[plt-scheme] Racket v4.90.0.1 is available for testing

Hi Eli,

I tried main trunk of libffi (3.0.9) and greeted with a wall of error messages. I'm afraid it's so mingled with Gcc that it will not be easy.
Nevertheless I managed to compile the version inside your source tree with the changes I submitted.

The reason to change in foreign.c was due to this error:
"./foreign.c", line 1799: cannot take sizeof void

Apparently RETSIZE() macro resolve to something like "sizeof(void)" which I suspect is not valid according to ISO C.

Is there any way to disable use of libffi?
Is it via "--disable-foreign" flag to configure script?

Thanks, regards
Semih Cemiloglu


-----Original Message-----
From: Eli Barzilay [mailto:eli at barzilay.org] 
Sent: Friday, 28 May 2010 1:23 PM
To: Semih Cemiloglu
Cc: plt-scheme at list.cs.brown.edu
Subject: RE: [plt-scheme] Racket v4.90.0.1 is available for testing

On May 28, Semih Cemiloglu wrote:
> Hi Eli,
> 
> I tried to compile it on OpenSolaris (x86 - version 2009.06) with
> Sun's C compiler (from Sun Studio 12 update 1).  My configure
> options were:
> 
> $ ./configure --prefix=/stow/racket-4.90.0 --disable-jit 2>&1 |tee conf.log
> 
> I order to advance, I did some modifications to the source tree as
> well as generated files (Makefile, libtool) which are attached to
> this message as diff file.

One of these changes code in libffi -- and I'm surprised that it
compiles using the sun compiler.  I'm not sure about changing it
without checking whether the newer version of libffi deals with it.

Also, why is this

  #if __SUNPRO_C          
        RETSIZE(void*);
  #else
        RETSIZE(void);
  #endif    

needed?  (See the definition of RETSIZE just before the switch
statement.)


> Later on I experienced following error:
> [...]
> env XFORM_USE_PRECOMP=xsrc/precomp.h ../racketcgc -cqu ./xform.rkt --setup . --cpp "cc -E -I./.. -I./../include -DMZ_DONT_USE_JIT   -DNEWGC_BTC_ACCOUNT  -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64 "  -o xsrc/salloc.c ./../src/salloc.c
> precompiled-header: line mismatch with precompiled: #"#ident \"acomp: Sun C 5.10 SunOS_i386 2009/06/03\"" (line 4015) versus #"#ident\t\"%Z%%M%\t%I%\t%E% SMI\""

(This is something for Matthew to look at.)

-- 
          ((lambda (x) (x x)) (lambda (x) (x x)))          Eli Barzilay:
                    http://barzilay.org/                   Maze is Life!