RE: [sv-bc] Interfaces in packages

Arturo,

 

An IP provider can put a package and interface in a common library and
use a configuration.

 

I'm not saying its bad idea; I just think it's a lower priority than a
lot of other things on our plate.

 

 

Dave

 

 

________________________________

From: Arturo Salz [mailto:Arturo.Salz@synopsys.com] 
Sent: Sunday, April 30, 2006 1:26 PM
To: Rich, Dave; Brad Pierce; sv-bc@eda.org
Subject: RE: [sv-bc] Interfaces in packages

 

Dave,

 

I believe the basic request is to put the interconnect definition in the
package, that is, the package does not include the interface
instantiation, only its definition. Such usage would make packages a lot
more convenient for IP providers who want to export both the physical
interconnect and the functional model of a component. Currently, the
functional model can be described in a package (as a set of class
declarations for example), but the interconnect (the interface) cannot,
thus, prohibiting references to the (virtual) interface within the class
methods.

 

            Arturo

 

________________________________

From: owner-sv-bc@eda.org [mailto:owner-sv-bc@eda.org] On Behalf Of
Rich, Dave
Sent: Saturday, April 29, 2006 10:48 PM
To: Brad Pierce; sv-bc@eda.org
Subject: RE: [sv-bc] Interfaces in packages

 

Interfaces, along with modules and programs have their own global
namespace, so what purpose does putting them in a package serve?

 

Packages were intended to contain constructs which exist in a local
namespace, and generally need to be declared before used.

 

But probably the most significant restriction about packages is that
they go through compilation without needing any elaboration. That is why
generate statements are not allowed in a package either 

 

Dave

 

 

________________________________

From: owner-sv-bc@server.eda.org [mailto:owner-sv-bc@server.eda.org] On
Behalf Of Brad Pierce
Sent: Friday, April 28, 2006 8:48 AM
To: sv-bc@server.eda.org
Subject: [sv-bc] Interfaces in packages

 

Users are expecting packages to contain interface declarations, and are
surprised when told that the LRM disallows that.

 

Conceptually, users seem to lump interface declarations together with
typedefs and class declarations, perhaps because an interface-type port
declaration can specify the name of an interface in place of the generic
'interface' keyword.  (However, unlike with classes, a particular
specialization of the interface cannot be specified in a port
declaration.)

 

Why does the LRM disallow interface declarations in packages?  If
there's not some serious principle forcing this restriction, user demand
indicates that the restriction should be lifted.

 

-- Brad