Meeting 09-13-04
Attendance

John Ries
Jim Lewis
Chuck Swart
Steve Bailey
Deepak Pant
Ajay Varikat
Eric Marschner
David Bishop
Frank Wolf
Peter Ashenden


FT7 
Package of functions vs. Alias.

Long discussion on elaboration issues with aliases.
Eric Marshner proposed that we just add a hierarchy reference
and all info just be inferred from the context. Issues
here are when the alias is resolved.

Spend an hour on this will defer.  We need some one to take
charge of it.

FT10A/B
  John and Jim will get together and determine 

FT11 
  AI from DAC was Eric going to take ownership of this.  Since he hasn't
responded, we are assuming that Eric is not taking ownership.

FT12

 Action Item for Steve update propose base on 07-26-2004 meeting. Allow reading
 of only signal ports and parameters of mode out.

FT15
 John thought this was deferred because we don't have a type for these slices.
 Will deferred.

FT16
  Jim, needs work but won't happen

FT19 
  John has action item to update.

FT20
  Action Item Jim to schedule for review.

FT22
  Action Item John based on 6-28-2004 meeting. Chuck work
  on locally static subtype is already done.

FT23
  Action Item John to update based on 6-28-2004 meeting.
  Dave' Action item completed.

FT24
  Action Items for John from 7-19-2004 meeting.

FT25
  Action Item for John and Chuck to update wording

Dave we need to review code.  Most important to review numeric_bit,
numeric_unsigned, numeric_bit_unsigned, and fix point packages.

Other not reviewed

Generics on Package
  Action Item of Jim to schedule review.

Peter will do review now.

 Eric where is generic list defined. Already defined but augmented to allow
 for types. Generic subprograms allow both type and value generics.

 Eric question, Ada has rules specifying classes of operators that are allowed
 on type.  Peter will address later.


 Eric: Why not remove 4.3.2.1 restriction everywhere not just generic clauses.
 Peter says why not, but was working on assumption of minimal changes.  Eric
 doesn't believe that allow use of interface list doesn't seem to be more work.
 Peter worries that there may problem with dynamic elaboration that would have to
 be explored.

 

 Peter proposes we relax access type restriction at end of section 5. This may be context dependent.
 More work is needed here.

 Section 6 only allows for assignment, equality and inequality.

  Eric's deferred question is addressed here.  Version 1 had Ada class specification, was
 removed based on previous feedback. Thoughts were to fit in classes definitions.

  Eric proposals to address class issue???

  Main reason we are not specifying addition class info is to prevent force a decision
  for Object oriented VHDL.

  Implicitly define operations for equality, inequality, and assignment are the only
  options "passed" with the actual type. If explicitly defined operators are needed
  they must be passed.

  Jim has question on example in section 7. Does example work if type is unsigned
and the instantiation of the package is as a primary.  The answer is that you have
a use clause for unsigned before the instantiation.

  Eric: Does the <> as subprogram parameter work any place as well?  Eric proposes
  that <> semantics for default subprogram be used in the unconstrained array 
  such that if an unconstrained array is given the we should use the range specified
  by the type mark in the unconstrained array type definition.  Peter see what Eric
  is say but Ada didn't do this and <> means that the the value is unspecified.

  Eric withdraws suggestion.

  Jim this proposal doesn't work for case where we want to set a parameter in the design.
  John responded that you can do this with Peter's proposal be separating the types into
  a package that is included in a generic package and made visible by the alias mechanism.