Re: IRIs and XLink/XPointer spec

On Wednesday, December 11, 2002, 12:19:53 AM, David wrote:


DGD> At 1:19 AM -0500 12/6/02, John Cowan wrote:
>>I think SPACE in IRIs (as opposed to %20) is a bogon, and shouldn't be
>>allowed, the xpointer() scheme notwithstanding.  The other schemes do
>>not require it.

Yes, I agree that the amount of breakage that is produced by allowing
space seems disproportionate.

DGD> I strongly agree with John on this issue, because. In the TEI we are 
DGD> seriously considering an option to perform all pointing with 
DGD> XPointers and URI fragments rather than the ID mechanism.

Hooray! Moving from single-instance pointing to scalable,
multi-instance pointing is a good design pattern. Pointing within the
same document to an element that happens to have an id already should
not use a separate mechanism to more general pointing using a URI
reference.

DGD> This really
DGD> depends on the need to escape spaces in URIs, because the best 
DGD> equivalent to IDREFS attributes is space delimited sets of URI 
DGD> references.

Yes. Without structured attributes and with the only attribute list
mechanism in W3C XML Schema being a space separated list, this really
is the only possible way to do it and keep the result as an attribute.

DGD> The use of the other characters as URI delimiters is pretty uncommon, 
DGD> but the same is not true for space.

Yes.

-- 
 Chris                            mailto:chris@w3.org