XPath host language for querying several XML fragments

Hello. I'm developing an XPath host language for querying XML fragments from
a repository
[http://www.ietf.org/internet-drafts/draft-godoy-webdav-xmlsearch-00.txt] .

Since there may be several XML fragments within the search scope, the
proposed approach would have implicated a terrible evaluation cost if
arbitrary complex queries had been allowed. So I'm trying to distinguish a
subset of "basic" XPath features (which would be REQUIRED) from other
features that would be OPTIONAL (the latter may be supported or not, as
necessary for each particular use-case). As a consequence, features from the
"basic" subset are intended to suffice for most use-cases, and their
evaluation may rely on a relational database for optimized processing.

I would appreciate your feedback about this proposal, as well as any
correction about XPath-related issues .

Regards,

Javier Godoy