[sync] Issues with MySync DS / Android

[Carsten Schumann]

Am 2012-09-02 22:41, schrieb Michael J Rubinsky:
> Quoting Carsten Schumann <grypho at tempus-vivit.net>:
>
>> Am 2012-09-02 20:44, schrieb Michael J Rubinsky:
>>> Quoting Carsten Schumann <grypho at tempus-vivit.net>:
>>>
>>>> Hi Mike,
>>>>
>>>> as fas as I  understand, the backends.php (for turba) does only   
>>>> cover the data providers for the adressbook. These providers   
>>>> (standard horde localsql database) work fine for me. My issue is   
>>>> related to SyncML which is (as far as I understand) not covered by  
>>>> the backends.php, is it?
>>>
>>> No, backends.php defines (among other things) the mappings 
>>> available
>>> and how they map between turba's fields and the backend storage's
>>> fields. The sync code will only export mappings present in this 
>>> file.
>>> E.g., the 'HOME' vCard attribute will only be exported if if the
>>> 'homePhone' mapping is present in backends.php.
>>>
>>>
>>>
>>
>> Ok, that was one thing I did not expect. I assume the  
>> $INSTALLDIR/turba/config/backends.php is the correct file. I had a  
>> look into that file but I found nothing that matches the VCARD  fields 
>> I found in my sync logs. I found some mappings that seem to  be the 
>> mapping between the sql data fields and the internal turba  fields.
>>
>> $cfgSources['localsql'] = array(
>>  [...]
>>     'map' => array(
>>        [...]
>>         'firstname' => 'object_firstname',
>>
>> So far, I understand the mapping mechanism. What I do not understand 
>> is the mapping between "object_firstname" and the corresponding  
>> VCARD field (I think it is the N field). Where can I find that  
>> mapping?
>
> These mappings are not configurable. To see which Turba field is
> mapped to which VCARD attribute see the Turba_Driver::toVCard()
> method. in turba/lib/Driver.php

Mike,

Thanks a lot. I have found the mapping and modified it to sync at least 
the phone number. I will go through all other fields this evening and 
fix their mappings.

Are you interested in a patch that covers both the current and the 
Funambol/MySyncDS mapping as a fallback if the client advertises it?

Carsten