7/22/04 Datatypes meeting minutes

From: Kathy McKinley (mckinley@cadence.com)
Date: Fri Jul 23 2004 - 11:15:17 PDT

  • Next message: Francoise Martinolle: "Fwd: RE: P1800 SystemVerilog WG F2F Meeting Agenda for August 9th"

    Attendees:

            Kurt Baty
            Shalom Bresticker
            Francoise Martinolle
            Kathy McKinley
            Dave Rich
            Stuart Sutherland

    Action Items:

            - Kathy to try to find out how we might make proposals
              for continuing datatypes extensions in new IEEE setup

            - Kathy to ask Steven Sharp to add agenda items to BTF meeting

    Summary:

    We spent the meeting discussing how we might proceed, given
    the uncertainty in the new organization under IEEE. It appears
    that there is no provision under consideration that would allow us
    to continue our datatypes extensions in the short term.

    Much of our current effort is focused on support for extended datatypes
    on nets, an area that is not covered by 1364 or SystemVerilog. This is
    a significant hole, and both Verilog and SystemVerilog users want this
    capability. Datatype orthogonality and the corresponding extended support
    that it would provide for nets was endorsed at the 1364 working group level.
    If there is no forum in which to continue this work, then there is a risk
    that different implementations will arise to fill this hole before we can
    arrive at a common standard solution.

    It is our understanding that the current goal is to have SystemVerilog
    standardized with very little change. Is there some way that we can
    propose a pipelined effort to continue the datatypes work, in a way
    that will not interfere with the short-term IEEE goals? We would like
    to discuss making a proposal for this at our next BTF meeting.

    We also discussed the need for an integrated document at some point,
    perhaps not immediately. This need is quite apparent in our current
    datatypes work, where the description of object kinds (what it means
    to be a parameter, variable, port) is in 1364, and the description of
    new datatypes (bit, structs, etc.) is in SystemVerilog. In order to
    properly discuss semantics of object kinds and the full set of datatypes,
    we need to modify the text for both parts. The extensions, even if
    they are made in two documents, need to be made by the same group.

    Integrating the 1364 and SystemVerilog documents would be a big effort,
    but it seems like it will be necessary at some point for making extensions.
    Could we propose to split the IEEE document into several volumes, some
    of which might be quite independent, and some of which would be the focus
    of serious integration effort? A separate volume would be different
    from a "dot standard" in that it would not be balloted separately.
    Perhaps the integration could be phased in over different versions
    of the standard. We would also like this document issue to be discussed
    at the next BTF meeting.

    This archive was generated by hypermail 2.1.4 : Fri Jul 23 2004 - 11:11:23 PDT and
    sponsored by Boyd Technology, Inc.