On Tue, 18 Jun 2002, Michael Rigby-Jones wrote:

> > -----Original Message-----
> > From: Andy Reynolds [SMTP:keehar@BTINTERNET.COM]
> > Sent: Tuesday, June 18, 2002 12:41 PM
> > To:   PICLIST@MITVMA.MIT.EDU
> > Subject:      Re: [EE] PD Pic compiler
> >
> > Michael,
> >
> > I went to www.dattalo.com and sourceforge but couldn't find a pic
> > download...
> >
> > Do you know where the link to the pic stuff is?
> >
> >
> Sorry, I don't.  I know the PIC port is still very much under developement,
> and AFAIK Scott is the only person working on it so you'll probably get some
> more usefull answers from him.


Yes, the SDCC PIC Port is being actively developed. At the moment, the
only way to get the code is from CVS. There may be a binary or two
floating around, but if they're more than a day old then they're obsolete.

The CVS code is here:

http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/sdcc/

If you want to see how much stuff works with the PIC Port, then examine
these working test files:

http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/sdcc/sdcc/src/regression/

and the PIC port portion of SDCC is here:

http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/sdcc/sdcc/src/pic/


If you go to SDCC's Home Page (http://sdcc.sourceforge.net/), you can find
instructions for getting SDCC out of CVS.

-----
At the moment, there are a handful of people experimenting with SDCC. One
person is trying to port an Ethernet stack! From the bug reports I'm
getting I can tell you the weakest areas of SDCC: Pointers and register
allocation. The algorithm for handling pointers is grossly inefficient.
Furthermore, not all cases of pointer accessing have been tested. I
definitely need to do some work here. Register allocation is also
relatively poor. This is the area I'm going to attack next.

Scott

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.