Re: pragmas


Subject: Re: pragmas
From: Paul Graham (pgraham@Cadence.COM)
Date: Tue Dec 18 2001 - 07:12:56 PST


I'm a tool developer, not an end user, so I may not be the best person to
comment. But based on customer test cases, I haven't seen too many spurious
pragma warnings. One thing you'll sometimes see is a comment in a verilog
design like:

    // do this funny kludge to work around a
    // synopsys bug

and our tool reports:

    // synopsys bug
       |
    Unrecognized synthesis pragma

But such spurious warnings probably aren't that frequent. With attributes,
I'd say there are fewer chances of this happening. And a tool could have
an option to ignore attributes with a given attribute identifier. Say,
ignore all attributes beginning with "simulation", process all attributes
beginning with "synthesis", and warn about all others.

Paul

> Right, which is why I do not like that solution.
>
> Paul Graham wrote:
>
> > > One thing which bothers me about pragmas implemented as comments or
> > > attributes is, what happens if the pragma is mis-spelled and therefore
> > > ignored by the tool ?
> > >
> > > How will such an error be detected ?
> >
> > The tool could issue a warning for any unrecognized pragma. This might get
> > annoying if, say, many non-synthesis pragmas appear in a synthesizable
> > module.



This archive was generated by hypermail 2b28 : Tue Dec 18 2001 - 07:19:24 PST