Re: BTF minutes, Oct 18 2004

From: Shalom.Bresticker@freescale.com
Date: Tue Oct 19 2004 - 01:24:47 PDT

  • Next message: Kathy McKinley: "SystemVerilog datatypes on nets"

    I took a quick look at some of these issues again.

    I think some of these can be done, at least with limited scope.

    For zero replication (547), I have draft text in my mind,
    just have to take the 15 minutes to type it out.
    It doesn't even require a BNF change.

    For allowing system functions in constant expressions,
    the problem is enumerating or characterizing the permitted functions.
    It gets complicated then in SV, because then they have to
    also specify the allowed functions.

    Lists in part-selects (409) is only nice-to-have, but it looks
    easy to do, both in BNF and in text.

    I did not review the entire list, though.

    I hope we can get some incremental, partial errata fixes done as well.

    Shalom

    On Mon, 18 Oct 2004, Steven Sharp wrote:

    > Minutes of BTF meeting on Oct 18, 2004, 10:40 AM PDT.
    >
    > In attendance:
    > Francoise Martinolle
    > Don Mills
    > Karen Pieper
    > Steven Sharp
    > Stuart Sutherland
    >
    > This meeting was informal, since there was insufficient notice given.
    > No official business was transacted or motions made. There was discussion
    > of how to organize the work.
    >
    > The BTF is being reconstituted under the new P1800 group, for continued
    > work on 1364 enhancements. The tentative schedule is to be done by
    > Dec 1 2004, at the same time that the P1800 first draft is scheduled.
    > P1800 did discuss whether a later date was possible for 1364 while still
    > matching the ballot date for P1800, but there was no clear resolution.
    > We will have meetings after the ETF meetings, and possibly others as
    > arranged via email. Given the limited time, most discussions will have
    > to take place via email.
    >
    > We discussed priorities on enhancement requests. Steven mentioned that
    > the ones Shalom considered highest priority were zero replication count
    > (erratum 547), allowing certain system functions in constant expressions
    > (erratum 387), and adding math system functions (erratum 390 or 580).
    > Steven suggested 354 and 297 were worth considering. It might be worthwhile
    > to consider 298 because it is already implemented in some simulators.
    > The datatypes group, which has moved to P1800, may wish us to consider
    > the "wone" single-driver net extension, as it fits into the 1364 LRM
    > better than the P1800 LRM. Stuart's `compatibility directive (erratum 287)
    > should also be considered. There was some discussion of that directive,
    > and Stuart will make a more complete proposal.
    >
    > Note: it might be possible to handle 183 and 414 as new "math" system
    > functions, as mentioned in 580.
    >
    > Erratum 350 (deprecating configs in Verilog source files) could be handled
    > as an enhancement request or as an erratum. At present, I am planning to
    > move it to the ETF as an erratum, since there is already a contradiction
    > in the LRM on whether this is legal.
    >
    > Due to disruption by the "hold music" of one of the participants, and
    > the limited business, the meeting was adjourned early.
    > 

    -- 
Shalom Bresticker                        Shalom.Bresticker @freescale.com
Design & Verification Methodology                    Tel: +972 9  9522268
Freescale Semiconductor Israel, Ltd.                 Fax: +972 9  9522890
POB 2208, Herzlia 46120, ISRAEL                     Cell: +972 50 5441478
  
[ ]Freescale Internal Use Only      [ ]Freescale Confidential Proprietary

    This archive was generated by hypermail 2.1.4 : Tue Oct 19 2004 - 01:15:02 PDT and
    sponsored by Boyd Technology, Inc.