Re: [sc] C# for SIL?



Re: [sc] C# for SIL?

From: Derek M Jones <derek_at_xxxxxx>
Date: Tue, 24 Oct 2006 18:58:42 +0100
Message-ID: <453E5452.6070401@xxxxxx>
Alvery,

> For my opinion, I would expect the safe subset to offer well-defined behaviour, but not necessarily complete consistency of translation. I.e. if we compiled the expression
> 	a=b+c
> it might be translated as

I think we all agree on that bit.

> I would expect that all constructs in the language subset had a well defined unique meaning. This probably means a formal definition using a widely accepted formalism. For the moment, I can't see how the well-defined meaning requirement can be met without the use of a formal semantics to prove that it has been met. However the use of formalism is not the requirement. 

So what kind of meaning is "well defined unique meaning"
exactly?  Is it different that a "well defined meaning"?

What do you think of my days of the week example?  I don't
see any of the formal definition languages having a problem
handling this definition as part of the overall definition
of a language.

> I would permit the definition of the safe subset to include not just syntactic constraints, but also constraints on the semantic interpretation, so long as those constraints didn't contradict the original specification for the full language. 

Does the days of the week requirement meet your idea
of "fully and precisely defined"?

> I've probably missed some of it. Any thoughts? 

I think you are still under the impression that your idea
of "fully and precisely defined" is the same as everybody else's.

-- 
Derek M. Jones                              tel: +44 (0) 1252 520 667
Knowledge Software Ltd                      mailto:derek@xxxxxx
Applications Standards Conformance Testing    http://www.knosof.co.uk
Received on Tue 24 Oct 2006 - 18:59:27 BST