Re: another charmod comment

There's a get-out clause that I increasingly see specs use, along the lines of:

[[
This specification describes certain behaviours.  Implementations of this 
spec are not required to actually conform to these behaviours as long as 
the results obtained are such as would be obtained by such conformance.
]]

If applied to charmod, does it help your concerns?

#g
--

At 02:00 PM 5/24/02 +0100, Jeremy Carroll wrote:


>This feels like a personal comment, but I would be happy if others in
>the WG wished to support it.
>
>[[[[
>
>Charmod does not conform with itself.
>
>This is intended as an issue with sections 1 and 2
>concerning the scope and applicability of charmod,
>and I apologize for the rather smartass way of
>articulating it.
>
>The exposition of the issue follows the mathematical
>style known as "reductio ad absurdum", the contradiction
>being the above.
>
>A particular non-conformance is that charmod
>puts no obligations upon implementors of charmod.
>It, does, of course, indirectly place many obligations
>on implementors of other W3C recommendations.
>An argument that charmod will have no (direct) implementors
>would be disingenious. e.g. I would expect/hope library
>writers to provide a function that checks that
>unicode strings are in normal form c and do not
>start with a composing character.
>Such an implementor would be able to claim charmod
>conformance without having been specifically obligated with
>
>[I] Specifications and software MUST NOT assume a one-to-one mapping
>between character codes and units of displayed text.
>
>since charmod has failed to say that implementations of charmod must
>
>[S] Every W3C specification MUST ...
>specify that implementations MUST conform to the requirements applicable
>to software,
>
>I find a similar problem when drafting text
>for the current round of RDF recommendations.
>Charmod is framed such that for charmod conformance
>a W3C recommendation must have a dictatorial relationship
>with its implementators. The RDF recommendations
>are more relaxed about their power. They define
>RDF documents and their meanings. They do not
>indicate what implementators should do, nor do they
>tell content writers what to do. It is difficult
>to keep reinterpreting the requirements of charmod
>into this less confrontational style of recommendation.
>This is made more difficult by charmod's own
>confrontational style, and expectation that
>other recommendations share such a stylistic
>preference.
>
>
>I believe that the charmod editors are trying to
>say something that is both meaningful and has my
>support.
>I fear that the framing of the scope of charmod
>in sections 1 and 2 has not done justice to the
>range of recommendations produced by the W3C
>nor the range of different relationships
>between recommendations and their users (both
>implementors and content producers).
>
>]]]

-------------------
Graham Klyne
<GK@NineByNine.org>

Received on Friday, 24 May 2002 09:40:03 UTC