[Tickets #3304] Registry losing applications after api calls from configuration screens

[bugs@bugs.horde.org]

DO NOT REPLY TO THIS MESSAGE. THIS EMAIL ADDRESS IS NOT MONITORED.

Ticket URL: https://dev.horde.org/horde/whups/ticket/?id=3304
-----------------------------------------------------------------------
 Ticket             | 3304
 Updated By         | Michael Rubinsky <mrubinsk at horde.org>
 Summary            | Registry losing applications after api calls from configuration screens
 Queue              | Horde Framework Packages
 Version            | HEAD
 State              | Feedback
 Priority           | 1. Low
 Type               | Bug
 Owners             | Horde Developers, Michael Rubinsky, Michael Slusarz
-----------------------------------------------------------------------


Michael Rubinsky <mrubinsk at horde.org> (2006-01-26 11:52) wrote:

Ok. For those keeping score this is what I've been able to find out so far. 


* Problem occurs (for me - and on multiple installations / machines) with
php 4.4.x including recent development checkout from cvs.php.net.  Problem
does not occur with 5.1.2.  Again, problem only occurs when calling a horde
application's api method from a configuration screen.

* The order that the function calls occur is different under php 4 and php
5.

In php 4.3.x -> 4.4.x I get:
===============
  Registry::clearCache
  Registry::_shutdowncache  (at which point $this->_nocache is false for
some reason)
  Registry::Registry (constructor)


In php 5.1.2 I get:
===============

  Registry::clearCache
  Registry::Registry (constructor)
  Registry::_shutdowncache (at which point $this->_nocache is once again
false)


I think what's happening is that even though it appears that php4 is
actually giving the expected call order (at least what I would expect it to
be), Registy::_shutdowncache is being triggered after the registry object's
class variables go out of scope somehow, but before the session data is
saved. So, when _shutdowncache is called, $this->_nocache is always false
and we are not left with an unset $_SESSION['_registrycache'] by the time we
get the to Registry constructor.  At this point I always have
$_SESSION['_registrycache']['_confcache'] still set.

In php5, $this->_nocache is still false when checked in _shutdowncache, but
at this point, since the Registry constructor is called before
Registry::_shutdowncache, we are always left with an unset
$_SESSION[_registrycache'] and are forced to reload registry.php 

Quick hacks to fix this would be to either call session_write_close() after
we call Registry::clearCache - but I guessing this is a Bad Thing.  The
other option, which I'm currently using is to unset
$_SESSION['_registrycache']['_confCache'] if it is set in the registry
constructor.

I've got code to commit to provide a drop down menu with address book
sources for Turba's 'client address book'  setting (see Bug: 3271)  I'm
hesitant to commit this as I'm sure I'm not the only one who would have this
problem...