Minutes of ISAC meeting held via telecom on 31 August 2005

Present: Larry Soule, Chuck Swart, Ajay Verikat

Absent: Peter Ashenden, Jim Lewis, Deepak Pant

Next Meeting: Thursday October 6, 2005, 7 pm Pacific Daylight Time
              (Friday October 7, 2005, 2 am GMT)

TOPIC: Status of Web Based IR System

The system is available at:

https://bugzilla.mentor.com

Please register there so that you can login.  Currently there is only
a test project available and only one bugzilla issue. The issue, #1,
is a summary of the requirements for an on-line system, and contains
Peter's suggested mapping between VASG IR status and Bugzilla states.

Chuck will be adding more stuff to this website.

TOPIC: IR 2071 Indexed name in case expression  

The submitter correctly complains about the following wording for case
statements:

"If the expression is of a one-dimensional character array type, then
the expression must be one of the following:

...

-- An indexed name whose prefix is one of the members of this list and
   whose indexing expressions are locally static expressions"

The requirement that the indexing expressions be locally static makes
sense for slices but is overly restrictive for indexed names, since
the size information comes from the (locally static) index subtype.

ACTION: All to review. We'll discuss this at our next meeting.

TOPIC: IR 2072 allow static operations on "ranges"

At first glance, this request seems a little strange. It is clearly an
enhancement request, but some analysis should be done to determine
whether or not there are technical issues. It seems that what the
submitter wants are operations between a range and a scalar. Does he
also want operations between two ranges?  If so, what is the
interpretation when the ranges have different directions.

Also, what effect would this  extension have on an already-complicated
grammar?

ACTION: All to check for subtle problems.

TOPIC: IR 2073 Index constraints and discrete range conversions from
universal_integer

The submitter claims, correctly, that the code:

    for i in 0 to 2 ** N -1 loop...

is technically illegal because the implicit conversion from universal
integer to INTEGER applies only to numeric literals or attributes (or,
presumable division of two values of a give physical type). This rule
makes

       for i in 1 to 10 loop ...

legal, but

       for i in -1 to 10 loop

illegal. This problem was recognized in Ada, and was solved by adding
new types (root_integer and root_real) and by changing the overload
resolution rules to favor these new types in ambiguous cases.

ACTION: Chuck to analyze


Categorization of old IRs:

It was agreed that (eventually) we will move open VHDL93 IRs into the
VHDL2002 database. Chuck will maintain a list showing the status of
VHDL93 IRs.

Status

1002    Open     Ambiguity as to when parameter subtype indications are elaborated.
1003    Open     Non-commutative and non-associative resolution functions are a source of non-determinism
1004    Enhancement-Analysis needed Are deferred constants still deferred after their full declaration?
1008    Resolved Function NOW is undefined during static elaboration.
1009    Open     Unclear context of evaluation for formal part of block map aspects.
1013    Open     Reference to a generate loop parameter in a configuration is unclear
1014    Open     The LENGTH attribute is ill-defined.
1017    Open     Why must a name in a wait stmt. sens. list be static?
1019    ??Compare with 2028 The signal updating steps of the simulation cycle is incorrect.
1024    Open     VPI Issue 20 -- Primary units with same name