Subject: Re: [sv-ac] Re: proposal for parameters on properties.
From: Adam Krolnik (krolnik@lsil.com)
Date: Thu Sep 18 2003 - 11:17:54 PDT
Hi Joseph;
Excellent!!
>But, one more thing, i am not clear about if the range of local variables for
> properties needs to be compile-time constants. I believe that is why
> parameters are for in SV.
The range of a variable must be a compile time constant. The LRM for property
instantiation speaks in terms that makes it sound like the property body is
inlined into the code. Thus the result looks like a compile time constant,
or is an error.
>In addition, the repetition expression requires either a constant or a range.
Same for this - there is nothing in the LRM that prevents the repetition parameters
from being a property parameter. Only that it must be a constant.
>If we are using existing SVA 3.1 approach, I can easily make a mistake by hooking up
>a signal to an actual parameter for a range identifier.
If you make a mistake, you will get an error message:
Error = temporal range must be a constant.
Contrast this to reporting the value of a variable in a $error() call instead
of referencing something to get you the sampled value. It will run and report
a bad error message, making you search in the wrong place or otherwise waste time.
What do you think about the need for passing properties through the ports of a module
or some other package to allow properties and satellite code to live together?
Adam
This archive was generated by hypermail 2b28 : Thu Sep 18 2003 - 11:19:33 PDT