Re: incredibly fast

Frederic Bonroy

Bart Bailey a écrit :

> I can see where hash scanning would be quicker, but don't those hashes
> have to be calculated prior to the scan and how much time does that
> consume, not to mention the chance of a hash collision? (dead beef)

I suppose that if they do employ hashing, then they precompute the data
in their lab and include it in some data file which is read at runtime,
and I also suppose they go drink coffee while the collisions are being
dealt with. ;-)

I have no idea how it is implemented in practice and unfortunately the
good old times, when people competent enough to demistify this
occasionally appeared here, are long gone.