|
|
Up |
|
|
  |
Author: Chuck HagenbuchChuck Hagenbuch
Date: Jan 25, 2007 21:46
Quoting Karsten Fourmont :
> I just realized that calling DataTree:getObject is extremly costly
> for nonexistent objects (with sql backend). If you do this:
Thanks for this analysis!
I was uncomfortable with the change at first, but I did some poking
around and committed this set, which includes your change but also
goes a bit beyond it:
http://lists.horde.org/archives/cvs/Week-of-Mon-20070122/064958.html
Now, when calling any of exists(), getId(), or getName(), if an object
isn't already loaded, we simply check the backend for it instead of
trying to load it and then seeing if it's there. This removes some of
the "on-demand" aspect of the datatree, but that's in all likelihood a
good thing. May result in a few bugs to iron out but I've already
tested it with 6-7 datatree using apps plus the admin screens.
Let me know how that looks/performs,
-chuck
--
"we are plastered to the windshield of the bus that is time." - Chris
|
| Show full article (1.16Kb) |
|
| |
no comments
|
|
  |
Author: Chuck HagenbuchChuck Hagenbuch
Date: Jan 25, 2007 21:04
Quoting Volker Then volkerthen.com>:
> here is an update on my ldap prefs "debugging":
>
> It seems that only the latest edited prefs got saved in the ldap
> backend and every other hordePref is getting deleted.
>
> For example, when I change the portal layout, only "portal_layout"
> is left in the hordePrefs, any other hordePrefs are gone.
|
| |
|
| |
no comments
|
|
  |
Author: Volker ThenVolker Then
Date: Jan 24, 2007 02:04
Quoting Volker Then volkerthen.com>:
> Quoting Chuck Hagenbuch horde.org>:
>
>> Quoting Volker Then volkerthen.com>:
>>
>>> seems that the LDAP prefs are still not working. All prefs are
>>> gone, no new prefs are getting saved.
>>>
>>> Let me know how I can help!
>>
>> I need debugging help with this then - I don't have an LDAP prefs setup.
>>
>> -chuck
>
Hi Chuck,
here is an update on my ldap prefs "debugging":
It seems that only the latest edited prefs got saved in the ldap
backend and every other hordePref is getting deleted.
|
| Show full article (0.91Kb) |
|
no comments
|
|
  |
Author: Chuck HagenbuchChuck Hagenbuch
Date: Jan 21, 2007 08:35
Quoting Jan Schneider horde.org>:
>> Log:
>> Fallback for line folding when no mbstring is available.
>
> That class' name is "Util".
We do have extensionExists() inside String though, I guess for
dependency reasons. Right now iCalendar.php loads String.php but not
Util.php, so not sure what the right call here is.
-chuck
--
"we are plastered to the windshield of the bus that is time." - Chris
|
| |
|
no comments
|
|
  |
|
|
  |
Author: Karsten FourmontKarsten Fourmont
Date: Jan 21, 2007 03:38
Hi,
I just realized that calling DataTree:getObject is extremly costly for
nonexistent objects (with sql backend). If you do this:
$o = $_datatree->getObject("unknown_fake_object");
you'll end up with this sql query being executed:
SELECT [*] FROM horde_datatree c WHERE c.group_uid = 'syncml'
which means that the complete datatree is loaded! Futile as the object
is not there no matter how much you load. (see end of post for detailed
horde.log)
The reason for that as far as I could figure it out:
_buildLoadQuery calls _buildParentIds in line 193 of sql.php:
$parent_where = $this->_buildParentIds($root, $loadTree, 'c.');
_buildParentIds' SQL query does not find anything and so returns '' in
line 267.
Then _buildLoadQuery goes on like this in line 194:
if (!empty($parent_where) && !is_a($parent_where, 'PEAR_Error')) {
$where = sprintf('%%s AND (%%s)', $where, $parent_where);
}
so the where part is not included and the query ends up being "get all".
|
| Show full article (3.64Kb) |
|
no comments
|
|
  |
Author: Jan SchneiderJan Schneider
Date: Jan 21, 2007 03:34
Zitat von Andre :
> Hello, I have the following problem:
>
> Complete dump of array vars are shown to the end user,
> how can I prevent this ?
Sorry? I have no idea what you're talking about.
Jan.
--
Do you need professional PHP or Horde consulting?
http://horde.org/consulting/
|
| |
|
no comments
|
|
  |
Author: Michael M SlusarzMichael M Slusarz
Date: Jan 20, 2007 21:27
Quoting Matt Selsky columbia.edu>:
You will need to update framework/MIME's package.xml file to indicate
that PHP 4.3.0 (at a minimum) is needed for the MIME package now (I
think it is currently set to 4.0.0).
michael
|
| Show full article (0.84Kb) |
|
no comments
|
|
  |
Author: Chuck HagenbuchChuck Hagenbuch
Date: Jan 20, 2007 11:34
Quoting Duck obala.net>:
> whups_attributes and whups_queries are still without indexes.
>
> I guess that whups_attributes should have a primary key of ticket_id +
> attribute_id.
Attributes need to be entirely rewritten. I'm not really sure it's
worth mucking with them for now at least on a main distribution basis.
> And if query_id in whups_queries is unique shoudl be a primay key too.
Not sure how I missed that. Adding it now.
-chuck
--
"we are plastered to the windshield of the bus that is time." - Chris
|
| |
|
no comments
|
|
  |
|
|
  |
Author: AndreAndre
Date: Jan 20, 2007 11:26
Hello, I have the following problem:
Complete dump of array vars are shown to the end user,
how can I prevent this ?
Thank you for any help in advance.
Best Regards,
Fernando Ribeiro
|
| |
|
no comments
|
|
|
|
|