Experiencing unreproducible internal compiler errors <<whinge>>

[Eli Zaretskii]

> The issue is that, if you can't repeat a compiler crash, odds are that
> it's not a bug in the compiler.  There's not any source of randomness
> in it that could cause it to crash only sometimes, given the same
> input.

There is one source of randomness: the contents of memory before the
compiler is invoked.  If GCC uses some uninitialized memory, it could
well exhibit seemingly-random-crash type of behavior, even for the
same input.

While in my experience such problems are more often than not a result
of faulty hardware (usually memory), IMHO the possibility of a GCC bug
that causes the same pattern cannot be easily dismissed.