We already have a working parser for TRL in an RFC822-like syntax.
We do not have a need to embed TRL documents in other XML.
An RFC822-like syntax is slightly easier for humans to read.
It is a stated future direction of the project to support XML, and the TRL interpreter code has been organized so that anybody who really cares about this issue could drop in an XML parser and make it go. The desireability of doing this is not a good topic for the list; please make your argument with working code or not at all.
Why don't you guys just use LDAP?
There are several reasons LDAP is inadequate for Trove's needs. Here is a non-exhaustive list:
Because LDAP directories are designed for machine-to-machine queries using an LDAP-aware front end; they can't be searched or browsed directly with a Web browser.
One of the things that was removed when X.500 DAS was stripped down to become LDAP was the ability for a server to hand off queries it can't answer to a ring of cooperating server sites. We need exactly this capability.
There's no way to extend or customize the semantics of the LDAP query language in the ways we need (for example, to support the right semantics of discriminators).