RE: [sv-ac] call to vote on 1549

Hi Lisa,

I think that assigning the max value is a bad idea in SVA context. I
think that the simplest solution for now is to forbid assigning $ to
typed arguments (and change the example).


Doron


-----Original Message-----
From: Lisa Piper [mailto:piper@cadence.com] 
Sent: Friday, August 17, 2007 6:32 AM
To: Bustan, Doron; Bresticker, Shalom; Yaniv.Fais@freescale.com
Cc: sv-ac@eda-stds.org
Subject: RE: [sv-ac] call to vote on 1549

Hi Doron,

I agree with Yaniv that it is a problem. We can look at this two ways:

1) $ means end of simulation as defined in section 16
2) in other sections, sometimes $ is defined to mean the max value of
that type (11.4.14 for example)

I think if $ were defined as in #2, then assigning $ to shortint would
mean the maximum value of a shortint. This would be consistent with
casting.
But I don't think it would be intuitive to the user.  It changes the
meaning of $. I also could not find any examples of initializing an
integer to $. The options are:

a) error
b) change the rules just for $ and say that $ is never cast. Also
confirm that assigning $ to shortint would mean that it is assigned the
maximum value. 

The new write-up says to error. That could be relaxed later if desired,
but we would need to work with BC to ensure it is legal. I prefer to be
very conservative at this point so it can be approved in time for the
2008 release.

lisa



-----Original Message-----
From: owner-sv-ac@eda.org [mailto:owner-sv-ac@eda.org] On Behalf Of
Bustan, Doron
Sent: Wednesday, August 15, 2007 2:02 AM
To: Bresticker, Shalom; Yaniv.Fais@freescale.com
Cc: sv-ac@eda-stds.org
Subject: RE: [sv-ac] call to vote on 1549

I hate to say it, but it looks like we have too much friendly amendments
and we need to postponed the vote. 

As for Yaniv's question, at 16.7 (end of p315) it says

A sequence is declared with optional formal arguments. When a sequence
is instantiated, actual arguments can be passed to the sequence. The
sequence gets expanded with the actual arguments by replacing the formal
arguments with the actual arguments. Semantic checks are performed to
ensure that the expanded sequence with the actual arguments is legal.
An actual argument can replace any of the following:
- Identifier
- Expression
- Event control expression
- Upper delay range or repetition range if the actual argument is $

I do not see any limitation on the type. I think that once we have the
const 
qualifier, we could limit the assignment of $  to const arguments, but
until then we are fine.

Doron

-----Original Message-----
From: owner-sv-ac@server.eda.org [mailto:owner-sv-ac@server.eda.org] On
Behalf Of Bresticker, Shalom
Sent: Tuesday, August 14, 2007 8:16 PM
To: Yaniv.Fais@freescale.com
Cc: sv-ac@server.eda-stds.org
Subject: RE: [sv-ac] call to vote on 1549

Yaniv's question is good.
I am not sure this is legal.
See Manti 966 and 1350.

Shalom
 

> I also like the usage of "$" as an actual argument in the 
> example on page 4 but this usage is allowed in a kind of 
> implicit way since in the LRM clause "6.20.2.1 $ as a 
> parameter value" allows $ to be assigned to a SV "parameter" 
> or untyped sequence/property argument, for example "parameter 
> P = $;" is legal, I don't think "shortint i = $;" is a legal 
> syntax, by this example it means $ can be passed to a 
> shortint type formal ?

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

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

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