Subject: FW: [sv-ac] Re: Straw Poll on issues
From: Bassam Tabbara (bassam@novas.com)
Date: Sun Nov 16 2003 - 14:03:26 PST
Formatting seems weird, trying to send gain with some line breaks ...
Hi All,
My votes (used same issues list as Adam sent out, although some proposal
headings need to be more descriptive). I put some comments after each
rejected item.
[Y] 1 additional property directive: assume
[N] 2 immediate assume
No need for this, does not add much value, may be even adds confusion.
[Y] 3 assume synchronization
This is the "constraint" proposal right ? ** Needs
discussion/refinement.
[N] 4 local variable extensions
I would rather we address this as an issue of argument passing. The
relaxation is needed but until we have a better means of doing this,
would rather keep the restriction strict (as it is 3.1).
[N] 5 clock_var_assign
Would rather have a uniform solution, that "opens up" interaction (i.e.
sampling of assertion results) throughout the language testbench
portion, not just in clocking domains. Related to item: 17 and 18. Merge
all together and have a uniform and coherent approach.
[Y] 6 clock variable access (this is the $sampled).
[N] 7 parameter in properties
I think may be we need to review the section and make sure it is clear
you can pass expressions. But I feel the function is already there.
[N] 8 sequence passing
I agree with the spirit. But I dislike the suggested mechanism ("import"
in arguments). We should move this to EC. Related to 16.
[Y] 9 pass $ (infinite) through property arguments.
[N] 10 enhance implication
Syntactic sugar that will only serve to confuse user, would rather keep
it simple.
[Y] 11 recursive properties
[N] 12 boolean property connectives
Would rather leave composition only at sequence level
[Y] 13 assertions in functions
[Y] 14 gated clock support
[N] 16 modports importing assertions
Move to EC related to 8.
[N] 17 event created from sequence/property for reactive functionality.
Related to 5 and 18. Coherent framework needed.
[N] 18 extend wait to work on sequence/property
Related to 5 and 17. Coherent framework needed.
[N] 19 embed assertions in structures.
Move to BC/EC.
[Y] ?? Proposal for action blocks using sampled variable values.
This one is same as 6, right ?
[Y] ?? Proposal for error message having access to local variables.
Here's John's (the ones not covered in items above)
[N] generalized implication
Keep it simple ..
[N] property instances
Keep at sequence level
[N] property recursion
I like the idea but:
A) would want to keep it at sequence level
B) recursion seems to be "tail recursion" ... Recursion is a tool
inefficient way to do this, seems like we can do this better by:
- looping construct
- some operator enhancement at sequence level (until).
[N] property negation
Keep at sequence level (same as 12 above)
-Bassam.
This archive was generated by hypermail 2b28 : Sun Nov 16 2003 - 14:04:09 PST