org, bugs.jpilot.org19:00
paul_nuclear_eclipse: so, question for you19:00
paul_in the lifetime of a request19:01
paul_within what function do you think authentication first happens?19:01
paul_or gets 'checked'19:01
nuclear_eclipse20:50 < paul_> ahh right19:01
nuclear_eclipse20:51 < paul_> we authenticate people for the first time by calling collapse_cache_token!19:01
nuclear_eclipseit's only incidental though19:02
nuclear_eclipseie, it just happens to be the first portion of the code that calls an API that through some chain looks for authentication info19:02
Heady|is there already a page which lists all available 1.2 plugins?19:04
nuclear_eclipsegit.mantisforge.org is the closest to that19:05
paul_anyway i've fixed versions19:09
nuclear_eclipsetook you long enough :P19:10
paul_but we've got some things to fix19:11
paul_nusoap needs fixing19:11
* paul_ can't test soap stuff as nusoap fails19:12
setuidpaul_, Any comments on performance on our side?19:12
paul_setuid: i'm surprised you'd need to use squid to cache the mantis pages19:13
paul_if your running apc19:13
paul_although, might be worth while seeing what 1.2 does in a few weeks19:13
paul_or months19:13
setuidWe don't "need" it, but it gives us a significant performance boost19:13
paul_brbfood19:14
setuidIs there any way to allow summary_graph_byseverity_mix.php and friends to be accessible by Guest?19:15
setuidi.e. not require a login?19:15
nuclear_eclipsesetuid search config_defaults for info about designating an account for anonymous access19:15
setuidWe have one, for browsing around the tracker19:15
setuidBut apparently the graphs require login19:16
nuclear_eclipse? that's odd19:16
nuclear_eclipseperhaps you could try looking in configs for thresholds to lower to VIEWER19:16
setuidYep, summary isn't available to anonymous users19:16
setuidah! g_view_summary_threshold19:17
setuidperfect...19:17
giallupaul_, what kind of nusoap fix? please DON'T touch upstrema code...19:54
nuclear_eclipse+119:55
* giallu has just been asked to move nusoap out of mantis package in fedora...19:56
paul_why?19:59
nuclear_eclipsebecause NuSOAP != MantisBT20:01
giallubecasue if nusoap has bug, they are fixing it. just report it, better with a patch20:01
paul_i meant recording moving it out of package20:02
paul_regarding even20:02
giallupaul_, I tried to explain it to you sever times. I failed20:02
gialluseveral20:02
giallusuffice to say, we have another package with the same embedded lib, we want to avoid duplication20:04
paul_anyway, we need to get a fixed version of nusoap20:12
giallupaul_, ok. I didn't catch what's the problem with nusoap20:19
paul_generates php strict warnings20:19
gialluok20:20
paul_although i'm not quite sure where nusoap is being developed now20:21
gialluhttp://code.google.com/p/nusoap-for-php5/ here maybe?20:21
paul_nope20:21
paul_that's a fork20:21
paul_http://sourceforge.net/projects/nusoap/20:21
paul_but also20:21
paul_there's a 1.0rc version of nusoap at http://projects.ez.no/nusoap20:22
gialluhow come we use the crappiest libraries around....20:22
giallu:P20:22
gialluyou first link doesn't work here20:23
giallu:(20:23
paul_sf.net is down i think20:23
gialluall of it?20:23
giallucool20:23
* paul_ shrugs20:23
nuclear_eclipsetell me again why we aren't using PHP's built-in SOAP libraries?20:30
giallunuclear_eclipse, good question20:33
gialluI think it has something to do with the fact it's not a default component of php20:34
nuclear_eclipseso? :P20:34
giallufor me it's irrelevant20:34
nuclear_eclipseif someone really wants to use SOAP with Mantis, why don't we just require them to have the extension enabled?20:34
