Re: Pointer Methods in RDF

On the whole, the pointer framework described by this document seems
to be well thought out. Some of the work they've done touches on the
work that the original "xpointer() XPointer Scheme" was going to do,
but I think that's ok. I noticed two areas on which we might want to
provide feedback.

The first is a bit trivial, but...they define an element (an RDF
property actually, but one that's serialized as an element) named
"ptr:XMLNamespace". I believe that technically we own all local names
that begin "[Xx][Mm][Ll]". 

The second point is one that I think we should encourage them to address.
Although they call out pointing into XML as one of the use cases, many of
the pointers that they provide operate (or appear to operate) on the
character serialization of the resource.

Maybe the answer is as simple as saying "don't use line/char, char
offset, byte snippet, etc. pointers when pointing into XML documents"
but I don't think they say that.

It's not clear what meaning one could associate with a pointer range
that started in the beginning of an end tag and ended in the middle of
an attribute name on some following start tag. Or even if it's
coherent to assert the existence of such a range in an XML document.


                                        Be seeing you,
                                          norm

-- 
Norman Walsh <ndw@nwalsh.com> | certain: adj., insufficiently analyzed
http://nwalsh.com/            |