BTF eeting minutes April 17, 2004

From: Francoise Martinolle (fm@cadence.com)
Date: Mon Apr 19 2004 - 12:01:02 PDT

  • Next message: Steven J. Dovich: "Verilog IP Encryption sub-group meeting (Wed, Apr 21)"

    Minutes of BTF meeting on April 19, 2004, 10:45 AM PDT.

    In attendance:
    Kurt Baty
    Cliff Cummings
    Ennis Hawk
    Atsushi Kasuya
    Michael McNamara
    Francoise Martinolle
    Karen Pieper
    Brad Pierce
    Dave Rich
    Steven Sharp
    Alec Stanculescu

    Steve made everyone aware of the IEEE patent policy.

    Minutes of April 5th: Francoise moves, Kurt seconds, Mac abstains, passes

    Review of action items:
    Kathy not here, we did not discuss

    Steve sharp action item regarding 538 was done.

    Alec has an action item to start reviewing the Verilog enhancements in the
    database:
      April 26 is the first scheduled meeting for reviewing and prioritize the
    enhancements.

    Alec wants to discuss 466 (separate compilation).
    Steve Sharp mentioned that no command line arguments are in the standard,
    there are all except +args implementation defined.
    Alec wants to have a list of restrictions documented in the standard for
    separation compilation for Verilog 2001.
    Steve thinks it is more an implementation issue rather than a standard issue.
    Alec brings up the example of C code compiled with a compiler it may not
    linked all the times with another compiled object from another compiler.
    Steve Dovich noted that ABI interoperability between compilers should allow
    separate compilation.

    Steve Sharp: nothing prevents vendors to do separate compilation.
    Alec: Restrictions are minimal restrictions. Effort by all vendors to have
    separate compilation.
    Dave: the restrictions are dependent on the implementation architecture.
    Alec: no

    Alec: separate compilation could be done without any restrictions
    In Verilog we need to determine the boundaries between the separate
    compilation units
    and define how to resolve external references.
    Usaful for IP delivery.

    Alec: code compiled separately by a simulator, some ip provider could
    compile his code and deliver his code if there rules were stated.

    Alec: if no restrictions are left
              then you can do separate compilation and that should be stated in
    the standard.

    Basically the discussion went around the question: Does the standard need
    to say that separate compilation must be supported?
    Alec wants to add a list of restrictions on the Verilog language usage
    which can allow separate compilation.

    Steve: if there are some language changes required to support separate
    compilation this should be mentioned in the language standard.

    Alec: SV has restrictions which match the proposed restrictions.
    Dave: Disagree, you can have XMRs between compilation units.
               In the separate compilation unit proposal in SV, we resolved
    named collision issues
                by creating a new name space: the compilation unit name space.

    Alec: at the very least we need to resolve the name conflicts.

    Steve: In Verilog 2001, module names are global
    Francoise: what about the directives?
    Alec: one solution would be be to have a .h to include all the compiler
    directive inherited.
    Steve: use the 'reset_all to turn off the directives, there is no way to
    turn off all the macros definitions right now so we may want to add
    possibility to remove their definitions

    Steve Sharp asked to stop this discussion for now as little time was left
    in the conference call to get other task force reports.

    Steve Dovich: this wednesday initial meeting for the IP protection, will
    determine if the proposal is complete and clear and if there is anything to
    do before passing it to the VSG.

    Steve Shap read the data type group progress report from Kathy:
    The datatypes subgroup continues to make slow but steady progress.
    We are sorting through some issues with primitive two-state and
    four-state logic types, and we will be preparing write-ups on some
    alternatives next week. When satisfied with our write-ups, we will
    present these issues for discussion and consideration at the BTF level,
    hopefully within the next month.
    We are also working on a prioritized list of datatypes issues that
    we plan to consider. This will help us to organize and get more work
    going in parallel.

    Kurts moves to ajourn

    Next meeting is May 3rd.

    This archive was generated by hypermail 2.1.4 : Mon Apr 19 2004 - 11:42:57 PDT and
    sponsored by Boyd Technology, Inc.