What is the differences between DSMLv2 and XED?
In our view DSML is half a solution. It provides an XML encoding for LDAP operations, but still assumes that directory attribute values are in the LDAP string format. If you try putting XML in directory attribute values it has to be escaped before it can be transported in DSML. Even worse, DSML represents LDAP extended operations and controls as the base 64 encoding of the octets of the BER encoding of the operation or control, which is practically indecipherable to a human reader.

The XED alternative to DSML is XLDAP (a 100% XML version of the LDAP protocol). XLDAP represents XML in directory attribute values natively as XML without escaping. It also provides an XML markup for values of existing complex syntaxes like directory schema definitions (attribute type definitions, object class definitions, etc). LDAP controls and extended operations are also represented as XML markup, never as base 64 or BER.

The XED framework allows directory attributes to be created with syntaxes defined in XML Schema, or with a DTD. The directory then understands the structure of the data and can do semantically driven verification and matching of the data. For example, the directory knows that "true" and "1" are equal values for the XSD boolean datatype, and knows how to order values of the XSD dateTime datatype with time zones. Component matching can also be invoked to match specific elements or XML attributes in directory attribute values. The best one can do with DSML is substring matching on the XML syntax, which is a bit hit and miss.

Because XLDAP is algorithmically derived from the underlying ASN.1 definitions of LDAP and X.500, every extension to LDAP automatically gains an XML representation in XLDAP. On the other hand, DSML is entirely hand crafted.