# On the NCR Tower 32 running SVR3, says ra@intsys.no :
# Do *not* enable optimization in CFLAGS when using the native cc, because:
# a) The optimizer seems to loop when invoked with -O2.
# b) The -O1 level does stack/frame pointer optimizations that make the
#    assembler alloca in libPW.a fail, and the C alloca eats *lots* of memory.
# c) gcc will eventually be recompiled with itself, so all this doesn't matter.
CLIB = -lmalloc -lPW