Re: profiles/modules/levels

I think I now see the motivation behind much of this discussion.  Andrew, 
you seem to perceive that adding new DOVs (outside of the ones SpecGL 
requires) to specs will be judged as extensions and, thus, thought of, in 
some bad light.

I think we have to look at "extensions" to SpecGL as a different kind of 
animal that extensions to target specs.  Extensions to target specs add new 
functionality that obviously harm interoperability.

Let's remember that SpecGL puts requirements on spec developers who then 
put requirements on (real) implementers.  The requirements we put on spec 
developers are only the ones that we thought of during the time of SpecGL 
development.  I think it would be very naive to think that we've thought of 
everything.  I believe almost all new specs will include things we haven't 
thought of, either because they were so trivial or because we weren't smart 
enough.

So my conclusion is twofold: 1) I think that just about every spec 
developed will have "extensions" to our list of requirements; 2) I don't 
think these extensions are necessarily harmful or a danger to 
interoperability if they are well thought out.

Mark



At 10:57 PM 4/29/2003 +0100, Andrew Thackrah wrote:



>Hi Mark,
>One concern I have about this explanation is that a spec author
>may not regard a DoV as being an extension. If a new-DoV [to use
>  Lofton's word] is engineered into the spec from the early days then
>I think that the authors would be confused if we told them that it
>counted as an extension.
>Even though I'm closer to SpecGL than many spec author's may ever be
>I still find this argument a subtle one, even though it seems
>technically good.
>
>-Andrew
>
>On Tue, 29 Apr 2003, Mark Skall wrote:
>
> >
> > At 02:32 PM 4/29/2003 +0100, Andrew Thackrah wrote:
> > >
> > >But what if the spec author choses another type of architecture that we
> > >have not thought of? My argument is about this case. I want to make sure
> > >that our
> > >checkpoints address this possibility. At the moment we have specific
> > >checkpoints for p, m and l. But if someone wants to use a new type of Dov
> > >called 'personality' or whatever then SpecGL is silent. So I am arguing
> > >that when we roll the checkpoints into a single GL, we should keep the
> > >specific checkpoints for the important concepts of p/m/l but ensure that
> > >we have general checkpoints too.
> >
> >
> > An implementation can be conformant to SpecGL and include extensions.  If
> > someone wants to use another form of DOV, they may.  That is a (conformant)
> > extension.  Since there is no way of pre-determining what kind of DOV (or
> > any other type of extension) someone may choose to include there is nothing
> > to gain by having general checkpoints.  The general checkpoint would be
> > redundant.
> >
> > Mark

****************************************************************
Mark Skall
Chief, Software Diagnostics and Conformance Testing Division
Information Technology Laboratory
National Institute of Standards and Technology (NIST)
100 Bureau Drive, Stop 8970
Gaithersburg, MD 20899-8970

Voice: 301-975-3262
Fax:   301-590-9174
Email: skall@nist.gov
****************************************************************

Received on Tuesday, 29 April 2003 18:11:39 UTC