Re: Running x86 Linux GNAT on MacOS under Virtual PC

[jim <jim_evart@yahoo.com>]

In article <5ee5b646.0108122056.2e61fad3@posting.google.com>, Robert
Dewar <dewar@gnat.com> wrote:

> That's misleading, in fact the Mac OS X port is quite tricky
> (ask Jim Hopper!) and is much more than just a recompilation!

there are a whole team of us working on this port right now.  myself,
Alan and Andrew reynolds, Stephen J. Bespalko, John Matthews, and
others who i am probably forgetting off the top of my head.  in
addition Stan Shebs at apple who is the keeper of the gcc 3.0 port is
also working with us.  we have also been talking to the apple pthreads
team.   In addition the folks at ACT have supplied us with gnat sources
and answers to some of our questions (those which were not apple
related)

we have had to do a number of things to the gcc port to make it support
gnat.  there were a few shortcuts taken in apples port of gcc that have
the backend calling the front end.  and while this works fine for the
c, and c++ front ends it was intended for and seems to work for
fortran, it doesnt work at all for ada.  so Stan, Alan, Andrew, and
myself have made changes to gcc to try to fix some of these.  we have
fixed a few of them but we are not done yet.  seems each time we get
past one of these another crops up. 

we also had to add trampoline support to apples gcc as it was never
made to work on the os x port for any language.  so nested functions
never worked in apples gcc, but they will in 3.0 if you have the patch
now.  Stan is going to get it into the mainline sources.  right now we
are working on another gcc problem.  we succeeded in building a gnat1,
but it won't bootstrap which we are working on.

Ptheads is another matter, apples implementation of pthreads is not
complete. However the apple folks are willing to look at adding what we
need and one of the team is corresponding with them about it.  In the
short term we have fsu threads ported so we will simply have to use
that in the short term.

i was off work on disability leave oever the summer and just got back
to work after getting the use of my arm back.  the upshot of this is
that i have been working on this at a reduced rate working one handed
until recently.  The Reynolds brothers have taken over the lead on
getting the gcc problems fixed in the last month or so while i was
recovering from getting my elbow cut up.  

we have set up a public mailing list for those who want to be kept up
to date with whats going on.  its at:

http://www.lists.apple.com/cgi-bin/mwf/board_show.pl?id=3&pg=2

apple is publicising it on there system and making it easy to get to
but its being hosted with the kind support of George Washington
University and Mike Feldman.  we have a developers list as well but we
are trying to keep people off that unless they can help with some phase
of this.

once we get the basic compiler going we will have a lot of work on the
bindings to carbon coco etc.  

its not near as neat as it would be if we had a gcc that was complete. 
but with the help of stan shebs and the reynolds brothers we are
getting there.

best jim