SV-BC Separate Compilation discussion 10/02/03

Attendees
100000000 Month
009876654
021012012 Day
236553928
000000000 Year
333333333
aaaaaaaaa Karen Pieper
aaa--aaaa Randy Misustin
a--a-a--a Dennis Brophy
--------a Danny Jacobi
aaaaaaaaa Brad Pierce
-a--aaa-a Dave Rich
a-a----aa Francoise Martinolle
a---a-aa- Matt Maidment
--a-a-a-- Tom Kiley
-----a--- Johny Srouji
---aaa--- Jay Lawrence
aaa-a---- Arturo Salz
--a-a---- Peter Flake
aa------- Nikhil

There was an issue in the table describing the error in the example 
following the table.  The Error condition needs to be in the cell
with the column import q::* and row import p.  So, Karen swapped
the two cells (with appropriate q/p changes).

A proposed solution is that when a q::* import is done, and a reference
is made to c, followed by an import either by q::* or import q, any
reference to c after the import of q is an error.

Why use ::?  To allow separate compilation, we need to be able to
distinguish between modules and packages in hierarchical reference.

Are we going to eliminate the semantics of import p;?
	For:  Randy, Matt, Francoise
	Against:  
	Abstain:  Arturo, Brad, Nikhil

Francoise moves that we require import p before any uses of "p::a",
    to specify that p is a package rather than a module, and then
    all references to "p::a" would be "p.a".
	There is no second.

	Matt, Nikhil, Arturo, Randy are for ::
	Francoise is for .

Table needs to be clarified when the references are occurring 
relative to the import statements.  And hte import p row gets
to be deleted.

We all agree that import statements will be allowed anywhere
a declaration is allowed.

1) Now, we need to define what happens in the case following the 
table.

2) imports in modules means that ports lists have to use :: reference

3) modules in packages?
Nikhil moves that we add modules to packages.  No second.


Straw poll on issue 1.  Randy and Francoise abstain.  Arturo,
	Matt, Nikhil are for an error.
        
The next meeting is 10/7/03 at 8:30am Pacific.