| ../irclogs/#mantishelp.2010-01-21.log | ||
| --- scribe started --- | 00:00 | |
| CIA-22 | Mantisbt: vboctor master-1.2.x * rdfbe8b563b8b /core/ (cfdefs/cfdef_standard.php gpc_api.php): Fixes #10670: Custom fields required do not work. | 04:06 |
|---|---|---|
| CIA-22 | Mantisbt: vboctor * re45c8298deed /core/ (cfdefs/cfdef_standard.php gpc_api.php): Fixes #10670: Custom fields required do not work. | 04:07 |
| rombert | Hey guys | 09:51 |
| rombert | Trying to debug an upgrade from 1.1.0 to 1.2.0rc | 09:51 |
| rombert | schema update fails at Schema UpdateFunction ( date_migrate ) | 09:52 |
| rombert | fails as in " The (GOOD) box is not printed, but no errors are shown" | 09:52 |
| rombert | nothing in the apache error log | 09:52 |
| rombert | I've set max_execution_time to 300, just in case | 09:52 |
| dhx_m | hmm is date_migrate getting called ok? | 10:07 |
| dhx_m | I moved it to install_helper_functions_api.php a while ago | 10:08 |
| rombert | just giving it 128M of memory worked | 10:12 |
| rombert | 150k issues ... | 10:12 |
| rombert | thanks anyway | 10:12 |
| dhx_m | it seems we need to implement date_migrate as a function that loads up pages in-between | 10:13 |
| dhx_m | ie. do 10k issues per page load | 10:13 |
| rombert | that would certainly help | 10:13 |
| rombert | Right now it loads all of them, right? | 10:13 |
| rombert | I think better error reporting would help as well | 10:14 |
| rombert | I was shooting in the dark, trying to guess why it failed | 10:14 |
| dhx_m | yeah I imagine the problem is every time the function is called (once per row) we're creating local variables/arrays/whatever | 10:14 |
| rombert | I see | 10:14 |
| dhx_m | and the garbage collector isn't executing in between | 10:14 |
| dhx_m | so those 150k+ variables are lingering around inefficiently (it's PHP) | 10:14 |
| rombert | right | 10:15 |
| dhx_m | are you able to modify the code to call gc_collect_cycles() every 10k cycles? | 10:15 |
| dhx_m | and see if that solves memory problems? | 10:16 |
| dhx_m | oh | 10:16 |
| dhx_m | it's PHP 5.3.0+ only | 10:16 |
| rombert | I'm running 5.1.6 here | 10:16 |
| rombert | so I'm afraid not | 10:16 |
| dhx_m | yep don't worry about it, because most people aren't using 5.3 (it was very buggy on release) | 10:16 |
| rombert | ok | 10:17 |
| dhx_m | hmm what I said doesn't make sense though | 10:18 |
| dhx_m | I'd be VERY surprised if variables aren't freed after return from a function | 10:18 |
| dhx_m | as their scope has ended | 10:18 |
| dhx_m | some info here: http://www.php.net/manual/en/features.gc.refcounting-basics.php | 10:19 |
| rombert | Is there any memory profiler available for PHP ? Or the possibility of triggering a memory dump / histogram? | 10:23 |
| dhx_m | maybe you can do a similar method as used here: | 10:24 |
| dhx_m | http://www.php.net/manual/en/features.gc.performance-considerations.php | 10:24 |
| dhx_m | memory_get_usage(); | 10:24 |
| rombert | OK, I skimmed that. Perhaps I'll get around to it when I do the real migration. This was just a test. | 10:25 |
| dhx_m | 150k is a large migration :) | 10:26 |
| dhx_m | I've always wondered what the largest/most active Mantis installation is | 10:27 |
| dhx_m | as I can't see Mantis coping well with more than about 20 concurrent users, at most | 10:27 |
| rombert | Well, this is one installation | 10:27 |
| rombert | I don't know about the concurrent users | 10:27 |
| dhx_m | (for an old dual core system that is) | 10:27 |
| rombert | Never heard anyone complain about responsiveness | 10:27 |
| giallu | dhx_m, are you sure it's PHP's fault? | 10:33 |
| giallu | that is, are we sure we're not caching stuff on purpose? | 10:34 |
| dhx_m | giallu: it's quite probably that our code is bad too :) | 10:34 |
| dhx_m | *probably | 10:34 |
| * giallu has been hit by cache issues quite often | 10:34 | |
| dhx_m | grr, *probable | 10:34 |
| dhx_m | 3 tries | 10:34 |
| dhx_m | project hierarchy caching is probably a big one | 10:35 |
| giallu | for most things, I don't really see why we are caching stuff. For instance, 32Mb is not enough to load our user list in adm_config_report.php | 10:37 |
| giallu | user list is used exactly _once_ that is, for building the list... | 10:38 |
| giallu | another example, my export plugin easily hits memory limit because bug data is cached when read, again I need to read it, dump it to XML and never use it again | 10:39 |
| * giallu thinks he will start a "killcache" git branch | 10:40 | |
| rombert | I'm not sure about the performance gain of these caches | 10:41 |
| rombert | perhaps we should have a hit count? | 10:41 |
| rombert | Then we can decide if a certain cache is worth it or not | 10:41 |
