ISAC: Re: IR2123 Process resumption and callbacks

From: Chuck Swart - MTI <cswart_at_.....>
Date: Thu Mar 27 2008 - 15:40:12 PDT
Attached is the response from John Shields. He obviously takes strong 
exception to the proposed comments.
I think that we need to resolve this for the next language version, if 
at all possible, so we need to meet to
discuss this. Perhaps Peter could chair the discussion? Also, it might 
be beneficial if John and/or Francoise or other
VHPI members could attend to help us reach consensus.

One small correction to John's response. Although Ajay did indicate that 
he was unhappy with the VHPI decision, the
note was written by me and any failure to be objective lies with me.

When can we meet? Let's start with Peter's availability and work from there.

I would ask John Shields to provide a use case in which the "double 
indeterminacy" is not an issue, so that we can
resolve the apparent misunderstanding.

Chuck Swart




-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.


attached mail follows:


Hi Chuck,

I am sorry but I find this flawed.  first, your comment in the first 
paragraph may be a representation of a perceived problem, but also 
speaks to a profound misunderstanding of the intent of the callback. I 
know it came from Ajay and I think Francoise should have that 
information.  She may wish to discuss it with him. For you not to state 
a balanced view calls to question the ability of the ISAC to evaluate 
the specification in this area.  Your judgment as the "double 
indeterminacy making the callback potentially less useful"  is 
particularly  unbalanced.

I also find the analysis flawed. It misrepresents the VHPI position and 
hides rationale in its summary, and shows some misunderstanding of, or 
ignoring, the purpose of the callback.  I can accept the 
misunderstanding.  It is the LRM strategy to approach this problem by 
specifying VHDL resumption in this way, and to be silent about 
non-determinism. It is an artificial point in the simulation cycle that 
indeed has healthy non-determinism with no bad behavior inherent in 
exploiting it. Too bad non-determinism is a dirty word. Fine, you said 
it a different way.  But then it is incumbent upon the author not to 
hide the real rationale that VHPI had. It is a disservice to any who 
would wish to review and understand this in proper perspective.  That 
part disturbs me. I presume that, in part, is why some ISAC reviewers 
see this in a bad light. IMHO, your comments only make it worse in some 
noble attempt to appease someone. 

The LRM text revision itself is well done, given the considerations.  
The note in 8.1 could have reflected that VHPI callbacks match the 
optimized behavior (which is the intent and has value), but instead it 
is just cast as fact and left to perceive whether that is a flaw. It is 
OK, but not great. If you want to appreciate the change, read the isac 
analysis and your comments, accept them as expert opinion, and then you 
almost have to view this as flawed decision.  If you believed that, you 
should have pushed it back. 

I trust you will convey to the ISAC my disappointment in the way this 
has been done.

Regards, John


Chuck Swart - MTI wrote:
> The attached comment will be added to the analysis of this IR. Please 
> review this ASAP and
> let me know of any changes you wish to make. I am including the most 
> recent version of the IR also.
>
> Chuck Swart
>
Received on Thu Mar 27 15:40:49 2008

This archive was generated by hypermail 2.1.8 : Thu Mar 27 2008 - 15:40:52 PDT