Dave Pawson <[EMAIL PROTECTED]> writes:

> On Tue, 2005-07-19 at 21:23 +0900, Michael Smith wrote:
> 
> > When some tool developers add support for it and others don't,
> > users start asking, "Hey, when I use tool X (which supports the
> > PI), things work the way I expect, but when I use tool Y (which
> > does not support the PI), things don't work the way I expect."
> 
> Your antagonism to this idea is exceeding strong Mike.
> Is there something else?

Nothing else. I simply do not think it is a good idea.

And one thing in particular I find a little worrisome about the
idea is that some people on the list have expressed only
reluctant/skeptical/guarded support for it.

It seems to me that kind of reluctant agreement is what ends up
giving us specs like the W3 XML Schema recommendation. I remember
one of James Clark's early comments about the W3 XML Schema rec
being something along the lines that he could not understand how
40+ extremely capable people could get together and produce such a
"disaster" of a specification.

I do not know how it happened in that particular case, but I would
speculate that in part at least it may have been the result of
some people reluctantly agreeing to things.

> The prime purpose seems to be to coalesce common usage
> using something an XML processor can happily ignore.

The current state of things as far as I can see is that we have
two RELAX NG-aware editing applications that have support for a PI
of this type. As far as I know, none of the other existing RELAX
NG-aware editing applications support such a PI, and there are at
least two (XML Buddy and nXML) that I think we can say (given the
public comments of their developers) will never support such a PI.

So I don't think that what we would end up with by publishing a
standard for a schema-association PI could ever be described as
"common usage". Because users will find that some tools support it
while others don't.

And I cannot buy the argument that because some editing
applications have already implemented support for a PI of this
type, but using a different name and syntax for it, we now ought
to go ahead and agree on a common name and syntax. I would
personally much rather see those editing applications de-support
the existing PIs.

> I see no harm, and benefits, not least of which
> is avoiding annoying users.

I see harm. And as far as the benefits, I think we really ought to
instead try to come up with with a standard mechanism that
provides the same benefits without requiring an explicit schema
location to be specified in individual document instances.

  --Mike

Attachment: smime.p7s
Description: S/MIME cryptographic signature

Reply via email to