SV: GCJ 3.3 garbage collection live-lock(conjencture) and GDB for iditos and
Wed Feb 26 21:50:00 GMT 2003


> I've found something which may be yet another clue to the
> live-lock problem.
> GDB discoveres a segmentation fault, which does not appear
> during normal execution of my program. My conjencture is
> that the normal Windows running environment is a bit more
> forgiving than GDB; this is intentionally so to make
> programs run in the field while making the bugs easier to
> find.

Maybe this is just because Windows and Linux employ
memory-management schemas. I've seen many C and Pascal
programs who do get a
SIGSEGV on Linux but runs without a hitch on many Windows
flavors. Many of them
were my students exercises (linked lists and so on). Every
time their code had
bugs (like not initializing a ponter). Have you ever
wonderded why viruses
(old-time viruses, not today worms) spread easily on Windows
but are infesible
in Linux? :)

[]s, Fernando Lozano

More information about the Java mailing list