Hi,
> Icarus is in a strange space because it is BOTH a simulator and
> synthesizer. I don't know of ANY other tool that does that.
Has consideration been given to separating the target-specific
portion of IVL (simulation and synthesis) from the parsing front-end in
a more decided way? I suppose what I envision would be a separate
synthesis application that uses IVL's parse-tree output (the same
parse-tree available to the targets right now). The advantage to being a
separate application would be that separate config scripts could be
used, and separate options could be used. Hey, a GUI could even be
attached to the synthesis app (my type of pet project). Another big
advantage is modularity. Need a verilog parser? Use Icarus's Verilog
Parsing Library with Your Application (tm).
Anyway, just my .02...
Regards,
Matt
On Tue, 2002-07-02 at 11:55, Steve Wilson wrote:
> A pre-processor trick as mentioned below might work - the simple fact is
> that the comments mechanism is the way it's done. If done otherwise -
> then we're the one guy in the marching band in step ;-)
>
> I HONESTLY don't think there is a way around this.
>
> Icarus is in a strange space because it is BOTH a simulator and
> synthesizer. I don't know of ANY other tool that does that. Think
> about it a second. You're Synoopsys. You parse the comments looking fo
> r your comments as part of your problem set. YOu don't have to worry
> about ALOT of the language since you only handle a small sub-set. Your
> life is easy. Now enter Cadence - you have to handle the full language
> - but certainly comments is comments ;-) We don't have to apply
> synthesis templates trying figure out what gates would make the best
> implementation, etc.
>
> Anyway - I vote for handling the comments in some manner (which
> OBVIOUSLY) I don't have to implement - ain't I brave with Mr. Williams'
> time ;-)
>
> Steve Wilson
>
>
> Larry Doolittle wrote:
> >>>Despite the obvious ugliness it seems at least the translate_on and
> >>>_off and parallel_case/full_case commands are pretty ubiquitous.
> >>
> >>Yeah, I'm aware of those. I might just translate them to attributes
> >>myself and let Synopsis follow me:-)
> >
> >
> > For those poor slobs who have existing code bases to deal with,
> > how about adding a preprocessing option to run the input through
> > a perl script that converts those yucky comment-based directives
> > into the elegant attribute-based directives?
> >
> > - Larry
> >
> >
>
>
>
