These bugs are known to me:

A. Implementation: 

I. The format mapping is not 100 % correct.

This concerns the ldap-format plugin.  It means, that the transformation
from any xmlformat-... format to an LDAP specific format and vice versa
is either incomplete or wrong.  The ldap-format plugin does not fully
comply with the requirements, yet, as laid out by 

/usr/local/share/libopensync1/schemas/xmlformat-*.xsd.


II. Referrals and continuations

Currently there is no support for LDAP referrals or continuations.
(LDAP_RES_SEARCH_REFERENCE).  This is a problem for distributed
LDAP systems, where, say, the addressbook is stored on one LDAP server,
and the calendar on another LDAP server.


III. Photos:

Photos are treated as every other kind of LDAP attribute:  Without any
special measures.  For example, paths are not being resolved.
This is probably wrong.


IV. Base64 encoding:

The plugin does not make any difference between ordinary strings and base64
encoded strings.  Both categories are treated as the same kind of
ordinary strings.



B. Design:

I. Storage scheme on the LDAP server:

I am not really happy with the way the data is stored on the LDAP server.
Only the object type "contact" maps to LDAP schemata that are already known
and used outside of the opensync project: 

	evolutionperson.schema
	inetorgperson.schema

All the other three object types are mapped in a way that robs
the LDAP attributes of their original meaning.  Just to get
a typical name-value pair stored.

And even with the "contact" object type there is currently almost
no support for the situation that a particular XML element occurs
more than just once.  The *.xsd files allow for such multiple 
occurrences.

I deliberate on whether I should make use of multivalued LDAP
attributes.  While this would be valid from the LDAP perspective,
it simply does not work, when XML attributes have to be dealt with,
or when I have to make use of those LDAP subentries for other reasons.

Maybe I should create opensync specific LDAP schemata? Don't know.
I would have preferred using schemata that are already existing and
used by other people...