Subject: Re: [sv-ac] First match example and match of past.
From: Adam Krolnik (krolnik@lsil.com)
Date: Fri Feb 14 2003 - 11:58:18 PST
Hi Cindy;
Thanks for taking the time to write out these good assertions.
>the long answer is that in many cases, if you
>have multiple matches of the antecedent, it is a bug that you will catch
>with another assertion. therefore, you are expecting only one match, and
>it is simpler visually and conceptually to leave out the first_match.
Here is where I disagree.
The example I proposed
assert @(posedge clk) (a; [1:3] b => {1; c; d}); (1)
Was designed as we reasoned about the idea of first_match. What we came up
with was that a user must looking for 'a' instead of just 'b'
because there could be more 'b's; without a preceeding 'a'.
Thus the example would need the first_match() prefix.
If you left off first_match() for a case such as I described, then you would
receive false failures.
If you put on first_match, then the assertion would not have that failure.
But it also seems that for your example, first_match() would not hurt to have
it in the assertion.
Thank you again.
Adam Krolnik
Verification Mgr.
LSI Logic Corp.
Plano TX. 75074
This archive was generated by hypermail 2b28 : Fri Feb 14 2003 - 11:59:13 PST