{5} Assigned, Active Tickets by Owner (Full Description) (36 matches)

While debugging a different problem, I stumbled across this PHP error:

array(4) {
  ["type"]=>
  int(2048)
  ["message"]=>
  string(141) "Non-static method midcom_baseclasses_core_dbobject::get_parameter() should not be called statically, assuming $this from incompatible context"
  ["file"]=>
  string(79) "midcom/midcom.core/midcom/core/dbaobject.php"
  ["line"]=>
  int(156)
}

I don't know if this is a problem or not, but I thought I'd post it anyways.

The big problem is that splitting an mbox to individual messages is based on just normalizing newlines to NL and then splitting on a regex that causes more than one message to be treated as a single message.

If xpath rules are used, fi.hut.htmlimport passes HTMLPurifier's output to simplexml, which fails on input that is broken enough.

see

http://trac.midgard-project.org/browser/branches/ragnaroek/midcom/fi.hut.htmlimport/importer.php

a partial fix would be to not pass htmlpurifier's output to simplexml if it is not needed, i.e. when no xpath rules are used

When emails are imported from a mailing list to n.n.discussion, they will be formatted via nl2br for display.

It would be great to blockquotize replies (lines starting with >).

MidCOM has phpdoc API documentation in:

http://www.midgard-project.org/api-docs/midcom/ragnaroek/

At the moment this is manually generated and uploaded. We should start generating it automatically with some given time period.

This would make the TODO list for instance a lot more useful: http://www.midgard-project.org/api-docs/midcom/ragnaroek/todolist.html

It appears the MidCOM caching service also writes things to cache that were already coming from there, possibly also on requests that were cache hits in content cache.

This should be changed so that cache is written only if there are cache puts or invalidates during the request. We should also safeguard no unnecessary puts are made.

cron runs in current ragnaroek svn report lots of warning like this one:

 ! ) Warning: Invalid ar gument supplied for foreach() in ragnaroek/midcom/midcom.core/midcom/services/cache/module/content.php on line 845
Call Stack
#
Time
Memory
Function
Location
1  0,0011  153472  {main}( )  ../61-305-68-0.php:0
  2  0,0464  1535564  midcom_application->codeinit( )  ../61-305-68-0.php:80
  3  0,0492  1565128  midcom_application->_process( )  ../application.php :475
  4  0,0493  1566108  midcom_application->_exec_file( )  ../application.php< b>:914
  5  0,0839  1539304  require( 'ragnaroek/midcom/midcom.core/midcom/exec/cron.php' )  ../application.php:2 516
  6  0,1297  1607368  midcom_services_cron->execute( )  ../cron.php:48< /td>
  7  1,8224  6226296  midcom_services_cron->_execute_job( )  ../cron.php >:329
  8  1,8225  6226296  midcom_baseclasses_components_cron_handler->execute( )  ../cron.php:364
  9  1,8225  6226296  midcom_services_at_cron_check->_on_execute( )  ../handler.php:126
  10  4,8884  7299556  midcom_core_dbaobject->delete( )  ../check.php: 80
  11  4,8884  7299720  midcom_baseclasses_core_dbobject->delete( )  ../db aobject.php:313
  12  4,9269  7466636  midcom_baseclasses_core_dbobject->delete_post_ops( )  ../dbobject.php:613
  13  4,9429  7466968  midcom_services_cache->invalidate( )  . ./dbobject.php:701
  14  4,9435  7467096  midcom_services_cache_module_content->invalidate( )  ../cache.php:162
   

Would be nice to make midgard dtrace-enabled. This would allow much easier debugging on Operating Systems, which have support for dtrace'ing. (Solaris, Mac OS X, FreeBSD)

Those routines were designed for proxies like midgard-apache module being responsible to set data for php-midgard extension. That's why both are not longer needed.

Old midgard used connection per process pattern, in newest Ragnaroek we copy connection structures and use connection per request one. However in some cases, MidgardConnection? structure might be copied twice by apache module. This makes MidgardConnection? unaware of current person/user being logged in, or sitegroup id for example.

I have the following schema type:

<type name="net_nemein_avaudu_contact" extends="midgard_person">

If I look at it using midgard_reflection_class, it seems the class extends midgard_object on PHP level instead of midgard_person

More decent and being under development python's Distribute should be used instead of not maintained setuptool. http://packages.python.org/distribute/

Vapi file for midgard-core should be provided in bindings.

Midgard has a feature where the "name" field is checked for uniqueness, so for example there can't be two articles named "news_item" under same topic.

However, this check does not check for multiple possible child types, which can lead into problems with get_by_path and MidCOM components.

For example, under same topic can now be article named "news_item" and topic named "news_item", which is obviously wrong.

non-sitegroup admin users (trusted site editors for example) should be able to create persons (but not user accounts?) if they have sufficient (i.e. midgard:create) privileges.

It caches the results and if you do get_parent(), move object and then get_parent() again you will get old parent.

This might affect a bunch of places so I would put this as blocker for .7 but not .6 since the incorrect behaviour has not affected anyone much yet.

when you use both chooser and composite widgets in a form (like f.x. in o.o.projects task), the chooser search fields are not displayed, thus making it impossible to enter data here. Firebug shows the following error:

id is undefined
midcom-static/midcom.helper.datamanager2/jquery.dm2_ajax_editor.js
Line 231

MidCOM should provide a CLI shell that would enable running operations with routes and services. For instance more granular cache invalidation or testing of route output.