ISAC: Minutes from meeting on 07 November 2007

These minutes are also available at eda.org.
Note that the minutes contain several comments that have been
expanded from the activities of the meeting.

Chuck Swart

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.


Minutes of ISAC meeting held via telecom on 07 November 2007

Present: Peter Ashenden, Larry Soule, Chuck Swart, Ajay Verikat

Absent: Jim Lewis, Lance Thompson

Next Meeting Wednesday, January 09, 8 pm Pacific Standard Time
             (Thursday, January 10, 4 am GMT)

TOPIC: IR 2123 Process resumption and callbacks
       IR 2124 Ordering of process execution and callbacks

Chuck reported that he had not received any response from VHPI chair.
The resolution of these IRs depends on the needs of VHPI.  Ajay said
that he would contact VHPI chair about these issues.

ACTION: Ajay to contact VHPI chair.
[Added: VHPI chair has contacted Chuck and Chuck will forward/resend
appropriate issues to her.]

TOPIC IR 2119 Can't declare a protected type and object of that type
in a single package.

The ISAC recommended that the LRM not be changed to allow this. The
VASG vote followed the ISAC recommendation, but there were two
negative ballots.

Jim Lewis stated that he thought "that having a protected type/class
in a package was a suggested way to do a static class member." Peter
responded that this impression was incorrect.

John Ries stated that the use case described was "common." Peter asked
if that meant that some tools supported this feature. Chuck replied
that [answer expanded from response given at the meeting] at least one
tool did accept this. When the tool producer realized that this was
non-conforming it added the capability as an extension controlled by
a switch.

The ISAC decided to submit this IR as a requirement, and to supply
preliminary analysis of some of the technical issues.

One solution is to allow protected type bodies in package headers.
This would work, but violates the principle that package headers
present only the interface, and that package bodies supply the
implementation.

A second solution (taken from comments of John Ries) is to provide
some explicit syntax for identifying this case.

So, for example, one could say something like

package p1 is
        type t1 is protected
             function count return integer;
             ...
        end protected;

        shared variable v1: t1 := OPEN;--new syntax

        constant c1: integer := v1.count;--still illegal

end package p1;

A third possibility (also from John Ries) is to treat a shared
variable in a package something like deferred constants--the variable
is not elaborated until after the protected type's body is
elaborated. Use of the shared variable would have the same
restrictions as if it were a deferred constant.

Peter commented that we might do something similar to the revision of
elaboration rules for driver identification given by hierarchical
references.

ACTION: Peter to incorporate a brief description of the general
issues, all to review, then the issue will be forwarded to the
requirements committee.

ITEM: Migration of pre-2002 IRs to Bugzilla

Peter observed that there will be a period of inaction while the
VHDL-200X LRM is transferred from Accellera to IEEE and the IEEE
approval mechanism is established. The ISAC could spend this time
working on legacy issues. It was decided that the ISAC would take one
month off, unless new issues were raised. During this month, Chuck and
Peter will work on transferring old IRs to Bugzilla, and perhaps,
reorganizing the Bugzilla entries as proposed by Peter several months
ago.It was also noted that we plan on implementing the IR analysis as
an attachment to provide flexibility and to prevent long, redundant,
obsolete analyses from cluttering the main text.

ACTION: Peter and Chuck to begin this transition.