Re: possible issue with LEIRI definition in draft-duerst-iri-bis-02.txt

Hello Paul, others,

At 02:31 08/03/05, Grosso, Paul wrote:
>
>I was just rereading the LEIRI section of
>http://www.ietf.org/internet-drafts/draft-duerst-iri-bis-02.txt
>where it says:
>
> The syntax of Legacy Extended IRIs is the same as that
> for IRIs, except that ucschar is redefined....
>
>In section "2.2. ABNF for IRI References and IRIs", it
>has a production for IRI (that has a required scheme)
>and another for IRI-reference.
>
>One could read section 7 to say that a LEIRI must match
>the production for IRI which would mean there could be
>no such thing as a relative LEIRI.  I'm quite sure we
>don't want this.

True indeed.

>I think section 7 needs to say:
>
> The syntax of Legacy Extended IRIs is the same as that
> for IRI-reference, except that ucschar is redefined....

That's unfortunately not good enough. There should
be a clear correspondence, as follows:

LEIRI               ->   IRI

LEIRI reference     ->   IRI reference

I have fixed this by adding the following short paragraph
after "The iprivate production becomes redundant.".

>>>>
Likewise, the syntax for Legacy Extended IRI references
(LEIRI references) is the same as that for IRI references
with the above redefinition of ucschar applied.
>>>>

Please tell me whether this is appropriate for you.
It may be that some of your specs currently use the
term LEIRI when they indeed mean an LEIRI reference,
in which case they should be adjusted.

It may be that indeed all or most of your specs want
to reference LEIRIs. In that case (especially if it's
all), it might be approriate to rewrite section 7 of
the current draft to concentrate on LEIRI references
(maybe as far as changing the title to Legacy Extended
IRI References). In particular if it's all your specs,
the rewrite should be straightforward. Please advise.

In general, both the URI spec and the IRI spec are careful
to use the correct terms where only one of them applies,
but they do not necessarily always use both terms if
both apply; doing so would make the spec unreadable.
This is usually covered by some general clause saying
that certain things also apply to references,...

Regards,    Martin.



>since the production for IRI-reference is:
>
>  IRI-reference = IRI / irelative-ref
>
>making IRI-reference the most inclusive one.
>
>paul


#-#-#  Martin J. Du"rst, Assoc. Professor, Aoyama Gakuin University
#-#-#  http://www.sw.it.aoyama.ac.jp       mailto:duerst@it.aoyama.ac.jp     

Received on Wednesday, 30 July 2008 01:27:54 UTC