Re: Idea: Multitasking graphical OS for the GS

["geoff" <geoff@a2central.com.remove-2dq-this>]

  To: David Schmenk
gcc already has 65816 support.  See 
http://sourceware.org/cgi-bin/cvsweb.cgi/src/bfd/?cvsroot=src

binutils is the first step to compile gcc.  It provides the assembler, linker, 
and basic set of libraries.  It is pretty trivial to make the few needed 
changes to get binutils to compile with w65 support (it didn't work out of the 
box for me when I looked at three years ago to set it up for cross compiling).
 The next step for IIgs functionality is to add GS/OS OMF support to ld.  
After that, a 16-bit gcc should just work to compile native IIgs apps (not to 
mention a very decent assembler and ar to build libraries).  A more useful gcc 
would be to add new 65816 code to binutils for 32-bit support (where ones 
passes the -m32 argument to gcc to force 32bit ints).  I since lost my notes 
that I wrote due to a bad OS upgrade.

Is there anybody who want to tackle adding OMF support to ld in binutils?  A 
java port of most of OMF can be found at http://code.google.com/p/libomf/ .  I 
don't know how complete or accurate it is.

Geoff

 > mdj wrote:
 > > On Jan 3, 3:52 pm, David Schmenk <dschm...@sbcglobal.net> wrote:
 > 
 > >> Another important piece to consider is the development environment.
 > >> Writing a core multitasking kernel isn't particularly difficult, but
 > >> having tools such as compilers and debuggers is a large part of the
 > >> project - sometimes larger than the kernel itself. If you want to make 
 > >> it open source, you'll want to find a tool suite that can be made easily 
 > >> available.  Do you want to self-host or do you want to cross-compile? >> 
 > Determination of the tools might even help dictate the design and
 > >> complexity of the kernel.
 > > 
 > > The reliance on a commercial tool chain has always been an impediment
 > > to IIgs development... Well, that plus the 5 manuals you'll need to
 > > buy. I guess there's Merlin 16, but a non-Unixy toolchain just seems 
 > > 'wrong' nowadays
 > 
 > > Matt
 > 
 > So who is going to add 65816 support to gcc ;-)