defensive-coding-guide/en-US/CXX-Std.xml
2017-10-24 08:15:28 +00:00

202 lines
9.7 KiB
XML

<?xml version='1.0' encoding='utf-8' ?>
<!DOCTYPE section PUBLIC "-//OASIS//DTD DocBook XML V4.5//EN" "http://www.oasis-open.org/docbook/xml/4.5/docbookx.dtd" [
]>
<section id="sect-Defensive_Coding-CXX-Std">
<title>The C++ Standard Library</title>
<para>
The C++ standard library includes most of its C counterpart
by reference, see <xref linkend="sect-Defensive_Coding-C-Libc"/>.
</para>
<section id="sect-Defensive_Coding-CXX-Std-Functions">
<title>Functions That Are Difficult to Use</title>
<para>
This section collects functions and function templates which are
part of the standard library and are difficult to use.
</para>
<section id="sect-Defensive_Coding-CXX-Std-Functions-Unpaired_Iterators">
<title>Unpaired Iterators</title>
<para>
Functions which use output operators or iterators which do not
come in pairs (denoting ranges) cannot perform iterator range
checking.
(See <xref linkend="sect-Defensive_Coding-CXX-Std-Iterators"/>)
Function templates which involve output iterators are
particularly dangerous:
</para>
<itemizedlist>
<listitem><para><function>std::copy</function></para></listitem>
<listitem><para><function>std::copy_backward</function></para></listitem>
<listitem><para><function>std::copy_if</function></para></listitem>
<listitem><para><function>std::move</function> (three-argument variant)</para></listitem>
<listitem><para><function>std::move_backward</function></para></listitem>
<listitem><para><function>std::partition_copy_if</function></para></listitem>
<listitem><para><function>std::remove_copy</function></para></listitem>
<listitem><para><function>std::remove_copy_if</function></para></listitem>
<listitem><para><function>std::replace_copy</function></para></listitem>
<listitem><para><function>std::replace_copy_if</function></para></listitem>
<listitem><para><function>std::swap_ranges</function></para></listitem>
<listitem><para><function>std::transform</function></para></listitem>
</itemizedlist>
<para>
In addition, <function>std::copy_n</function>,
<function>std::fill_n</function> and
<function>std::generate_n</function> do not perform iterator
checking, either, but there is an explicit count which has to be
supplied by the caller, as opposed to an implicit length
indicator in the form of a pair of forward iterators.
</para>
<para>
These output-iterator-expecting functions should only be used
with unlimited-range output iterators, such as iterators
obtained with the <function>std::back_inserter</function>
function.
</para>
<para>
Other functions use single input or forward iterators, which can
read beyond the end of the input range if the caller is not careful:
</para>
<itemizedlist>
<listitem><para><function>std::equal</function></para></listitem>
<listitem><para><function>std::is_permutation</function></para></listitem>
<listitem><para><function>std::mismatch</function></para></listitem>
</itemizedlist>
</section>
</section>
<section id="sect-Defensive_Coding-CXX-Std-String">
<title>String Handling with <literal>std::string</literal></title>
<para>
The <literal>std::string</literal> class provides a convenient
way to handle strings. Unlike C strings,
<literal>std::string</literal> objects have an explicit length
(and can contain embedded NUL characters), and storage for its
characters is managed automatically. This section discusses
<literal>std::string</literal>, but these observations also
apply to other instances of the
<literal>std::basic_string</literal> template.
</para>
<para>
The pointer returned by the <function>data()</function> member
function does not necessarily point to a NUL-terminated string.
To obtain a C-compatible string pointer, use
<function>c_str()</function> instead, which adds the NUL
terminator.
</para>
<para>
The pointers returned by the <function>data()</function> and
<function>c_str()</function> functions and iterators are only
valid until certain events happen. It is required that the
exact <literal>std::string</literal> object still exists (even
if it was initially created as a copy of another string object).
Pointers and iterators are also invalidated when non-const
member functions are called, or functions with a non-const
reference parameter. The behavior of the GCC implementation
deviates from that required by the C++ standard if multiple
threads are present. In general, only the first call to a
non-const member function after a structural modification of the
string (such as appending a character) is invalidating, but this
also applies to member function such as the non-const version of
<function>begin()</function>, in violation of the C++ standard.
