If I did std::hash
using libstdc++
and then did one on the upcoming C++11
VS 2012 library - would they match?
I assume that hash implementations are not part of the C++ specification and can vary by distribution?
If I did std::hash
using libstdc++
and then did one on the upcoming C++11
VS 2012 library - would they match?
I assume that hash implementations are not part of the C++ specification and can vary by distribution?
The standard only says this:
20.8.12 Class template hash The unordered associative containers defined in 23.5 use specializations of the class template hash as the default hash function. For all object types Key for which there exists a specialization hash, theinstantiation hash shall:
- satisfy the Hash requirements (17.6.3.4), with Key as the function call argument type, the DefaultConstructible requirements (Table 19), the CopyAssignable requirements (Table 23),
- be swappable (17.6.3.2) for lvalues,
- provide two nested types result_type and argument_type which shall be synonyms for size_t and Key, respectively,
- satisfy the requirement that if k1 == k2 is true, h(k1) == h(k2) is also true, where h is an object of type hash and k1 and k2 are objects of type Key.
in 17.6.3.4, this is of most importance (table 26):
Shall not throw exceptions. The value returned shall depend only on the argument k. [ Note: Thus all evaluations of the expression h(k) with the same value for k yield the same result. — end note ] [ Note: For two different values t1 and t2, the probability that h(t1) and h(t2) compare equal should be very small, approaching 1.0 / numeric_- limits::max(). — end note ]
So in general, no, the computation itself is not defined and the result is not required to be consistent over implementations. For that matter, even two different versions of the same library may give different results.
No, this is not guaranteed. std::hash
only has to respect the following conditions:
- Accepts a single parameter of type Key.
- Returns a value of type size_t that represents the hash value of the parameter.
- Does not throw exceptions when called.
- For two parameters k1 and k2 that are equal, std::hash()(k1) == std::hash()(k2).
- For two different parameters k1 and k2 that are not equal, the probability that std::hash()(k1) == std::hash()(k2) should be very small, approaching 1.0/std::numeric_limits::max().
The standard only says this:
20.8.12 Class template hash The unordered associative containers defined in 23.5 use specializations of the class template hash as the default hash function. For all object types Key for which there exists a specialization hash, theinstantiation hash shall:
- satisfy the Hash requirements (17.6.3.4), with Key as the function call argument type, the DefaultConstructible requirements (Table 19), the CopyAssignable requirements (Table 23),
- be swappable (17.6.3.2) for lvalues,
- provide two nested types result_type and argument_type which shall be synonyms for size_t and Key, respectively,
- satisfy the requirement that if k1 == k2 is true, h(k1) == h(k2) is also true, where h is an object of type hash and k1 and k2 are objects of type Key.
in 17.6.3.4, this is of most importance (table 26):
Shall not throw exceptions. The value returned shall depend only on the argument k. [ Note: Thus all evaluations of the expression h(k) with the same value for k yield the same result. — end note ] [ Note: For two different values t1 and t2, the probability that h(t1) and h(t2) compare equal should be very small, approaching 1.0 / numeric_- limits::max(). — end note ]
So in general, no, the computation itself is not defined and the result is not required to be consistent over implementations. For that matter, even two different versions of the same library may give different results.
The requirements (17.6.3.4 Hash requirements [hash.requirements]) on the value returned by a Hash
function are:
Table 26 — Hash requirements [hash]
The value returned shall depend only on the argument
k
. [ Note: Thus all evaluations of the expressionh(k)
with the same value fork
yield the same result. —end note ] [ Note: For two different valuest1
andt2
, the probability thath(t1)
andh(t2)
compare equal should be very small, approaching1.0 / numeric_limits<size_t>::max()
. —end note ]
In practice, it's quite common that for integral types std::hash(k)
would equal k
, as that is the simplest possible implementation conformant with the standard. For other types, anything is possible.
© 2022 - 2024 — McMap. All rights reserved.
std::hash
are most likely shipped in header-only libraries, too, so you're getting the entire package anyway. – Sy17.6.3.4
but just wanted to be sure. There are three answers that are identical - what is the standard procedure that I should follow to accept the correct answer? – Coxcombry