Minutes of the sv-sc sub-committee meeting, June 17, 2008

=================Attendance=============================

   1020220 Day
   7306817
   0000000 Month
   6655444
   0000000 Year
   8888888

--[------a]   Arturo Salz          - Synopsys
vv[aaa-aa-]   Abigail Morehouse    - Mentor
--[------a]   Bassam Tabbara       - Synopsys
--[------a]   Brad Pierce          - Synopsys
--[--a-aaa]   Cliff Cummings       - Sunburst Design
-v[--aaaaa]   Dave Rich            - Mentor Graphics
vv[aaaa-aa]   Dmitry Korchemny     - Intel
--[--a-aa-]   Don Mills            -
-v[--aaaaa]   Eduard Cerny         - Synopsys
tt[aaaaaaa]   Erik Seligman        - Intel (chair)
vv[aaaaaaa]   Francoise Martinolle - Cadence
vv[aaaaaaa]   Gordon Vreugdenhil   - Mentor Graphics
vv[-aaaaa-]   Jin Yang             - Intel
vv[-aaaaaa]   John Havlicek        - Freescale
--[------a]   Jonathan Bromley     - Doulas
--[-a--a-a]   Karen Pieper         - Accellera
vv[aaaaaaa]   Lisa Piper           - Cadence
vv[aaaaaaa]   Manisha Kulshrestha  - Mentor Graphics
vv[aaaaaaa]   Mark Hartoog         - Synopsys
vv[-aaaaaa]   Mehdi Mohtashemi     - Synopsys
vv[aaaa...]   Michael Burns        - Freescale
vv[-aaaaaa]   Mirek Forczek        - Aldec
vv[-aaaaaa]   Neil Korpusik        - Sun Microsystems
--[-----a-]   Ray Ryan             - Mentor
--[-----aa]   Shalom Bresticker    - Intel
vv[aaaaaaa]   Steven Sharp         - Cadence
--[----aaa]   Stu Sutherland       - Sutherland HDL
--[-----aa]   Surrendra Dudani     - Synopsys
vv[aaaaaaa]   Tom Thatcher         - Sun Microsystems (co-chair)
||
||---- Voting eligibility for current meeting
|------Voting eligibility for next meeting


====================Agenda==============================

Agenda:

1.     Review the patent policy

2.     Approve the minutes from the last meeting, available at
   http://www.eda.org/twiki/bin/view.cgi/P1800/SvScMeetingMinutes20080603

3.     Action Item Review
   * 2398: New semantics for concurrent asserts in procedural code (Erik, Gord)
     - Discussion: global const casts, inferred_const, reset handling?
   * 1728: Let proposal (Dmitry)
     - CC expert (Abi) help on VPI aspects
   * 1900: Checkers (Dmitry, Erik, Mike, ...): 6/24 to write
   * 2088: Covergroups in Checkers (Tom?): 6/24 to write
   * 2089: Final in Checkers (Tom?) : 6/24 to write
   * 2182: VPI in Checkers (owner?): 7/1 to write
   * 2370: $past in procedural code (Jin) : 7/1 to write
   * New proposal for self-determined types (Gord): 6/24 to write
   * New proposal for "@edge clk" (owner?): 6/24 to write

4.     Other technical topics

o        Opens?



====================Notes==============================

1.  Patent Policy:
    Move:  Gordon  Consider the patent policy read
    Second:  Mark
    Approved unanimously
    Ref:  http://standards.ieee.org/board/pat/pat-slideset.ppt


2.  Approve minutes.
    Move:  Gordon:  Approve the minutes from the last meeting.
    second:  Tom:
    Approved unanimously

3.  Action item Review

    * 2398: New semantics for concurrent asserts in procedural code (Erik, Gord)
      - Discussion: global const casts, inferred_const, reset handling?
    Dmitry   - Had proposed in E-mail to allow syntax to allow all variables
            to be const cast.
    Gord   - Don't really like this extension
    Tom      - Will it really save effort?
    Erik   - Could we create a "$inferred_procedural" to let the checker
            /property know where it is called.
    Erik   - Provide a parameter/argument to let checker know?
    Dmitry   - possible.
    Abby   - Seems a const assertion doesn't make any sense.
    Mike   - Still not seeing the interesting use case.
    Gord   - Doesn't seem to be too inconvenient.
    Erik   - Could use a macro to do this.