</para>
<para>
Particular care is necessary when invoking the
<function>c_str()</function> member function on a temporary
object. This is convenient for calling C functions, but the
pointer will turn invalid as soon as the temporary object is
destroyed, which generally happens when the outermost expression
enclosing the expression on which <function>c_str()</function>
is called completes evaluation. Passing the result of
<function>c_str()</function> to a function which does not store
or otherwise leak that pointer is safe, though.
</para>
<para>
Like with <literal>std::vector</literal> and
<literal>std::array</literal>, subscribing with
<literal>operator[]</literal> does not perform bounds checks.
Use the <function>at(size_type)</function> member function
instead. See <xref
linkend="sect-Defensive_Coding-CXX-Std-Subscript"/>.
Furthermore, accessing the terminating NUL character using
<literal>operator[]</literal> is not possible. (In some
implementations, the <literal>c_str()</literal> member function
writes the NUL character on demand.)
</para>
<para>
Never write to the pointers returned by
<function>data()</function> or <function>c_str()</function>
after casting away <literal>const</literal>. If you need a
C-style writable string, use a
<literal>std::vector&lt;char&gt;</literal> object and its
<function>data()</function> member function. In this case, you
have to explicitly add the terminating NUL character.
</para>
<para>
GCC's implementation of <literal>std::string</literal> is
currently based on reference counting. It is expected that a
future version will remove the reference counting, due to
performance and conformance issues. As a result, code that
implicitly assumes sharing by holding to pointers or iterators
for too long will break, resulting in run-time crashes or worse.
On the other hand, non-const iterator-returning functions will
no longer give other threads an opportunity for invalidating
existing iterators and pointers because iterator invalidation
does not depend on sharing of the internal character array
object anymore.
</para>
</section>
<section id="sect-Defensive_Coding-CXX-Std-Subscript">
<title>Containers and <literal>operator[]</literal></title>
<para>
Many sequence containers similar to <literal>std::vector</literal>
provide both <literal>operator[](size_type)</literal> and a
member function <literal>at(size_type)</literal>. This applies
to <literal>std::vector</literal> itself,
<literal>std::array</literal>, <literal>std::string</literal>
and other instances of <literal>std::basic_string</literal>.
</para>
<para>
<literal>operator[](size_type)</literal> is not required by the
standard to perform bounds checking (and the implementation in
GCC does not). In contrast, <literal>at(size_type)</literal>
must perform such a check. Therefore, in code which is not
performance-critical, you should prefer
<literal>at(size_type)</literal> over
<literal>operator[](size_type)</literal>, even though it is
slightly more verbose.
</para>
<para>
The <literal>front()</literal> and <literal>back()</literal>
member functions are undefined if a vector object is empty. You
can use <literal>vec.at(0)</literal> and
<literal>vec.at(vec.size() - 1)</literal> as checked
replacements. For an empty vector, <literal>data()</literal> is
defined; it returns an arbitrary pointer, but not necessarily
the NULL pointer.
</para>
</section>
<section id="sect-Defensive_Coding-CXX-Std-Iterators">
<title>Iterators</title>
<para>
Iterators do not perform any bounds checking. Therefore, all
functions that work on iterators should accept them in pairs,
denoting a range, and make sure that iterators are not moved
outside that range. For forward iterators and bidirectional
iterators, you need to check for equality before moving the
first or last iterator in the range. For random-access
iterators, you need to compute the difference before adding or
subtracting an offset. It is not possible to perform the
operation and check for an invalid operator afterwards.
</para>
<para>
Output iterators cannot be compared for equality. Therefore, it
is impossible to write code that detects that it has been
supplied an output area that is too small, and their use should
be avoided.
</para>
<para>
These issues make some of the standard library functions
difficult to use correctly, see <xref
linkend="sect-Defensive_Coding-CXX-Std-Functions-Unpaired_Iterators"/>.
</para>
</section>
</section>