Re: [xml-dev] XML vocabulary for expressing constraints?

[davep <davep@dpawson.co.uk>]

On 13/12/13 17:02, Steve Newcomb wrote:
> On 12/13/2013 11:11 AM, davep wrote:
> > On 13/12/13 14:46, Steve Newcomb wrote:
> > > Unambiguous non-interactive communication is hard to do.
> > Are you taking two extremes Steve?
> > If you (initially) take a mid ground and deal in what you say it isn't
> > it seems doable.
> >    It's an image of X
> >    It's a text document describing Y
> >    It's a video showing Z
> >
> > I'm less interested in the stream of bytes, more in what it is
> > representing.
> >
> > Your abstract idea of the constitution is .. harder, too abstract for XML?
> >
> > Is it this abstract idea you were getting at?
> I was only trying to demonstrate that the href value's semantic is
> extremely ambiguous, unless and until there's sufficient context, shared
> by the author/sender with the user/recipient, to make it unambiguous, or
> at least to restrict its ambiguity.

This is a point Rick has made many times on this list.
Two issues here. Machine to machine interaction and human to human
interaction.

> When people speak sincerely, they generally intend to be speaking about
> specific things, and they *believe* that those things are what they are
> talking about.  In a dialogue, it's possible to detect and resolve
> misunderstandings and confusions.  In a non-interactive context -- one
> that doesn't involve a reverse channel from the recipient to the sender
> -- it's much harder for either end of the channel to be sure of anything.
Two way XML? That's novel. Liam?
email? My machine gets something from you, Logic says it is 
incomprehensible (for some definition of incomprehensible). The machine
sends us both an email and we 'talk' to resolve the conflict?
This bridges the gap, machine to human? I really don't think (today)
two machines could resolve ... US vs UK semantics.


Re context surely the root element of the xml instance provides *some* 
context, perhaps not the same as two people having spoken, but hopefully
before you send me an XML instance I'm expecting it, hence have some
sort of expectation?


> No, I don't think it's too abstract for XML, any more than it's too
> abstract for English.  Both English and XML can be used to convey
> thoughts about abstractions.  In my own work, I often see XML used to
> express kind-of and instance-of relationships, and, yes, to attach
> thoughts to those relationships.
Yes.... but again I'd constrain this to the human to human interaction,
even if via a machine to machine transaction? I.e. I can read the XML
and gain more from it (as a person) with again the background of
previous interactions.


> Indeed, some such semantics are built into XML itself: it has element
> types and element instances, for example.  The latter kind of
> relationship is a pretty abstract thing, but the only reason it's hard
> to talk about is that XML lacks a rhetoric for talking about it.
Sorry. Over my head Steve.


> Digression: It could have such a rhetoric inherently, but I think would
> be the wrong way to go about it; it would be a one-off solution to a
> much more general problem, one that would leave the general problem, er,
> unaddressed.  Anyway, in all my years of working with this stuff, I've
> never had an occasion to talk about the *relationship* between a
> specific element type and a specific instance of that element type.
>
> Nevertheless, it should be possible to talk about such a relationship,
> and to do so unambiguously. Question: What would make that possible?
> Answer: A way to establish a context for references to such
> relationships -- a context in which the necessary rhetorical facilities
> exist.  And then, of course, one would actually use those facilities.

Sounds much closer to be a human activity than something a machine could do?

> Note that I do NOT claim that that's easy or simple in any sense.  I
> estimate that it's just about as hard as living with others in peace and
> prosperity.  In fact, I suspect the two problems are closely related.
Sounds like a fair summary <grin/>

Interesting subject




regards

--
Dave Pawson
XSLT XSL-FO FAQ.
http://www.dpawson.co.uk