Re: 2001-09-07#5 Literals

DanC:
> These proposals are sufficiently long and complex that I'm
> pretty sure it would be more straightforward to code
> first and describe the code after the fact.


I think it is clear that I am going to need to break this up into smaller
chunks.

The vast majority is intended as just an (optimistic) description of current
practice, and clarification of M&S.

An example of what I mean "optimistic" is the lang tag thing. In M&S
language tags are part of the literal, in RFC1766 and RFC3066 lang-tag
comparison is case insensitive. That doesn't mean that those few RDF
implementations that support lang tags actually do the comparison case
insensitively. But I think the implementators would accept that they had
made a mistake.

The unicode normalization stuff is an exception to that "description of
current practice", except that charmod which inspires it has been
deliberately written so that code that is designed to work with the US-ASCII
subset of Unicode is OK.

The other exception to "current practice" is canonicalization as the
preferred treatment of rdf:parseType="Literal". There the description of the
code (use XML Canonicalization) is much more straightforward than the code
itself, and it is helpful to have a statement of strong interest from the WG
before coding it.

Jeremy

Received on Thursday, 27 September 2001 11:50:34 UTC