RE: [sv-ec] Re: [sv-bc] Name resolution and imports

Limiting this discussion to SV-EC...

Because inheritance (or extension)of a class is defined as the chaining
of visible symbols. An extended class does not need to know the name of
its parent. That's why we have 'super'.

Dave


> -----Original Message-----
> From: Greg Jaxon [mailto:Greg.Jaxon@synopsys.com]
> Sent: Monday, September 11, 2006 2:02 PM
> To: Francoise Martinolle
> Cc: Rich, Dave; Arturo Salz; Vreugdenhil, Gordon; Logie Ramachandran;
Mark
> Hartoog; SV_BC List; SV_EC List; sv-ac@verilog.org
> Subject: Re: [sv-ec] Re: [sv-bc] Name resolution and imports
> 
> What do you see as support for that claim? The identifier "base"
> is not bound inside the scope of class base, or its descendants;
> it is bound in the package scope that is exporting it.
> 
> I agree that the base class can be used freely by anyone who
> imports a class derived from it, but they surely did only know its
> name if the producer of the derived class passes it along.
> Of course most folks would want that effect, which is why
> package chaining seemed so natural.
> 
> Greg
> 
> 
> Francoise Martinolle wrote:
> > I agree with Dave, the base class is visible through the class
> > inheritance of p3.
> >
> > Francoise
> >     '
> 
> > I might disagree that is not legal. Class p3 will get access to the
> > symbols via the class hierarchy mechanism. It does not need to go
> > through the package.
> >
> > Dave
> >
> >
> >> -----Original Message-----
> >> From: owner-sv-bc@server.eda.org
[mailto:owner-sv-bc@server.eda.org]
> > On
> >> Behalf Of Arturo Salz
> >> Sent: Monday, September 11, 2006 11:00 AM
> >> To: Vreugdenhil, Gordon; Arturo Salz
> >> Cc: Logie Ramachandran; Mark Hartoog; SV_BC List; SV_EC List; sv-
> >> ac@server.verilog.org
> >> Subject: RE: [sv-ec] Re: [sv-bc] Name resolution and imports
> >>
> >> You are correct in that simple case, but, consider the following
> > example
> >>    package p1;
> >>       class base;
> >>          int x;
> >>          virtual function void F(base x); ...  endfunction
> >>       endclass
> >>    endpackage
> >>
> >>    package p2;
> >>       class derived extends p1::base;
> >>          int y;
> >>       endclass
> >>    endpackage
> >>
> >>    module top;
> >>       import p2::*;
> >>
> >>       class p3 extends derived;
> >>          virtual function void F(base x);	// this is not legal
> >> without the re-import
> >> 	 ...
> >>         endfunction
> >>
> >>         ...
> >>    endmdodule
> >>
> >> In this case, the base class is not visible to the importing scope
> >> without explicitly importing package p1. Or is it? Why should
> > end-users
> >> need to add an explicit import of p1 (import p1::*) or explicitly
> > write
> >> (p1::base)?
> >>
> >> 	Arturo