Hasher

A Hasher implements a hash algorithm. This is used to compute the small digests NuDB needs to effectively implement a hash table. NuDB provides the default implementation xxhasher, which is suitable for most use cases. For advanced applications, a user supplied hash function may be supplied which must meet these requirements.

In the table below:

X denotes a hasher class
a denotes a value of type X const
s denotes a value of type std::uint64_t
p denotes a value of type void const*
n denotes a value of type std::size_t

Table 2. Hasher requirements

operation	type	semantics, pre/post-conditions
`X a{s}`		`a` is constructed with a seed value integer. To achieve resistance from algorithmic complexity attacks, an implementation of Hasher should ensure that values returned from the hash function will be distinctly different for different values of `s` given the same inputs. If algorithmic complexity attack resistance is not a requirement, the seed may be ignored upon construction.
`a(p,n)`	`std::uint64_t`	Returns the digest of the memory `n` bytes in size and pointed to by `p`. `n` will never be zero. A good hash function will return values with these qualities: * Values are uniformly distributed in the full range * Values for the same input are distinctly different for different seeds * Small changes in the input produce unpredictable output values

operation

type

semantics, pre/post-conditions

X a{s}

a is constructed with a seed value integer. To achieve resistance from algorithmic complexity attacks, an implementation of Hasher should ensure that values returned from the hash function will be distinctly different for different values of s given the same inputs. If algorithmic complexity attack resistance is not a requirement, the seed may be ignored upon construction.

a(p,n)

std::uint64_t

Returns the digest of the memory n bytes in size and pointed to by p. n will never be zero. A good hash function will return values with these qualities:

* Values are uniformly distributed in the full range

* Values for the same input are distinctly different for different seeds

* Small changes in the input produce unpredictable output values