I've read Greg's "Proposed Language Requirements" document
dated July 1999.
IEEE has published a standard (830) as to how a Requirements
document should be written. Although a language
requirement specification is not quite the same as
a software requirement specification, I'd prefer
to have our requirements in something
resembling the Std 830 format, rather than in free form.
Many of Greg's suggestions don't provide enough information
to verify whether or when the requirement has been fulfilled.
Also, I don't understand the final appendix, which
lists "priorities": Would these be priorities for
completion of design specifications by the OO
group? I assume it would not be the task of a standards
group actually to implement a conforming compiler?
I'd like to suggest that agreement on requirements will
be easier, I think, if we avoid priorities and just
decide on the constructs to be required in OOVHDL.
Remember, less important language constructs always
might be added later.
I think Greg's paper could form a good basis of a
language requirements document if it merely
listed by name the constructs required.
--
John
jwill@pacbell.net
John Michael Williams
Peter J. Ashenden wrote:
>
> Dear colleagues,
>
> The review process for SUAVE and Objective VHDL required publication of
> the Requirements document. You can find Draft Version 2.0, prepared by
> Greg Peterson, at
>
> http://www.cs.adelaide.edu.au/users/petera/suave/OOReqVer2.pdf
>
> Cheers,
>
> PA
>...