Re: AW: Peter's slides about the MOF metamodel

On 6 Aug 2008, at 11:23, Alan Ruttenberg wrote:
[snip]
> It is my impression that we are doing neither. My take is that we  
> have a model related to, but not of, the syntax,

No. It's a model of the syntax. Models abstract.

> as the order of elements in the expressions is not captured and in  
> the functional syntax order matters.

The order of elements in the expressions is not part of what should  
be modeled. The part that is modeled is that e.g., a ClassAssertion  
has a class expression part and an individual part. They way those  
aspect of the model are *realized* in the functional syntax is by  
order of arguments, but, of course, it needn't be *that* order, nor  
need it have parenthesis, or lack a comma between the arguments, or  
*have* a comma between the arguments.

These are given by a *grammar*.

> So to get to 1 we would pull the model out and make mapping  
> explicit or to get to 2 we would add ordering to the model.


I do not believe we need to do either of these things. I also believe  
we are in a rathole prompted by a seemingly innocuous but rather  
naive question (from Michael). Now we're off in some there's A Big  
Deal mode.

I personally believe that reasonable developers can make sense of the  
spec. I also believe that we can do a reasonable job of ensuring what  
is, after all, the key point: that the syntaxes conform to the model  
(when viewed from the right level of abstraction). We have an  
existence proof in the form of the OWL API.

I do not believe it is necessary to produce scripts either. Actually,  
I don't find a strong need to heavily document the relationship  
between the various bits where it's quite obvious, though I don't  
object to adding some text.

Could someone point to actual problems? The order thing just isn't one.

Cheers,
Bijan.

Received on Wednesday, 6 August 2008 16:33:20 UTC