Re: [046-lc-edit-relative-URI] proposed patch

I have integrated this change into the IRI spec. Here are my
observations:

- At one point, I was rather affraid that this wouldn't work very
   well, because with just "relative reference" in the URI spec,
   there is nothing to distinguish the URI version from the IRI
   version.

- However, I found out that URI-reference/IRI-reference are still
   okay terms, which include relative 'stuff', and so the term
   'relative IRI reference' is available, and I have used it.

- This suggests that instead of 'relative reference', using 'relative
   URI reference' in the URI spec would make things clearer and more
   straightforward. This does not have to affect the name of the
   production rule (relative-URI-ref would be too long, and rel-URI-ref
   rather clumsy). I'll leave it to Roy whether he wants to make this
   change or not.

- Drawing a diagram to make sure I got everything right, I found the
   following anomaly:
   absolute-URI is used to refer to 'stuff' that is both absolute and
   doesn't have a fragid. This seems to make sense, except that URIs
   are now defined as 'stuff' that is absolute. So the only distinguishing
   feature of 'absolute-URI' against 'URI' is that they don't have a
   fragid.

   Using the 'stone lion' example from an earlier discussion, we have
   URI = lion
   relative = stone
   absolute = living
   with fragid = adult
   no fragid = baby

   Before, we were fighing with the issue of whether using 'lion' in the
   term 'stone lion' was appropriate, and decided it was not (because,
   after all, a stone lion isn't really a lion). Now, we are faced with
   the question of whether 'living lion' is an appropriate term for
   'baby lion'. While of course baby lions are living lions, it doesn't
   seem to be appropriate to use a general term (living lion) for something
   specific (baby lion).

   What do others think?

Regards,     Martin.


At 17:54 04/08/27 -0700, Roy T.Fielding wrote:

>A request has been made to remove all usage of the term relative-URI
>from the specification, now listed as issue 046-lc-edit-relative-URI.
>
>The following patch will make that change to draft-06.xml. In spite of
>its length, the change remains IMHO editorial in nature.  If you do
>not think it is allowable in the author's 48 hours of modifications
>prior to RFC publication, or if you disagree with the patch, or if
>you feel that this level of churn isn't worth it just to satisfy
>confusion, then please tell us now by replying to this message.
>
>Cheers,
>
>Roy T. Fielding                            <http://roy.gbiv.com/>
>Chief Scientist, Day Software              <http://www.day.com/>

Received on Monday, 20 September 2004 03:43:19 UTC