[Gnu-arch-users] Re: [GNU-arch-dev] Re: [ANNOUNCEMENT] /Arch/ embraces `git'

[Matthieu Moy]

Jacob Gorm Hansen <address@hidden> writes:

> That is a problem that can be fixed at the hardware-level, i.e. with
> error-correcting memory. 

This does not _fix_ the problem. It just reduces probability of memory
corruption by using redundancy. If you consider redundancy as a
solution for the memory corruption problem, then, you have to admit
than a probability of 10^-150 is not worth considering.

If you prefer another example, the probability of accidental collision
is probably also lower than the probability of all developers dying
suddenly and all their computers exploding with complete data loss.

(I hope you don't use any cryptographic software. All cryptographic
software are based on such low probabilities : If you try a secret key
at random, and if "by chance", you guess the right one, then you've
broken the encryption.)

> Using this argument to justify hash
> collisions is like trying to convince the cop who just pulled you
> over for speeding that he should go and chase some 'real criminals'
> instead.

Why? Do I have a probability of 10^-150 of being catched by a cop for
speeding? Your example should be "Using this argument to justify hash
collisions is like trying to convince the cop who hunts martians that
he should go and chase some 'real criminals' instead.". Then, the
probability are a bit more comparable ;-)

-- 
Matthieu