Re: RDF/XML Syntax Working Draft - Internal Draft V1.16

At the teleconference, I got the impression that Dave was after feedback, so
here's some more :-).

I do not believe that we have discussed the following issues (although I
have mentioned some privately to Dave). As such, it is appropriate that they
are not in the current WD; but I would hope they would be addressed before
the next WD.

1: Whitespace
=============
M&S does not describe how to process whitespace in RDF/XML. (Well it does,
but what it says doesn't work).
The grammar in the WD carries that error too.
I have two possible fixes:
a: extend the grammar with an explicit whitespace production
b: add text adequately capturing the whitespace rules.
e.g. "Whitespace is significant in attribute values, where XML whitespace
processing applies, and within literal values. Elsewhere it is ignored."

An example of (a) is the ARP grammar. I prefer (b) for the spec.

2: XML Attributes
=================
M&S and the new WD do not specify that attributes whose name starts in
[xX][mM][lL] are discarded when RDF processing, (although they may impact
XML processing, e.g. xml:lang, xml:space,
xml:someFutureThatHasNotBeenThoughtOf). Again, the grammar can be modified
to capture this, or text added. Again I prefer text in a specification, and
modifying the grammar in an implementation.

3: "(can be empty)"
==================
This phrase occurs a couple of times and should be emphasised in the
supporting text, since XML does not allow this, i.e. the XML Infoset of the
document is missing the supposed text string rather than having an empty
text string as suggested by the grammar. Again, elegance and clarity of
exposition in my mind takes priority over having an exact correspondence
between the Infoset and the grammar.


Jeremy

Received on Friday, 24 August 2001 11:18:00 UTC