ISAC: Minutes of iSAC meeting held on 20 April 2006

Attached are the minutes, which are also available
at the database.

Chuck Swart



`Minutes of ISAC meeting held via telecom on 20 April 2006

Present: Peter Ashenden, Chuck Swart, Ajay Verikat

Absent: Jim Lewis, Larry Soule, Lance Thompson

Next Meeting: Thursday May 04, 2006, 7 pm Pacific Daylight Time
              (Friday May 05, 2006, 2 am GMT)

TOPIC: IR 2091 Translation between std_logic_vector based types and
std_ulogic_vector

Peter supplied the analysis, which concluded that all the examples
were legal VHDL.

ACTION: All to vote, Chuck to send IR to Jerry K for review.

TOPIC: IR 2038 Minor semantic errors

Ajay has reviewed about half of the items. He raised questions about
some of the items:

2a and 2b, which deal with determining the index range of array
objects.  Both of these issues have been superseded by LCS-2006-113.

3. The submitter pointed out that some LRM text refers to the "name of
a variable designated by <an> access value..." which contradicts the
fact that access variables don't have names.

A careful examination revealed that the phrase came from Ada 83 as a
note. It was probably inserted because this was an apparently rare
case in which the type of the expression was needed to determine the
type of the target in a variable assignment. VHDL uses other
mechanisms, such as aggregates as targets, which also require
determination of the type of the expression to determine the target
type, so this case is not particularly noteworthy.

The recommendation to Ajay is to make this area a note, and because it
is informative, and not to bother to correct the slight inaccuracy in
the wording.

ACTION: Ajay to continue analysis. If he believes that he cannot finish
the analysis by the next meeting he will contact Chuck, who will
acquire "volunteers" to take over some of the issues.

TOPIC: IR 2054 Individual assoc. rules for array formal are not valid

It was agreed that this IR has been superseded by LCS-2006-113

ACTION: All to vote.

TOPIC: IR 2062 Range staticness

It was agreed that this IR is ready for an ISAC vote.

ACTION: All to vote.

TOPIC: IR 2074 Problem with direct/select visibility in formal part

This IR deals with a port map of the form X01(X01) => expression. In
one case the intended interpretation is that X01 be a type conversion,
in a second case the interpretation is that the leftmost X01 is a port
name.

The basic principle is that select visibility (i.e., as a port name) takes
precedence over direct visibility (as a subtype name).

It was decided that both examples are illegal, for different reasons.

In the recommendation visibility by selection was extended to the entire
formal part or actual part instead of just to the formal designator
or actual designator in order the let visibility rules be used to disambiguate
the expressions in question.

This is ready for an ISAC vote.

ACTION: All to vote.

TOPIC: IR 2082 Elaboration of unconstrained interface objects

This IR deals with problems in the current LRM in elaborating unconstrained
interface objects.

LCS-2006-118 resolves this issue for generic constants used in generic maps.

The suggested resolution extends the wording of the LCS to port maps
and interface parameters. Here are some of the high points of the IR:

a) Its not clear whether clause 12.3.1.4, Object declarations applies
to interface objects. The proposal explicitly states that the clause
does not apply to interface objects and updates 12.5 to specifically
apply to subprogram parameters.

b) Its not clear whether parameters of mode OUT are initialized.  The
proposal updates 12.5 to specify that they are initialized.

c) A major feature of LCS-2006-118 deals with order of elaboration of
generics, so that a generic value can be used to determine subsequent
generic values in the same generic map. In the proposal, this was not
extend to ports, in part to bypass the issue of when OPEN or
unassociated ports are evaluated. Ordering similar to that of generics
could easily be added to the proposal.

ACTION: All to review.

TOPIC: IR 2085 What happens when a parameter of mode out is not assigned in a
procedure?

This issue has been subsumed by IR 2082. If, and when, IR 2082 is
approved, this IR will be superseded.

ACTION: None until IR 2082 is resolved.

TOPIC: IR 2086 Incorrect description of type mark in disconnection
specification

The technical rewording has been done and this IR is ready for an ISAC
vote.

ACTION: All to vote.

TOPIC: Review of old ISAC issues

It appears that we soon will be ready to continue our examination of
older ISAC issues.

ACTION: Chuck to look into assigning 5 or so issues to each ISAC
member to determine status.