Re: 21.5b9 on MacOSX build problem
Stephen J. Turnbull wrote:
Note that we will probably have a Carbonized version (by Pitts
Jarvis)
in CVS in a week or so.
Very good news indeed.
[]
It's also possible that we simply have broken code and a wild pointer
(eg, a boundary problem) somewhere. If that's the case, we need
somebody to trap the bug in the debugger. Anybody have a MacOS system
to donate to the cause? ;-)
The sourceforge compile farm has a Mac OSX system.
Try (1) --use-union-type (which is as the gods intended things to
be,
and should cause no problems with aliasing restrictions, or (2) (if
your compiler supports it) --no-strict-aliasing (or whatever the
option is called on your compiler that suppresses the options that
make strong assumptions about aliasing).
I tried -fno-strict-aliasing. Instead of
gcc -c -g -O3
the compiler flags are now
gcc -c -fno-strict-aliasing
Unfortunately, the result is still the same, bus error when the newly
dumped xemacs is used for the first time.
--
Martin