Re: relativeIRIs (draft-duerst-iri-02)

Hello Carlos,

Many thanks for your comments.

At 10:14 02/10/24 +0900, Carlos Viegas Damasio wrote:

First, apologies for my mailer mangling your name.

>I have to questions regarding relative IRIs, which apparently have not
>been addressed.
>
>1)Is it relative IRI of the form "?y" (or even "?") allowed in IRIs ?
>Apparently not, since irel_segment requires at least one character
>
>                 irel_segment   = 1*( iunreserved | escaped |
>                                  ";" | "@" | "&" | "=" | "+" | "$" | ","
>)
>
>This is related to an example of RFC 2396 (section C.1 normal
>examples), where "?y" is implicitly assumed to be a valid relative URI:
>
>       ?y            =  http://a/b/c/?y

The IRI syntax, in this respect, tries to reflect the RFC 2396
exactly. Looking at RFC 2396, it seems to me that it has exactly
the same problem. I don't know whether the example is wrong, or
the syntax is wrong, or there is some other problem.

I have copied the uri list, where the experts on RFC 2396 can
help us.


>2) Also, shouldn't irel_segment obey to the syntax of isegment with the
>exception of ":" not being permitted?
>    After all, irel_segments will be isegments after resolution of
>relative references.
>
>So, my suggestion is to replace irel_segment production by an
>appropriate version of isegment, something like:
>
>              irel_segment       = *irpchar *( ";" irparam )
>                 irparam       = *irpchar
>                 irpchar       = iunreserved | escaped |
>                                 "@" | "&" | "=" | "+" | "$" | ","

I just pointed out the lack of *( ";" irparam ) in rel_segment
recently on uri@w3.org. Please see the thread starting at
http://lists.w3.org/Archives/Public/uri/2002Oct/0003.html.


Regards,    Martin.