Hi Uwe, On Thu, Jun 14, 2012 at 1:11 AM, Uwe F. Mayer <uwe_f_ma...@yahoo.com> wrote: > You did not check in the change below (you may have left that one out on > purpose, I don't know). If I recall correctly, having > write_barrier_failing_case_ptr uninitialized caused an error when compiling > with --gc=hybrid --opt=jit. I doubt this is Cygwin specific.
Ah, I see now. But the error is deeper; your fix merely causes the JIT to call a NULL function pointer at run-time, as far as I can tell. We indeed don't have any test for hybrid+JIT, which means that over time it just broke. I "fixed" it by definitely requiring --gc=minimark with the JIT. A bientôt, Armin. _______________________________________________ pypy-dev mailing list pypy-dev@python.org http://mail.python.org/mailman/listinfo/pypy-dev
