Re: peace and quiet

-----Original Message-----
From: keshlam@us.ibm.com <keshlam@us.ibm.com>
To: xml-uri@w3.org <xml-uri@w3.org>
Cc: xml-uri@w3.org <xml-uri@w3.org>
Date: Wednesday, May 24, 2000 10:13 AM
Subject: RE: peace and quiet


>Note that Literal-and-discourage was the conclusion that the Plenary
>Group's straw poll reached,

No, it was litteral and warn. This was a plenary-confidential document, but
if the plenary
is happy to discuss it in this forum let us be precise. The litteral bit:

"When relative URIs are used in namespace declarations, then for purposes of
Namespace Recognition and Attribute Checking, they should be compared
literally; two values are identical if and only if their unnormalized,
unexpanded, uncanonicalized, raw forms are identical, regardless of whether
their absolutized or canonicalized forms are identical or different."


>for more or less the reasons we've seen in the
>previous notes under this subject. This solution is something that both
>those who think relative syntax is evil, and those who understand the
>objections but are already using it, can live with.
>
>
>If it helps, think of Literal-and-discourage as really being Forbid with a
>migration path.

No, the only warning is:

"The Namespaces Recommendation should be changed to warn users that,
although relative URIs are syntactically legal in namespace declarations and
as namespace names, they do not in themselves have the characteristics of
universality, uniqueness, and persistence which are desirable in namespace
names, because their effective value can change when the base URI of their
enclosing document changes. "

That warning is one which assumes the correct URI RFC interpretation of
URIs, and warns that t is relative to the document base address.  That is
motherhood and apple pie for anyone who knows what a relative URI is -- or a
relative unix path of course.

What is NOT said is: "Processing of relative namespace URIs according to
this specification may lead to failures in systems which also process them
according to the URI specification."

> And it gives us a direction for the future without breaking existing
>practice, at the slight cost that those who really want to absolutize the
>name have to accept responsibility for doing so explicitly.


And the major cost that they will be inconsistent with any XPath and XSLT
processing which has occurs. And chemical plants blowing up all over the
place ...
(see my previous posting)

Tim BL

Received on Wednesday, 24 May 2000 17:40:42 UTC