[
Lists Home |
Date Index |
Thread Index
]
I leave the first question for Steve Newcomb because he
is the arch forms expert. A point about contracts:
it is a means to constrain a communication and behaviors.
Without an actual process, there is no sensible answer,
that is, when the child cites its parents to the
school teacher, they are using that citation differently
than they would if they cite it to the parent themselves.
It is easy with attributes to cite parents, or to use
#FIXED defaults. It is having a standard means to
cite the process or handler (eg, what we did with
NOTATIONs) that is the problem.
Yes, a lot of the early works for using SGML to tag
the windowing tools look the same because in any markup
language we egenCode (tag the stuff
we discover) or abstract (create tags for a system
of abstract entities and relationships). Many systems for
that come down to a view/controller sort of paradigm.
In MID, we just spec'd a way to navigate different
views given a database. Not all that revolutionary
because it looks like any system of containers with
navigational controls. At the time, DoD already
had the database standard (87269) and we were
adding a viewing tool. The dilemma is
getting a portable definition of the behaviors.
len
-----Original Message-----
From: Gavin Thomas Nicol [mailto:gtn@rbii.com]
On Thursday 24 January 2002 11:47 am, Bullard, Claude L (Len) wrote:
> Steve's message is pertinent. Maybe XML needs to, as
> a human does, be able to go in both directions. That
> is, a parent constrains the element, but an element should
> be able to cite its eligible parents.
How is that contract to be obeyed?
> Packaging the application was what the MID project was
> about. There turn out to be (in my experience) only a
> few solutions if one can bet past arguing over PIs
> and pointy vs curly brackets.
I'd agree with that. The MID was an interesting application... the
system I did years ago was similar in some ways to it.
|