    Dmitry   - Problems with reset?
          - Arguments to checkers are sampled.
      - What about disable iff of properties
    Dmitry   - Need to make argument passing to checkers sampled
    Tom      - Value passed to checker sampled if used in concurrent
            assertion, not sampled when used in assignment.
      - Was never aware of any intent to make checker arguments
        sampled regardless of use.

    Manisha   - 16.14.5.1:  Language issue.  "Attempt" is not when clock tick
            occurs.
    AI:  Manisha will send e-mail on this.

    * 1728: Let proposal (Dmitry)
      - CC expert (Abi) help on VPI aspects

    Erik   - Abi was helping on the VPI arg
    Abi      - Actual arguments to let can be expressions.  How can we
            access formal arguments of let?  Will we have to invent
        something.  Note in proposal says that iterator returns
        vpiExpression.
    Gord   - complexity due to untyped arguments.  You don't have much
            information
    Erik   - What about sequences?  Do they have same problem?
    Abi      - Yes, Sequences have similar problems.
    Erik   - This seems to be a more general problem
    Abi      - Yes.
          - Diagrams and text contradict each other.  Don't like either
      - Several competing/overlapping proposals in play.  We could
              take VPI out of this proposal.
    Fransoise   - Not many people in CC familiar with assertions.  Would be
            better to have a joint proposal.

    Dmitry   - changes: constant value functions might be used in let.
            changed to properties
    Gord   - Have to think about this change.
    Dmitry   - I could split out this change, and create a new Mantis item.
    Erik   - Seems like a good idea.

    * 1900: Checkers (Dmitry, Erik, Mike, ...): 6/24 to write

    Dmitry   - Some outstanding issues:  free keyword, assignment semantics
          - const free vars, etc.
      - Part 2 is ready.
    Mike   - Free variables in simulation, randomization
          - Haven't contributed anything yet.
    Erik   - What about section 17.6.  It's still in checker proposal.
            A lot of text about free variable assignments
    Dmitry   - Mike's proposal will be an addition, not a re-write.
          - Pink colored text will need to be changed.
    Mike   - On track for getting proposal out on 6/24
          - Discussed causality issue with John H.
      - Idea is not to use sampled values for solving. 
        Solve in active

    Tom      - Solve in Observed or Reactive region, using current values.
            This gives you the value for the next clocking event.
    Gord   - Not sure this is completely valid.
    Mike   - Feeling less and less comfortable about this.
    Gord   - Would it make sense to solve in preponed region?
    MIke   - Doesn't this violate assumption that nothing happens in 
            preponed region?
          - Would we have to solve at every time step?  Because we don't
            know if the clocking event will occur until observed region.
    Gord   - If nothing changes, don't have to solve.  But, could end up
        solving more often than otherwise.

    AI:  Mike to send out some examples


    Other checker issues

    Dmitry   - Error in BNF:  deferred assertion does not have a label.
    Gord   - What's the purpose of the label?
    Dmitry   - Reporting for assertion violations.
    Gord   - OK

    Dmitry   - Problem with ended and triggered
          - ended and triggered are pretty much the same.
    Steven   - named events have triggered triggered
          - As long as it's a distinct type
      - 
    Dmitry   - 
    Tom      - Current text says must be used in a wait.
            Would there be a problem including triggered in a sensitivity
        list?
    steven   - If a named event .triggered appears by itself in a sensitivity
        list, the block will execute only once.  It executes once
        when triggered becomes true.  It doesn't execute when
        it becomes false.  Then when it becomes true again, nothing
        is changed, so no new event generated.
        But, if other items are in sensitivity list, it should be OK.
    Tom      - So no new event happens when triggered set back to zero at
            end of timestep?
    Mike   - It would be a contradiction, because then you would have to
            be processing events at the end of the timestep.

    Dmitry   - clock inference in sequences
          - Sequence passed to checker
      - Will it infer clock from checker context?

    * 2088: Covergroups in Checkers (Tom?): 6/24 to write
    * 2089: Final in Checkers (Tom?) : 6/24 to write

    Tom      - Dmitry will be including the bodies of these proposals in 
            the 1900 proposal.

Next meeting:  6/24

-- ErikSeligman - 19 Jun 2008