Subject: Re: DASC Scope
From: Evan Lavelle (eml@riverside-machines.com)
Date: Thu Nov 13 2003 - 02:11:26 PST
Ron Waxman wrote:
> At 04:28 PM 11/12/03, John Michael Williams wrote:
>> If several variants of SystemC or SystemVerilog
>> evolved (literally)
> They will.
>> , and differences created
>> incompatibilities,
> They will.
>
>> then IEEE should consider resolving
>> such differences by publishing a standard.
>
> Which is why an IEEE standard for a widely used "industrial strength" de
> facto standard is needed before the problem exists.
I don't think this will fix the problem, because we don't appear to have
the procedures in place to handle this. Your point is basically that an
IEEE standard is in some way independent and fixed, and is impermeable
to vendor pressure, while a vendor or industry group standard is none of
these things, and is subject to drift (correct me if I'm wrong).
However, consider that one year from now DASC will very likely contain
several hundred members, primarily composed of representatives of the
major EDA vendors. If a group within DASC wants to push through a change
which is beneficial to a particular vendor or group of vendors, then
there is a very real possibility that this will happen. We will simply
have become exactly what we are trying to avoid.
How do we protect against this possibility? I can see 3 possible routes:
1 Strengthen the procedures to reduce vendor influence. This is already
possible to some extent, but the WG chair has to apply this rule
arbitrarily, in a way which is not defined. And what if the WG chair is
an employee of the vendor who wishes to push through a change? I don't
see that this is a viable route.
2 Is this what the Steering committee and NesCom are already attempting
to do, in a rather ill-defined way? If so, this needs fixing. Their
procedures need to be transparent, we need to know who's on these groups
without resorting to Google, we need to know what their affiliations
are, and we need to be convinced that they represent a cross-section of
*users*, and that they represent the interests of *IEEE members*, not
EDA vendors. The DASC does not exist to advance the interests of EDA
vendors.
3 This is my personal favourite. The DASC needs a new advisory group.
This should be composed of about 50 - 100 members, whose purpose will be
to advise on the future direction of EDA, to arbitrate on issues such as
whether a language is suitable for IEEE standardisation, to advise on
whether an existing standard needs modification, and to ensure
transparency and impartiality. The current DASC and the SC will remain
unchanged, and the DASC will contain the hundreds of paying and voting
members required for actually creating the standards. The advisory group
should be composed of EDA experts, with demonstrable impartiality, and
experience of a range of languages. The group should be composed of
academics, and experts from within the industry. Voting needs to be
secret to ensure that industry members are not unduly influenced by
their employers.
Evan Lavelle
This archive was generated by hypermail 2b28 : Thu Nov 13 2003 - 02:13:04 PST