Subject: Re: [sv-ac] R29a/b Optional or mandatory name for assertions/properties/assumptions.
From: Rajeev Ranjan (rajeev@realintent.com)
Date: Thu Sep 19 2002 - 18:34:55 PDT
Simon --
Your analogy is not completely accurate. You have a name and so does your
car. The random serial number you are referring to (I am assuming VIN) is
canonical and has been assigned by the manufacturing company (with some
global convention which keeps the naming space of different manufacturing
companies disjoint). If you take the car to different repair shops and if
they need to refer to VIN, they all will be using the same number -- as
opposed to every repair shop creating their own "random" number.
And regardless of how many additional cars you buy/rent/least, the VIN of
each of the car remains unchanged. Let me illustrate why I make this point.
I would argue that if I write an assertion and a name is automatically
assigned to it then :-
- The name of the assertion should not change if I change my mind and move
it around the code for readability purposes. This can be freely done if the
assertion is declarative -- or I could choose to move the whole procedural
block.
- The name of the assertion should not change if I add new pieces of code
before or after it -- i.e., additing new processes, adding new conditionals
and so on.
I hope I don't have to argue as to why it is desired to have the name of the
assertion to be constant.
I would contend that ANY auto generation scheme could be broken (i.e. would
not meet the two requirements stated above), albeit with small probability.
-rajeev
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Rajeev K. Ranjan Tel: (408) 982-5418
Director, R&D Fax: (408) 982-5443
Real Intent
3910 Freedom Circle, Suite 102A rajeev@realintent.com
Santa Clara, CA 95054 http://www.realintent.com
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
----- Original Message -----
From: "Simon Davidmann" <simond@co-design.com>
To: "Adam Krolnik" <krolnik@lsil.com>
Cc: <sv-ac@server.eda.org>
Sent: Thursday, September 19, 2002 5:08 PM
Subject: Re: [sv-ac] R29a/b Optional or mandatory name for
assertions/properties/assumptions.
> Adam
>
> Sorry to be slow to catch up - but surely this generation of an auto-name
> for me is wrong - you do not need to generate a name for me - that just
> adds confusion.
>
> My parents gave me a name as they knew you would want to use it - thus I
> have been named.
>
> However my car is not named - it just has a random serial number that I
can
> use if I ever need to.
>
> there is a big difference.
>
> Simon
>
> At 11:50 AM 9/18/2002, you wrote:
>
>
> >Hi ovi1751;
> >
> >[BTW, that's my autogenerated name for you...]
> >
> > >I agree with Simon. Why force someone to name an assertion
> > >if he or she is happy using the default generated names?
> >
> >Gail has presented many reasons for why a 'default generated'
> >name can create problems.
> >
> >I've been thinking about proposing a requirement:
> >
> >'Simulators should create a signal from an assertion name
> >and set it to 1/0 based on the pass/fail of the assertion
> >so that a waveform viewer can display this pass/fail
> >status of an assertion for easy identification of the time
> >when an assertion fails.'
> >
> >It is highly useful to be able to see (in waveforms)
> >
> >If the usage of assertions by external tools necessitates
> >a name (for consistency of results.) Then based on experience
> >let's require one. The cost is not very great and this can
> >prevent problems.
> >
> >
> > Adam Krolnik
> > Verification Mgr.
> > LSI Logic Corp.
> > Plano TX.
>
>
This archive was generated by hypermail 2b28 : Thu Sep 19 2002 - 18:36:52 PDT