gialluI have it with a "yum install php-soap"20:35
nuclear_eclipseexactly20:35
giallunow20:35
gialluI wonder why for instance even ZF is providinig a Zend_Soap class...20:36
nuclear_eclipsebecause Zend decided to provide everything20:36
nuclear_eclipseand to make it all as verbose as possible while they were at it20:37
gialluwell, ok, but if they felt compelled to wrap php-soap I bet there is a better reason20:37
giallunuclear_eclipse, no, verbosity champion is Microsoft20:37
nuclear_eclipseperhaps, C# + .Net is les verbose than Java + Java Lib20:38
nuclear_eclipseC#/.Net is probably my least hated thing to ever come out of Microsoft, with a possible exception for the gaming studios they published for...20:39
nuclear_eclipseactually, C# combined with Xna actually makes for a really nice framework for making video games; it's just a total shame that it only ever works on Windows and Xbox...20:40
paul_nuclear_eclipse: apparently the php stuff doesn't do some thing21:06
CIA-61Mantisbt: paul * re17bd57d7534 /graphs/ (8 files): Remove unused graph files (half of these don't work)22:04
CIA-61Mantisbt: paul * r5343a7f722a5 / (4 files in 3 dirs): Store User id of users adding files (work towards #0003437)22:04
nuclear_eclipsepaul_: just fyi, you can't just put the bug # in, you need to do "fix|issue|bug #" type of thing22:06
nuclear_eclipseor at least, if you want the source integration to pick it up as a bug reference :P22:07
paul_did was use \ to escape % in config api?22:09
nuclear_eclipseye22:10
paul_oops22:10
paul_grr22:10
paul_that was wrong ;/22:11
paul_I can't amend commit i've pushed can I? ;/22:11
giallubug 343722:11
mantisbotBug 3437 - mjacek - open - acknowledged22:11
mantisbotUser should be able to edit his own documentation - http://www.mantisbt.org/bugs/view.php?id=343722:11
nuclear_eclipsenot without cauing headaches to anyone who's pulled from it, including mantsiforge22:11
nuclear_eclipsedon't worry bout it22:12
giallupaul_, IMHO commit 648f388 is wrong22:17
paul_oh?22:18
nuclear_eclipsewhat's commit 648f388?22:19
gialluOne goal in the newlayout world is to get rid of g_core_path22:19
gialluhttp://git.mantisbt.org/?p=mantisbt.git;a=commit;h=648f388e1c076c431495aa2b18f77d9d5adb6ba622:19
nuclear_eclipseoh, an old news commit :P22:19
gialluyeah, I just noticed it22:19
giallubecause it conflicts with my current stuff :(22:20
paul_you need to allow users to move core path outside of webroot22:20
paul_iirc, that's what that was fixing22:21
gialluwhen I'm done, core will be outside webroot...22:21
giallubut it's ok... I'll deal with that later22:22
paul_well, it should be optional22:22
paul_personally, I use mantis with core in webroot22:22
paul_as it's easier to set up22:22
paul_or more, I dont mind as long as one can just extract mantis and run it22:24
paul_without requiring server config access22:24
CIA-61Mantisbt: paul * r544e7d9a8cd1 /core.php: add autoload function to spl register - this allows plugins to define their own (additional) autoload functionality22:39
CIA-61Mantisbt: paul * r8663867b27e3 / (9 files in 3 dirs): Changes for bug 1056522:39
mantisbotBug 10565 - jreese - open - assigned22:39
mantisbotUser input date strings are munged with new schema - http://www.mantisbt.org/bugs/view.php?id=1056522:40
paul_nuclear_eclipse: can we fix tests?22:41
siebrandpaul_: inconsistent message formatting in your latest commit:23:01
siebrand $MANTIS_ERROR[ERROR_INVALID_SORT_FIELD] = 'Invalid sort field.';23:01
siebrand+$MANTIS_ERROR[ERROR_INVALID_DATE_FORMAT] = 'Invalid Date Format.';23:01
paul_ohhi23:03

Generated by irclog2html.py