[
Lists Home |
Date Index |
Thread Index
]
>
> > However, does this then mean that Xalan and Saxon are wrong in
> > accepting
> >
> > file:myDir/myFile
> >
> > as a relative URI?
>
> The new draft for RFC 2396bis[1] clarifies this
>
> <q>
> Some parsers allow the scheme name to be present in a
> relative URI if it is the same as
> the base URI scheme. This is considered to be a loophole in
> prior specifications of partial
> URI [RFC1630]. Its use should be avoided, but is allowed for
> backwards compatibility.
> http:g = http:g ; for validating parsers
> / http://a/b/c/g ; for backwards compatibility
> </q>
>
> So if Saxon and Xalan *require* the file: scheme on relative
> URLs they are wrong
> (Eddie, what was the status of this?) but it looks like they
> may provide it for backwards
> compatability. Anyway, we use it at our peril. Half a step
> forward :-)
>
Saxon largely delegates its URI manipulation to the JDK, except for a
few cases where the JDK doesn't offer the required functionality. If the
JDK gets it wrong, I will need a lot of persuasion to implement
workarounds at the Saxon level.
Saxon certainly doesn't require any scheme name to be present in a
relative URL, and it's a surprise to me that it allows it, but if Java
allows it, then I do.
Frankly, the RFC specs on URIs are so incomplete and buggy that there's
absolutely no hope of ever getting a clean implementation, or two
implementations that are 100% interoperable. So I'm happy to leave the
JDK folks to write the bugs rather than writing them myself.
Michael Kay
Software AG
home: Michael.H.Kay@ntlworld.com
work: Michael.Kay@softwareag.com
|