LC comments on CURIE syntax 1.0

Dear all,
I just had a look at and I
have the following comments on the draft:

1) QNames can be expressed without a prefix, in which case they bind to
the default namespace as defined by Namespaces in XML. CURIE have two
forms without a prefix, one with and one without the colon. It seems to
me that without a colon, the CURIE is intended to denote a reserved
symbol in some list predefined by the host language, and with the colon
(but no prefix), there should be some predefined default prefix value.
It should be clarified that what QNames do without the colon, CURIEs do
with it (and that the support for XML Namespace default namespaces is
not required by the spec), and that without the colon, CURIEs work very
differently from QNames.

What is the rationale for introducing the "set of reserved values"? What
is the rationale for dropping support for XML NS default namespaces? And
what is the rationale for changing QNames' default namespace syntax
without a colon to CURIEs' syntax with a colon? 

I think all three issues here would confuse potential users, if they are
not explained and motivated better in the spec.

2) Section 3 says 
        "The CURIE prefix '_' is reserved for use by languages that
        support RDF. For this reason, the prefix '_' SHOULD be avoided
        by authors." 

This short paragraph must be expanded to say whether languages that do
not support RDF are actually allowed to use this prefix (and to do
xmlns:_="..."); and what exactly is meant by "use by languages that
support RDF". I suspect that the intended use is for bnodes, but the
spec should say so.

In fact, I would suggest that if interoperation with RDF is desired, and
I think it should be, then _:something should always denote a blank node
in the RDF sense, and that the use of the _ prefix in languages that do
not support RDF blank nodes should be forbidden. This would enable a
language that hasn't supported RDF blank nodes to evolve easily into one
that does.

3) Appendix A defines a set of datatypes, which I suggest should be
expanded with URIsOrSafeCURIEs - a list of URIorSafeCURIE. The omission
of such a datatype is curious, even though I'm unsure if my proposed
capitalization should be the final one.

Please note that I'm not subscribed to www-html, so any replies should
be directed (or CC'd) to me.

Hope it helps,
Jacek Kopecky

Received on Friday, 23 May 2008 19:10:39 UTC