Minutes SV-AC 01/12/04
Written by: Surrendra Dudani

SV-AC v3.1a Meetings:
Next meeting: 01/19/04

code for the conference call:

        Domestic: 888-635-9997
        International: 763-315-6815
        Participant: 274372#

Attendance Record
	Legend:
		x = attended
		- = missed
		r = represented
		. = not yet a member
		v = valid voter (3 out of last 4 or 75% overall)
		n = not valid voter

v[-x-xxx] Faisal Haque (Cisco, Chairman)
v[-xxxxx] Arif Samad (Synopsys - Co-chair)	
v[xxxxxx] Roy Armoni (Intel)
v[xxxxxx] Surrendra Dudani (Synopsys)
n[------] Cindy Eisner (IBM)
v[xxxrxx] John Havlicek (Motorola) - Kurt Schultz
n[------] Richard Ho (0-in)
v[x----x] Adam Krolnik (LSI)
v[xxxxxx] Joseph Lu (Sun)
n[------] Rishiyur Nikhil (Blue Spec)
v[x--x-x] Koushik Roy (Cadence)
n[------] Andrew Seawright (0-in)
v[-xxx-x] Bassam Tabbara (Novas)
n[------] Tej Singh (Mentor)
n[------] Connie O'dell (Consultant)
v[-----x] Hillel Miller (Motorola)
==|||||+-------------------------- 11/17/03
==||||+--------------------------- 11/24/03
==|||+---------------------------- 12/01/03
==||+----------------------------- 12/08/03
==|+------------------------------ 12/15/03
==+------------------------------- 01/12/04
-----------

1)Summary

- Proposals 8,10 and 19 were discussed.
- The proposals would be modified in accordance with the discussion, and submitted by the end of today.
- Proposed and accepted to have an email vote with the deadline of 1/18/04(end of the day)
- If necessary, voice vote will be taken in the next meeting on 1/19/04.

2) Minutes
* Proposal 19 was presented by John.
Adam: Can task be called multiple times?
John: Any time there is a sequence match, the task would get called.
Adam: If it was desired to have only one call, then you would use first_match operator on the sequence
John: For detecting failures, you would need to construct the failure witness to invoke the task.But, for coverage, the task gets executed naturally for every match
Adam: What is the BNF cahnge?
John: Variable assignment needs to change to sequence_match_item which includes variable assignment, task call, system task call and function call.
Adam: Suggested to explicitly state that the sampled value gets passed in via call by value parameter. Should we allow ref. or only const ref?
John: It is best to leave the flexibility to allow updating information based on the results, such as coverage results.
Surrendra: Agreed.
Adam: What is the order between task execution and the corresponding action block execution?
John: It is undefined, just like the order between muliple executions of a task due to multiple matches.
John: Proposal will be revised to expcitly state the argument types and the execution semantics of the tasks.

* Proposal 8 was presented by Adam.
Kaushik: what is the meaning of @clk in a function, since functions are timeless.
Adam: Assertions are simply extracted with the inferred enabling condition and clock, as if it was written outside the function.
Kaushik: Would sequential expression be allowed?
Adam: Since the assertion is extracted out, it doesn't matter as it doesn't execute within the function.
Kaushik: It is different than a function, as a function is just timeless. May be need more discussion.
Surrendra: What would it mean to have an assertion in a function, if the function is used on the rhs of a continuous assignment.
Adam: I guess it would create enabling condition based on the signals on the rhs of the assignment.

* Proposal 10 was presented by Adam.
Kaushik: how would the results be different, if modport is instantiated multiple times.
Adam: You would be able to identify the source of the error message when the assertion fails.
John: Can we not just instantiate the assertion along with the import statement?
Adam: You could do that, but there is no existing syntax to allow it. Currently, you can only import identifier syntactically.
Surrendra: Does the assertion gets resolved with respect to the signals used in the interface where it is declared?
Adam: Yes.
Surrendra: So the results would be identical wherever the assertion gets imported, excepr for the name of the assertion.
Surrendra: Suggestion is to modify the semantics so that the execution of the assertion never takes place if the assertion is imported by any modport.
Adam: Will modify the proposal and submit it by the end of today.

     

3) Next meeting

Next meeting will be held on 01/19/04 at 9AM PST.