Minutes of BTF meeting on June 28, 2004, 10:45 AM PDT.

From: Alec Stanculescu (alec@fintronic.com)
Date: Mon Jun 28 2004 - 14:35:13 PDT

  • Next message: Stefen Boyd: "Fwd: BOUNCE btf@wa.boyd.com: Non-member submission from ["Chris Spear" <Chris.Spear@synopsys.com>]"

    Kurt moves to accept the minutes sent out on June 14th.
    Steve Dovich seconded.

    Present at this conference call are:

            Kurt Baty
            Geoffrey Coram
            Tom Fitzpatrick
            Steven Dovich
            Francoise Martinolle
            Don Mills
            Karen Pieper
            Brad Pierce
            Dave Rich
            Steven Sharp
            Alec Stanculescu
            Stuart Sutherland
            Shalom Bresticker

    Patent Policy was read at the 8:30am PST meeting. THe people in
    attendance agreed that the reading still applies.

    Next meeting will be organized by Karen Pieper on July 12.
    It will be Chaired by Alec Stanculescu.

    Steve Dovich reported on Encryption.
    Progress was made on boundary of encryptions.

    Steve Dovich also reported that his group located a registry of
    encryption and encoding. Get the Verilog std to
    point to std algorithms provided by "Internet Engineering Std
    process". There is a WEB site but was not reported yet as being
    premature at this time.

    The encryption group is still working under 1364.

    Alec reported on Prioritization Group.

    Issues 416 457, 421, 422, 427 were discussed within the group.

    Steve discussed issue from Data type group. Should reg be an object
    kind or a type equivalent to logic. This would require another keyword var
    be added.

    Steve Sharp said that SystemVerilog uses reg as a type
    declaration. However, as objects of complex data type can be also
    formal ports there may be problems.

    Kurt: reg should become type.
    Stu: if reg becomes type then PLI will be affected.
    Stu: Reg has special properties: can be of any size.
    How do you get the properties of the object? Left Range, Right Range?
    Kurt: implementation range
    Francoise: PLI does not provide you access to range of real because it is not
    user definable.
    Kurt: extend vpi to get ranges of non-user defined types
    Francoise: vpi_reg has the same enumeration as vpi_var. SystemVerilog
    addressed the backward compatibility with V2K.
    Returns vpi_logic_var.
    Francoise/Ric,Stu/Steve: agreed that there may be big confusion
    between kind and type in PLI applications.
    Francoise, Dave Rich: it may be workable.
    Stu: it will be difficult.

    Reg is already perceived as a kind in vpi. With either decision we
    will break one rule or the other in vpi.

    Francoise, Alec: Old iterators will work on existing applications and
    new iterators will work on new designs.

    Francoise: new iterators will work also on old applications.

    Steve Sharp: One solution would be to deprecate reg in any typedefs.
    Stu: use logic in structure declarations and not reg.

    Steve Sharp: variable should be viewed as the union of reg, integer,
    parameter, etc.

    Steve Sharp further invited people to make some final comments.

    Brad Pierce: I have not seen reg used that much. People are using logic.
    Alec: The distinction between kind and type must be clearly
    made. This distinction must be supported by new iterators in vpi with
    the old iterators continuing to support the existing applications.
    I agree with Kurt that non-user declared types (such as real and
    integer) shall still have their left and right ranges reported by the
    vpi and I agree with Dave Rich who pointed out that the new reg data
    type must be unified with integer.
    Stu: There is a need for a full PLI specification regarding changes
    that need to be made to prove backward compatibility with existing
    applications. I like var as a type. I hate wire reg!
    Shalom expressed concern about the fact that forced backward
    compatibility with SV may lead to make registers type rather than kind
    which he as a user would rather like to see. A reg is more
    natural as a kind but it may not pass because this does not match SV.

    This issue will be discussed by the PLI task force this afternoon.

    Steve Sharp: I will not attend next meeting on July 12 at 8:30am because I
    will be on vacation.

    This archive was generated by hypermail 2.1.4 : Mon Jun 28 2004 - 14:34:36 PDT and
    sponsored by Boyd Technology, Inc.