In message <20000728204309.B14AC2A22@orchard.arlington.ma.us>
Bill Sommerfeld writes:

>> Doesn't that make for yet more very-low-payback nuisance changes
>> when importing kernel source which originates elsewhere, and
>> tracking changes from elsewhere?  (For new netbsd-originated code,
>> sure.)
>
>Not when it eliminates an entire class of buffer overrun bugs.

Which happen how often, again?  Whats the cost in fixing them,
compared to the cost of maintaining yet-another-gratuitious change?