On 09-Jul-2015 00:26, Martin Nowak wrote:
On 07/08/2015 08:06 PM, Yuxuan Shui wrote:
switch(x.toHash()){
case a.toHash():
if(x!=a) goto default;
...;
break;
case b.toHash():
if(x!=b) goto default;
...;
break;
default: break;
}
(a and b are known at compile time here, and so are their hash values.
Hash collisions between e.g. a and b would need to be dealt with of
course.)
Yes, this is what I meant.
Very interesting. How would you deal with collisions?
You can already do some nice things during compile time, e.g. creating
perfect hash functions and lookup table, but it doesn't seem to be
possible to achieve what you suggest.
In the worst case just if-else chain collided items per hash value?
Given that in this case it may use the full range of size_t I won't
expect high collision rates.
--
Dmitry Olshansky