On 2008-03-29 22:57:36 -0500, Robert Spykerman
<robert.spykerman@[EMAIL PROTECTED]
> said:
> I finally got 0.6.2 compiled with the long long switch
>
> http://www.complang.tuwien.ac.at/forth/gforth/Known-problems.html
>
> It runs, it links to ffcall dynamically but there are a few issues I
> have yet to sit down and figure out ie the word see seems to crash the
> system.
>
> *** Has anyone build a gforth 0.6.x on an old PPC G4 set up like mine
> succesfully and can tell me about their experiences?
>
> I know it's been done :
Robert, I have been running gforth on G3/4/5 systems for years. You
found one of the issues with 0.6.2. There are others, but I don't
remember what they were at the moment. Anton and I worked most
recently on MacIntel boxes, and last I checked, his latest CVS version
has all the wrinkles worked out. I'm sure Anton would agree that you
should be using it. On PPC systems you might be able to get away with
the 0.6.9 snapshot, but as others have commented, waiting for gforth
releases or sup****t is problematic. The CVS approach is your best bet.
>
> Bit of googling and I note Ian Osgood posted a link to a .dmg with a
> prebuilt binary (0.6.2 with ffcall statically linked in) circa
> 2006. ).
>
> http://www.quirkster.com/forth/gforth.dmg.gz
>
> This appears to work beutifully - it may solve your problem but it's
> not likely G5 optimised - it ran on my G4.
What kind of optimizations are you looking for? I haven't played with
it, but making it be a 64-bit Forth would be interesting, but even
then, getting OSX to let you use 64-bit calls will be problematic (last
I looked).
>
> I don't actually know if it has a PPC asm built in - can't check I'm
> out on my Eee at the moment (great piece of kit).
The CVS version does, but not sure about 0.6.9 and 0.6.2 does not.
>
> Hmm... come to think of it I recall I had some slight problems
> building gforth on my Eee now come to think of it - I'll post my
> workaround i case anyone's in the same boat.
>
> Robert Spykerman
DaR
|
