Re: boehm-gc .comm problem

[Tom Tromey <tromey at redhat dot com>]

>>>>> "Tony" == Tony Kimball <alk@pobox.com> writes:

Tony> It's hard to know which path to take, and difficult to justify
Tony> continuing down any one path to overcome an obstacle, when one
Tony> knows that there are other, untried paths, which might avoid all
Tony> obstacles and provide a cheap win.

If your goal is to get gcj-compiled Java programs running on Windows,
then my advice is to do a native port.  One reason for this is that
the GC needs a lot of platform-specific information.  Eliminating
Cygwin means that is one less layer to worry about; debugging will
probably be easier.

This doesn't mean a Cygwin port wouldn't be useful.  In fact, I'd like
to see both a Cygwin port and a native Windows port of libgcj.

In some ways doing a full native port is going to be more work than a
Cygwin port.  There are plenty of POSIX-y assumptions in the current
code that will need to be cleaned up.  However I think the work
involved in this route, while there is more of it, is likely to be
easier than porting to Cygwin.

Maybe I'm wrong though.  This is just a guess.


As far as how to host it, when I did Windows development in the past I
did all my programming on Linux and did cross-builds.  This was far
more comfortable for me.  This was in 1997, too, when Cygwin was
(presumably) less reliable than it is now.  On the other hand, I was
working with a unified tree so I didn't have to worry about separately
building and installing binutils.

Tom

--
Want to unsubscribe from this list?
Check out: http://cygwin.com/ml/#unsubscribe-simple