[comp.sys.amiga] GNU C compiler

mjp@spice.cs.cmu.edu.UUCP (03/26/87)

Is anyone thinking of doing an Amiga port of the GNU C compiler?  this
would be a boon to everyone in the audience on a limited budget, plus
it might give Lattice and Manx incentive to compete, given a free
competitor.  Does anyone know what kind of demands the GNU C compiler
makes on system resources (like memory and disk space)?

I assume that along with the port someone would have to write a set of
libraries for accessing the Amiga ROM routines as well.
-- 

Mike Portuesi / Carnegie-Mellon University Computer Science Department
ARPA:	mjp@spice.cs.cmu.edu
UUCP:	{harvard | seismo | ucbvax | decwrl}!spice.cs.cmu.edu!mjp
BITNET:	s314mp1u@cmccvb, rainwalker@drycas (pick one)

"Amiga hackers do it graphically, with lots of sound effects"

"Mac owners dream in black and white
 Atari owners dream in color...
 but Amigoids dream using Hold and Modify!"

rokicki@rocky.UUCP (03/26/87)

mjp@spice.cs.cmu.edu (Michael Portuesi) writes:
> Is anyone thinking of doing an Amiga port of the GNU C compiler?  this
> would be a boon to everyone in the audience on a limited budget, plus
> it might give Lattice and Manx incentive to compete, given a free
> competitor.  Does anyone know what kind of demands the GNU C compiler
> makes on system resources (like memory and disk space)?

I've got it, and I'm hacking on it.  (It's called GNU gcc.)
Three comments:
   1.  The executable for the first pass on a 750 is 1.2M long.
   2.  It optimizes very well---it generates better code than Manx
       Lattice.
   3.  It optimizes too well---there are still some bugs to be removed.

If anyone wants to hack on it with me, drop me a line.  (mail to
...!decvax!decwrl!sushi.stanford.edu!rokicki, for instance, or
rokicki@sushi.stanford.edu preferred.)   For instance, I haven't
located a nice assembler yet . . .

-tom

jdg@elmgate.UUCP (03/28/87)

We've compiled GCC on a Sun-3 with optimizations on.  The STRIPPED
executable was 400k+.  Further the compiler likes to take it's memory from
the stack. No sweat with an MMU,  but without you'd have to set a fixed
stack size that is possitively huge.  The GCC optimizer really eats memory
fast.  There is a bug or two in it's 68000 (not 020) code generation for
long division, etc...  Anyways good luck.

As a side note: Why does it begin to appear that when your C program
,compiled with Manx 3.4, bombs, you had better check the code it
generated first instead of checking your code first??   8^)


-- 
Jeff Gortatowsky       {seismo,allegra}!rochester!kodak!elmgate!jdg
Eastman Kodak Company  
These comments are mine alone and not Eastman Kodak's. How's that for a
simple and complete disclaimer? 

gcglan@sdrc.UUCP (frank glandorf) (12/13/89)

With the recent discussion on new versions of C compilers from
Manx and Lattice I thought I'd check if there is a port of GCC,
the C compiler from GNU. Unfortunately they don't have one but
Leonard Tower Jr. suggested someone may be working on one (they
distribute source with the compiler). There are ports to other
MC680x0 machines. All you have to do is develop the runtime
libraries :-) 
 
In addition to GCC, a linker and source code debugger, the GNU
project includes a number of un*x style utilities like gawk
(awk), flex (fast lex), bison (an american yacc :-)), etc. They
also have a unique 'licensing' agreement which encourages the
free exchange of software. 

Frank Glandorf -- gcglan@sdrc.uu.net 
"If the end of the world was annouced tomorrow, I'd move to Cincinnati
since everything happens there twenty years later" -- Samuel L. Clemens

ckp@grebyn.com (Checkpoint Technologies) (12/14/89)

In article <985@sdrc.UUCP> gcglan@sdrc.UUCP (frank glandorf) writes:
>
>With the recent discussion on new versions of C compilers from
>Manx and Lattice I thought I'd check if there is a port of GCC,
>the C compiler from GNU... 

NeXT has a GNU C Compiler for the 68030 -- you should (according to the
GNU licencing agreement) be able to get source from them.   I would
assume that it must be fairly robust to be sold as a comercial product.