...one of the most highly
regarded and expertly designed C++ library projects in the
world.
— Herb Sutter and Andrei
Alexandrescu, C++
Coding Standards
The header file 'equal.hpp' contains two variants of a the stl algorithm
equal
. The algorithm tests
to see if two sequences contain equal values;
Before (the proposed) C++14 the algorithm std::equal
took three iterators and an optional comparison predicate. The first two
iterators [first1, last1)
defined a sequence, and the second one
first2
defined the start
of the second sequence. The second sequence was assumed to be the same
length as the first.
In C++14, two new variants were introduced, taking four iterators and an
optional comparison predicate. The four iterators define two sequences
[first1, last1)
and [first2, last2)
explicitly, rather than defining the second one implicitly. This leads
to correct answers in more cases (and avoid undefined behavior in others).
Consider the two sequences:
auto seq1 = { 0, 1, 2 }; auto seq2 = { 0, 1, 2, 3, 4 }; std::equal ( seq1.begin (), seq1.end (), seq2.begin ()); // true std::equal ( seq2.begin (), seq2.end (), seq1.begin ()); // Undefined behavior std::equal ( seq1.begin (), seq1.end (), seq2.begin (), seq2.end ()); // false
You can argue that true
is
the correct answer in the first case, even though the sequences are not
the same. The first N entries in seq2
are the same as the entries in seq1
- but that's not all that's in seq2
.
But in the second case, the algorithm will read past the end of seq1
, resulting in undefined behavior
(large earthquake, incorrect results, pregnant cat, etc).
However, if the two sequences are specified completely, it's clear that they are not equal.
The function equal
returns
true if the two sequences compare equal; i.e, if each element in the sequence
compares equal to the corresponding element in the other sequence. One
version uses std::equal_to
to do the comparison; the other
lets the caller pass predicate to do the comparisons.
template <class InputIterator1, class InputIterator2> bool equal ( InputIterator1 first1, InputIterator1 last1, InputIterator2 first2, InputIterator2 last2 ); template <class InputIterator1, class InputIterator2, class BinaryPredicate> bool equal ( InputIterator1 first1, InputIterator1 last1, InputIterator2 first2, InputIterator2 last2, BinaryPredicate pred );
Given the container c1
containing { 0, 1,
2, 3, 14, 15 }
,
and c2
containing { 1,
2, 3 }
, then
equal ( c1.begin (), c1.end (), c2.begin (), c2.end ()) --> false equal ( c1.begin () + 1, c1.begin () + 3, c2.begin (), c2.end ()) --> true equal ( c1.end (), c1.end (), c2.end (), c2.end ()) --> true // empty sequences are alway equal to each other
equal
works on all iterators
except output iterators.
Both of the variants of equal
run in O(N) (linear) time; that is, they compare against
each element in the list once. If the sequence is found to be not equal
at any point, the routine will terminate immediately, without examining
the rest of the elements.
Both of the variants of equal
take their parameters by value and do not depend upon any global state.
Therefore, all the routines in this file provide the strong exception guarantee.
equal
is part of the C++14 standard. When C++14 standard library implementations
become available, the implementation from the standard library should
be used.
equal
returns true
for two empty ranges, no matter what predicate is passed to test against.
The header file 'mismatch.hpp' contains two variants of a the stl algorithm
mismatch
. The algorithm
finds the first point in two sequences where they do not match.
Before (the proposed) C++14 the algorithm std::mismatch
took three iterators and an optional comparison predicate. The first two
iterators [first1, last1)
defined a sequence, and the second one
first2
defined the start
of the second sequence. The second sequence was assumed to be the same
length as the first.
In C++14, two new variants were introduced, taking four iterators and an
optional comparison predicate. The four iterators define two sequences
[first1, last1)
and [first2, last2)
explicitly, rather than defining the second one implicitly. This leads
to correct answers in more cases (and avoid undefined behavior in others).
Consider the two sequences:
auto seq1 = { 0, 1, 2 }; auto seq2 = { 0, 1, 2, 3, 4 }; std::mismatch ( seq1.begin (), seq1.end (), seq2.begin ()); // <3, 3> std::mismatch ( seq2.begin (), seq2.end (), seq1.begin ()); // Undefined behavior std::mismatch ( seq1.begin (), seq1.end (), seq2.begin (), seq2.end ()); // <3, 3>
The first N entries in seq2
are the same as the entries in seq1
- but that's not all that's in seq2
.
In the second case, the algorithm will read past the end of seq1
, resulting in undefined behavior
(large earthquake, incorrect results, pregnant cat, etc).
However, if the two sequences are specified completely, it's clear that where the mismatch occurs.
The function mismatch
returns
a pair of iterators which denote the first mismatching elements in each
sequence. If the sequences match completely, mismatch
returns their end iterators. One version uses std::equal_to
to do the comparison; the other lets the caller pass predicate to do the
comparisons.
template <class InputIterator1, class InputIterator2> std::pair<InputIterator1, InputIterator2> mismatch ( InputIterator1 first1, InputIterator1 last1, InputIterator2 first2, InputIterator2 last2 ); template <class InputIterator1, class InputIterator2, class BinaryPredicate> std::pair<InputIterator1, InputIterator2> mismatch ( InputIterator1 first1, InputIterator1 last1, InputIterator2 first2, InputIterator2 last2, BinaryPredicate pred );
Given the container c1
containing { 0, 1,
2, 3, 14, 15 }
,
and c2
containing { 1,
2, 3 }
, then
mismatch ( c1.begin(), c1.end(), c2.begin(), c2.end()) --> <c1.begin(), c2.begin()> // first elements do not match mismatch ( c1.begin() + 1, c1.begin() + 4, c2.begin(), c2.end()) --> <c1.begin() + 4, c2.end ()> // all elements of `c2` match mismatch ( c1.end(), c1.end(), c2.end(), c2.end()) --> <c1.end(), c2.end()> // empty sequences don't match at the end.
mismatch
works on all iterators
except output iterators.
Both of the variants of mismatch
run in O(N) (linear) time; that is, they compare against
each element in the list once. If the sequence is found to be not equal
at any point, the routine will terminate immediately, without examining
the rest of the elements.
Both of the variants of mismatch
take their parameters by value and do not depend upon any global state.
Therefore, all the routines in this file provide the strong exception guarantee.
mismatch
is part of the C++14 standard. When C++14 standard library implementations
become available, the implementation from the standard library should
be used.