Re: Why do we have a component model?

On Mar 7, 2005, at 6:57 PM, Yalcinalp, Umit wrote:
[snip]
> I am quite worried about replacing the component model with something
> which may be as hard to decipher this late in the game.

No one wants that, I'm sure. This is why I'm against Arthur's clever 
proposal...it only might *seem* less complex.

> I am not yet convinced that we need to make drastic changes like this
> instead of fixing bugs at this point. Component model provides an
> abtraction to talk about composition within the language (i.e. WSDLs),
> composition with another (i.e. Schema), composition with constraints...
> By the time you get rif of the abstraction away, the things that one is
> abstracting from, ie. Documents will have to be described in such a way
> that we will lost the conciseness of the abstraction, and I fear that 
> we
> will end up with a more complicated thing in our hands.

Yes. But we should make *clear* the source of the complexity and how 
the component model helps. I hope the "Whys of WSDL" will do that.

> Just try to import two separate WSDL documents with embedded schemas 
> and
> talk about them using Infoset(s) only. Put F&P, couple of styles, and
> inheritence on top. I am not sure where one ends up is "simpler" per 
> se.

As Arthur said, it's not at all clear except that you have fewer kinds 
of entity (though you do have to subclass infoset items).

> Complexity of the current spec is not necessarily due to the component
> model, but due to the notion of having different languages, 
> composition,
> constraint mechanisms, inheritence within abtraction and the
> flexibilities that we provide. We can toss away some of the 
> flexibility,
> (like the discussion we had about using multiple languages with WSDL at
> the f2f) and it may be more tractable.

I would hope that in the next last call we can make clear to the 
community the actual source of the complexity so that they can decide 
*what* they really can live with or without. Rich said, "Get rid of 
inheritance" (and what ever else forced the component model), but I 
think that if the reason for getting rid of a feature is *solely* to 
get rid of a certain way of explicating it...well...that's daft. 
However, if the community looks at the complexity of the *langauge* and 
says, "Well, you know, we only need one kind of document 
composition...", then we have a point of genuine simplification.

> I would like to get to a point of closure for WSDL. I am concerned that
> we are not getting to that point, personally.

Amen.

Cheers,
Bijan.

Received on Tuesday, 8 March 2005 02:00:50 UTC