"John S. Gage" wrote:
> Thomas Beale wrote:
>
> > I knew this, but I also knew you would respond with the latest status
> > on this! (BTW - just wait till they try and do UML -> XML-schema.
> > We've already been through the loop once, and it's painful. I've been
> > told not to deride XML-schema too much, but let's jsut say it's
> > missing a couple of key OO concepts).
>
> I, personally, think that this is an extremely key point. I have heard
> the benefits of XML touted to the rooftops, but increasingly I have
> begun to believe that XML is very much at the bottom of the food chain.
(I'm smiling at this analogy)
> Thomas seems to clearly advocate using a *programming language* to model
> the domain. He wants Eiffel, but the real point is that he wants a
Well, I want a formalism which can act as both a modelling language and a
programming language. Eiffel does both, out of the box, whilst one might
claim that UML + Java + Java-capable UML case tool is nearly equivalent -
UML gives you the modelling bit, Java the programming bit, the tool the
error-free transformation in between.
Eiffel/UML conversion already exists, BTW, but a better integration with
Rose is on the horizon.
> programming language. Am I correct? In addition, despite David's use
> of Corba, if I'm not mistaken, all his code is written in Java. If, in
> fact, it's the programming language that counts, then (here comes the
> controversy) IMHO Java wins going away. Just as a single example:
> medicine is supposed to become paperless, yet most computers in medicine
> generate *far* more paper than computer-less areas. Printing is always
> going to be extremely important in medicine. Well, guess who has by far
> and away the best printing API? Why would I want to program in Eiffel,
> when it probably doesn't even have a print API? Please tell me I'm wrong.
Printing is an activity fairly far "down the food chain". By the time you
see your content from a GEHR kernel implementation of COAS running on Zope,
you are probably looking at an XML browser on a PC in another wing of the
hospital; the printing capabilities will be to do with this browser, not the
back end. Or else you'll be using a custom VB app; same argument.
> > Well, I think that UML is ... not bad for a diagramming notation. It's
> > been though enough iterations to cover a lot of things. Its big
> > drawback is that it has no formal grammar or definition.
>
> i.e. it's not a programming language. It's just the rich man's XML.
I wouldn't compare it to XML - XML does actually have a grammar (not a nice
one though); it's just a diagramming standard.
- thomas beale
