Changes have been made in CVS for this ticket:

http://cvs.horde.org/diff.php/turba/lib/Driver.php?r1=1.57.2.58&r2=1.57.2.59&ty=u

Changed the order of the conditional checks to put the (expensive) 
array_key_exists() call at the end....and committed, thanks!



I left the similar code in toHash() since I'm not sure off the top of 
my head if any other code paths would be affected.

Changes have been made in CVS for this ticket:

http://cvs.horde.org/diff.php/turba/lib/Driver.php?r1=1.195&r2=1.196&ty=u

Thanks, you are right, I didn't notice that code path before.



Attached is an updated patch to fix this issue. I also updated

the comment to be more precise about it.

Yes, but follow the code in toDriverKeys() through some more and 
you'll see that if we are not saving the composite field to storage 
(in other words, if $this->map[$key]['attribute'] is empty) then the 
composite field is parsed out into it's individual fields, and then 
*those* values are saved to the $fields array when the arrays are 
merged.



To test this, just try to add a new contact in the "out of the box" 
localsql source using Turba's UI....it nulls out all the individual 
name fields.

Do you mean this part?



+                if (isset($hash[$mapfields])) {

+                    // Add composite field

+                    $hash[$key] = null;

+                    break;

+                }



It sets $hash[$key] which points to the composite field, not the 
individual part.

This patch causes the individual fields that compose the composite 
field to be overwritten with null values when the composite field is 
not being saved as a separate field in the backend storage...

Hello together,



composite fields like "name" are not computed properly f.e. during import.

Attached patch adds support for that. Note: It is intended

to use array_key_exists() in the code.



I've seen there is "similar" code in the $driver->toHash() fucntion.

Maybe that can be remove if the new code gets in place.



Thomas