Re: URIEquivalence-15 and IRIs

At 11:45 02/07/09 -0700, Roy T. Fielding wrote:

>In other words, simply require that all case-insensitive parts of
>the URI be lowercase and all non-reserved data characters be unencoded
>as part of the definition of valid xmlns values.  Then all of the
>parsers can simply perform byte comparison.

Did you just pick lower-case at random? In general (i.e. domain names),
it would make more sense than upper case. Has there ever been any
preference/convention for the hex letters?

The alternative that would also work is: Never touch namespace URIs/IRIs,
even if you 'know' it shouldn't hurt. I.e. never change %7e to %7E,...

Or even a combination of both:

Never touch a namespace URI/IRI, and if you create one, use lower case.

Does somebody know about cases where namespaces didn't match due to
case/escaping problems? If not, why do you think these didn't come
up until now? If yes, what can we learn from them for this discussion?


Regards,   Martin.

Received on Tuesday, 9 July 2002 21:33:34 UTC