Re: Deprecating wreal

From: Kevin Cameron <edaorg@v-ms.com>
Date: Wed Feb 17 2010 - 16:09:10 PST

http://www.verilog.org/mantis/view.php?id=2378

Wreal as it originally appeared is used pretty much for communicating
the equivalent of a voltage. As such you can use the regular methods of
connecting signals to do that. Wires only have types for syntactic
reasons, the type really belongs to the drivers, contributions or
receivers attached to it.

This whole issue should be left for later, and handled with SV
integration because it has to do with how you classify drivers and
receivers and how they interact with the type system. Wreal is
essentially an unresolved type (real) with discrete drivers and
receivers. If you wanted to (say) have multiple current drivers in a
discrete context that would add together that would require a resolved
type (still real). Since SV adds more types it should all be
rationalized at the IEEE committee along with user defined types and
generalized resolution/connect-module schemes.

The original AMS design objectives required that the language was (as
much as possible) plug-and-play so that analog and digital modules could
be swapped. Wreal breaks that by saying if you want to communicate a
voltage between digital contexts that's different from analog (or mixed)
contexts. It's an unnecessary distinction that dates back to when
somebody's simulator couldn't do something and a quick hack was necessary.

Kev.

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.
Received on Wed Feb 17 16:09:25 2010

This archive was generated by hypermail 2.1.8 : Wed Feb 17 2010 - 16:09:26 PST