[
Lists Home |
Date Index |
Thread Index
]
From: Gavin Thomas Nicol [mailto:gtn@rbii.com]
On Tuesday 25 February 2003 03:36 pm, Bullard, Claude L (Len) wrote:
>> and Gavin concludes:
>>
>> "The standardization efforts within a given domain might well normatively
>> define themselves in terms of the syntax and infoset defined elsewhere..."
>Note that I said "might well", I didn't say "should", or "must"
Yes.
>and you also left off an important bit:
"...but that doesn't mean the syntax and infoset are responsible for
interoperability."
>In other words (and as I've said) I do not think that having a single syntax
>or infoset results in interoperability at the application/software component
>level*. It might *simplify* things for a specific set of applications, but
>that's about it.* Scheme/LISP proved that...
I agree (thought I did anyway). But what I'm aiming at here is determining
which subset simplifies the most things for the most applications at the
cost of sharing the pain across the board. If that can't be determined,
there is little sense in a subset and if "the subsetting is begun" it
better never be formally recognized by the W3C outside an application
specification. SOAP is happy in its own bath, one permathread is
reduced to cotton, and Liam's task won't be a long one.
>> XML-SW is as close as any proposal I've seen put forward that
>> gets the most benefits for the best sharing of the pain of the
>> implementation. I would think it in the best interests of the
>> W3C and the XML community to start there.
>I disagree. As I said at the start of the thread, XML-SW bundles namespaces,
>xml:space, xml:lang, xml:base and infoset, which I think is a mistake.
Fair enough. I said it is a place to start. If all of the subset supporters
and detractors are arguing about the same document, then that is a satisfactory
way to begin. If they can't do that, there is little use in beginning. Somewhere
in the beginning of my entry to this thread, I used the term "strawman" not
inopportunely. Consider, however, that unbundling namespaces and the syntax
will mean more committee work, maybe multiple authors, and so on. It makes
the task longer, but that is not my concern anyway. We are just here in a
place where people often cite the aledgedly irrefutable proposition that
small self-contained specs are better than spiderWorkz. Mileage varies.
len
|
