Re: [sv-ac] Re: [sv-ec]e-mail ballot Closes Wednesday February 20 2008, 11:59pm PST

Hi Gord,

I have re-written the proposal to allow only covergroup instances within 
a checker.  Take a look and see if that answers some of the most serious 
problems.  The proposal is attached.

Thanks,

Tom

Gordon Vreugdenhil wrote:
> 
> 
> Steven Sharp wrote:
>>> From: Gordon Vreugdenhil <gordonv@model.com>
>>
>>> But that is immaterial to my question.  If you have an auto declaration
>>> in the enclosing scope, can a checker instance refer to that?  No
>>> one has (yet) made any restrictions regarding the resolution rules,
>>> so I am assuming that the answer is yes.
>>
>> In defense of this part of the proposal, I think the same issue arises
>> for static variable initializers.  An initializer on a local static
>> variable could be within the scope of an automatic variable, and I don't
>> see anything in the LRM restricting it from referencing one.  However,
>> clearly it cannot be allowed to do so.  Automatic variables are created
>> by procedural execution entering their scope, and static variable
>> initializers are not executed in the context of a procedural entry into
>> the scope.
> 
> 
> I agree on both fronts.  My concerns are mostly that while I can
> reason about why something like:
>     initial begin:b
>         automatic int x = 1;
>         static int y = x;
>       ...
> should be illegal, it is harder for me to reason about things
> in the AC space since in various contexts more "synthesis"
> like behaviors seem to be expected and I haven't quite gotten to
> a point where I can predict what AC expects the behavior to be.
> 
>> If concurrent assertions and checkers in procedural code were similarly
>> static objects, independent of any procedural execution of the code, then
>> things would be simplified.  However, Mantis 1995 and 2110 break that
>> conceptual model.
> 
> This is also what I'm struggling with.  Some of the aspects
> of checkers appear to in fact be static (the text requires
> that for checker variables, etc), but it isn't quite clear
> to me yet exactly what the boundaries are and whether those
> boundaries are well formed.
> 
> Part of my difficulty lies in lack of time to get everything in
> my head; part of it is also that various relevant pieces of
> the semantics and rules appear to be distributed in a half-dozen
> Mantis items and I don't know what is all relevant.
> 
> Gord.

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