> But it has also happened often to me that an old bug I was trying to fix was simply gone, either because the error mode has gone away, or the feature that the bug related to has been removed.
And this is why groups like the OpenBSD team and the Linux kernel team are so eager to remove support for old hardware. Removing support for the 80386 cleaned up code which no longer had to have special cases to handle opcodes and features the 80386 doesn't have but later x86 processors do.
Sometimes, yes. But it has also happened often to me that an old bug I was trying to fix was simply gone, either because the error mode has gone away, or the feature that the bug related to has been removed.