Meeting 07-26-2004

Attendance
Jim Lewis
Karl Eisnhofer
Chuck Swart
John Ries
Charlie
Dave Bishop

FT25
  Even with the proposal there are places where qualified expressions are needed
  due to normal ambiguity. For example concatenation of std_logic. The result
  could be at least std_logic_vector and std_ulogic_vector. We don't have a
  problem FT15, multidimensional array slices, because the slicing will either
  produce a one-dimensional type which would be handled just like any other
  one-dimensional array or another type which would be an error.  Chuck believes
  that the wording of the proposal needs to be updated to correctly reflect the
  checks on choice lengths.  
 
  Action item for John to work with Chuck to update the proposal's wording.

FT12
  Ada now allows reading of output variables. There are two proposals here,
  which one should we use?  In previous discussions, it was proposed that we use
  the attribute, but if an output appears in a place where it must be read, the
  attribute is implicitly used.  The attribute would allow for selecting which
  value is used when reading a inout. There was a discussion on what value an
  output has for composite parameters.  A number of issues on reading subprogram
  parameters of mode out.  Charlie suggested that we don't allow reading of
  subprogram out parameters. There is a clear need in the concurrent context,
  but do we have a need in subprograms? Jim proposed that only reading of signal
  ports of mode OUT be fast tracked and subprogram parameters be deferred.
  Chuck points out that this will make LRM work a lot harder because of the
  commonality of ports and parameters.  Chuck's counter proposal is to allow for
  reading of signals parameters/ports and no allow it for variables.  Chuck's
  proposal was agreed on.

  Action item for Steve  Bailey to update proposal to reflect this.


FT11
  No new proposal, Eric had indicated that he would do this. Skipping.

FT16
  Write up does not specify how the LRM is to be changed but the intent is
  clear.  Proposal needs to be updated, there is no longer an default context.
  If updated, we will review again, otherwise FT16 will be removed from fast track.

FT19
  Main issue is sensitivity to non-local signal access. Discussion on what this means and possible
  work arounds.  One options is to compute the usage of global signals at
  elaboration time.  Option 2 would be not allow any references to subprogram in
  packages and impure functions in packages.  Option 3 would allow calls to
  procedures and function in packages along as they don't access global signals.
  Option 4 is just ignore the access to global signals.

  David doesn't care about the use of globals.  He has only seen them when the
  design is translated from Verilog to VHDL.

  Option 3 is the consensus.

  Attribute vs keyword. discussion.  Group likes keyword. 

  Action item for John to rewrite proposal adding the restrictions of option 3.

Meeting closed.