| rombert | IMO we should cache 'reference data', like category names | 10:42 |
| rombert | If we cache bugs it sounds a little bit like we're not properly remembering what we've retrieved from the database | 10:43 |
| rombert | but then again, I don't know much about the internals | 10:43 |
| dhx_m | back | 11:03 |
| dhx_m | the purpose of cache is to prevent continuous queries to the SQL server | 11:03 |
| dhx_m | otherwise every time you need to know about a user account, you have to hit the database | 11:03 |
| dhx_m | at least that is the reason I *suspect* we use cache ;) | 11:03 |
| giallu | dhx_m, I think I understand the "cache" concept. however, we should use cache where needed, i.e. only in pages where we measure a negative impact from not doing it | 11:15 |
| giallu | the current approach is more like: let's just cache everything I read from the DB, just in case | 11:16 |
| dhx_m | yep I agree, the caching needs to be much smarter | 11:16 |
| dhx_m | and only done where absolutely needed (ie. not for single use transactions) | 11:16 |
| dhx_m | for instance, you may want to print a list of project hierarchies, but that doesn't indicate a need to cache that information | 11:17 |
| dhx_m | you may only print it once and be done with it | 11:17 |
| TrahDivad | Hi people. I'm trying to find a way for my application to programmatically post an issue on my Mantis server when an unhandled exception occurs. | 11:23 |
| TrahDivad | I seems like I need to use Mantis Connect, but it seems a bit complicated to use in Cocoa (didnt find any example). I was wondering if there was a simplier way, similar to BugzScout http://www.fogcreek.com/FogBugz/docs/60/topics/customers/BugzScout.html | 11:24 |
| TrahDivad | I'm trying to find a solution that is as lightweight as possible | 11:28 |
| rombert | TrahDivad: have you tried just using the SOAP API to create a new issue? | 11:30 |
| rombert | the WSDL is at server/mantis/api/soap/mantisconnect.php?WSDL | 11:31 |
| TrahDivad | I don't know anything about SOAP, but it seems that it's a bit heavy. And Cocoa dosn't have good support for SOAP | 11:37 |
| TrahDivad | As it's happening in the Exception Handler of the application, it needs to be fast and simple. | 11:38 |
| TrahDivad | I'll try looking into SOAP if ther's no other possibility, but I xanted to make sure beforehand. | 11:39 |
| rombert | There's no alternative that I know of | 11:48 |
| alemayo | Hi! | 12:44 |
| alemayo | we are just about to create our third plugin :-) | 12:45 |
| TrahDivad | I'm trying to use the SOAP web service and I keep getting Access Denied. Any idea if the user I'm trying to post with must have special priviliges? | 13:26 |
| alemayo | http://git.mantisforge.org/w/Terminology.git | 13:35 |
| alemayo | there is something wrong with the update-columes function in 1.2.x | 13:58 |
| alemayo | we want to set columes for a project. we selected that projects, but the button "update columes for project XXX" is hidden | 13:58 |
| alemayo | I think the if block around line 144 in mange_columns_inc (if ( $t_account_page ) { ) must just be removed -- does not seem to make sense in there ?! | 13:59 |
| bolshi | I've got some problems configuring EmailReporting via IMAP | 15:04 |
| bolshi | Does someone know an configuration sheet? | 15:06 |
| bolshi | crontab delivered the following: <p style="color:red">SYSTEM WARNING: require_once(/Applications/MAMP/htdocs/mantisbtrczwo/core/mail_api.php): failed to open stream: No such file or directory</p> | 15:54 |
| bolshi | we added imap support | 15:55 |
| bolshi | instead of mail_api.php the file is named mail_api_imap.php | 15:56 |
| nuclear_eclipse | bolshi: we can't support 3rd party modifications to mantis here, sorry | 15:56 |
| bolshi | sorry | 15:56 |
| bolshi | i didnt know that | 15:56 |
| nuclear_eclipse | no worries | 15:57 |
| aa3132 | hello. i'm trying to find the correspondence between the bug resolution field in the table, which is an id and the string for it. can anybody help? | 16:16 |
| paul | dhx_m: tbh, caching depends on what you deem important | 22:07 |
| paul | i.e. our AD at work is liek 1GB in size, yet you've gotta buy a server with 8gb of ram or something as that's the lowest amount available type of thing | 22:08 |
| paul | on that basis, ram = cheap, a remote call to a sql server over a network would therefore be slow+expensive | 22:08 |
| paul | so at the moment it serves two purposes | 22:08 |
| paul | a) the stuff we cache we tend to use >1 on a page anyway | 22:09 |
| CIA-22 | Mantisbt: s.mazeland * rb2c3d1db4ac4 / (46 files in 4 dirs): Localisation updates from http://translatewiki.net (2010-01-21) | 22:25 |
Generated by irclog2html.py