gnupic: Thread: SDCC status

For those of you following...

The pcode stuff I talked about a couple of weeks ago is now in CVS. The compiler
is still not useful (for generating PIC code) but it's getting there. The files
in the sdcc/src/regression directory of CVS do compile using the new pcode
logic. However, they are not optimized by the peephole optimizer. But that's
what I'm working on now.

Scott

I've been plugging away at SDCC and have made some progress. As copied
from http://www.dattalo.com/gnupic/sdcc.html :


    Comparisons with literals has been highly optimized.
    Signed addition and subtraction now work.
    Signed comparisons now work.
    And'ing with a constant now works
    signed casting now works
    compound if statements now work
    And the usual bug fixes

The PIC port of SDCC is still in alpha stage. I'm going to complete the
rest of the operators over the next couple of weeks. After this, SDCC
might be actually usable for very simple applications. This should be
complete by mid November.

Once the operators have been completed, I'll begin working on (better)
parameter passing.

Scott

Scott Dattalo wrote:
> 
> I've been plugging away at SDCC and have made some progress. As copied
> from http://www.dattalo.com/gnupic/sdcc.html :
> 
>     Comparisons with literals has been highly optimized.
>     Signed addition and subtraction now work.
>     Signed comparisons now work.
>     And'ing with a constant now works
>     signed casting now works
>     compound if statements now work
>     And the usual bug fixes
> 

I have ported/fixed the small-c compiler that was placed on the
gnupic website (by John Fatava?). The version I have working generates
assembly code for MPASM or gpasm. Also, I've modified some of the
code generation (slight improvement) and debugged a few of the
bugs inherent on the compiler. I have used this compiler on a couple
of small to medium projects and so far works ok. If anyone is
interested,
or if Scott wants to put this version on the gnupic website, please
let me know.

Erasmo

i am very interested. do you support more than the original
16C84? if you do not have this on a website, i'd be happy to
put it on mine.

let me know how i can get it,

michael


On Wed, 17 Oct 2001, Erasmo Brenes wrote:

> Scott Dattalo wrote:
> >
> > I've been plugging away at SDCC and have made some progress. As copied
> > from http://www.dattalo.com/gnupic/sdcc.html :
> >
> >     Comparisons with literals has been highly optimized.
> >     Signed addition and subtraction now work.
> >     Signed comparisons now work.
> >     And'ing with a constant now works
> >     signed casting now works
> >     compound if statements now work
> >     And the usual bug fixes
> >
>
> I have ported/fixed the small-c compiler that was placed on the
> gnupic website (by John Fatava?). The version I have working generates
> assembly code for MPASM or gpasm. Also, I've modified some of the
> code generation (slight improvement) and debugged a few of the
> bugs inherent on the compiler. I have used this compiler on a couple
> of small to medium projects and so far works ok. If anyone is
> interested,
> or if Scott wants to put this version on the gnupic website, please
> let me know.
>
> Erasmo
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: ####@####.####
> For additional commands, e-mail: ####@####.####
>

On Wed, 17 Oct 2001, Erasmo Brenes wrote:

> I have ported/fixed the small-c compiler that was placed on the
> gnupic website (by John Fatava?). The version I have working generates
> assembly code for MPASM or gpasm. Also, I've modified some of the
> code generation (slight improvement) and debugged a few of the
> bugs inherent on the compiler. I have used this compiler on a couple
> of small to medium projects and so far works ok. If anyone is
> interested,
> or if Scott wants to put this version on the gnupic website, please
> let me know.

Sure, send it over. I'll be glad to put it up on the gnupic webpage.

I'm really interested to see how you did this. When I looked at it a year
or so ago I almost had to take anti-nausea medication. I reckon though if
you don't care about speed or processor utilization, then small-c is just
fine!

Scott

Scott Dattalo wrote:
> 
> On Wed, 17 Oct 2001, Erasmo Brenes wrote:
> 
> > I have ported/fixed the small-c compiler that was placed on the
> > gnupic website (by John Fatava?). The version I have working generates
> ...
> 
> I'm really interested to see how you did this. When I looked at it a year
> or so ago I almost had to take anti-nausea medication. I reckon though if
> you don't care about speed or processor utilization, then small-c is just
> fine!
> 
> Scott
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: ####@####.####
> For additional commands, e-mail: ####@####.####

Well, the first time I looked at the code, I got the same reaction.
As a matter of fact, I stopped for a while developing it because it is
such a nightmare to follow how the code actually works and there is
little documentation. But on the other hand, there wasn't anything
else out there (for free) and the development of the sdcc was/is still
in progress. Also, I needed to work with 16F87x and there wasn't
anything there for them.

As you point out, it doesn't have any real optimization code in it. I
have improved the code generation module to detect simple constructs,
such as incf, etc. as opposed to the original code. Nevertheless, there
are still "dumb" code generation outputs, such as c=0; being translated
as movlw #0, movwf c_, instead of clrf c_. But, for small to medium
projects, these dumb translations can be filtered out after to speed up
and shrink the code.

Should I tar the sources and send them to you?

Erasmo

It's still happen'. The PIC Port of SDCC (http://sdcc.sourceforge.net/) is
making progress.

The latest addition has been with parameter passing. Now parameters can be
passed to/from functions. There have also been numerous enhancements and
bug fixes. (e.g. ^ was always working, but ^= wasn't ). I also changed the
default process to a 16F877 (from an f84). I still don't have a mechanism
for selecting the processor yet.

On the gpsim side, Ralf has got the beginning of C source debugging
working! This is really impressive. There are still a couple of "minor"
details to work out. :). Once we get these done, I'll post a few screen
shots and perhaps another gpsim release.

Scott

It's been a while since I posted anything about the status of the PIC port
of SDCC. But that shouldn't be misconstrued as no work being performed.
Over the last couple of weeks I've been working on a "real life" C program
and trying to get it to compile under SDCC:

http://www.bobblick.com/techref/projects/lcdterm/lcdterm.html

This is Bob Blick's LCD C program for driving LCD's. I figured that since
gpsim has an LCD module, then this would be a good example to try to get
working with SDCC. Here's a screen shot of Bob's program compiled with
SDCC and simulated with gpsim:

http://www.dattalo.com/bblick_lcd.jpg

I did have to modify his program slightly to accommodate the differences
in the way Hitech and SDCC declare bits. I also had to strip out the RS232
and keyboard code. But the underlying LCD stuff was untouched.

Before starting this I thought SDCC was almost ready. However, there were
numerous detail bugs that had to be fixed. And I might add, there are
still numerous bugs... There are also still some fundamental design issues
to be resolved as well (e.g. interrupts are not yet supported).

I'm still plugging away and I think that a March Beta release of the PIC
Port of SDCC is still feasible (It just may be midnight PST March 31 :).

Scott