Gitiles
Code Review Sign In
gerrit-public.fairphone.software / platform / external / clang / 64f857b4efe543026334dbe99c5c22a4b40b517f / . / docs / LanguageExtensions.html
blob: dc70c696d2215fdac1eb67a859f892a7a9b361e2 [file] [log] [blame]
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
3<!-- Material used from: HTML 4.01 specs: http://www.w3.org/TR/html401/ -->
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]4<html>
5<head>
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]6 <META http-equiv="Content-Type" content="text/html; charset=ISO-8859-1">
7 <title>Clang LanguageExtensions</title>
Sean Hunt64f857b2011-06-23 01:22:53 +0000[diff] [blame^]8 <link type="text/css" rel="stylesheet" href="../menu.css">
9 <link type="text/css" rel="stylesheet" href="../content.css">
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]10 <style type="text/css">
11 td {
12 vertical-align: top;
13 }
14 </style>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]15</head>
16<body>
17
18<!--#include virtual="../menu.html.incl"-->
19
20<div id="content">
21
22<h1>Clang Language Extensions</h1>
23
24<ul>
25<li><a href="#intro">Introduction</a></li>
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]26<li><a href="#feature_check">Feature Checking Macros</a></li>
John Thompson92bd8c72009-11-02 22:28:12 +0000[diff] [blame]27<li><a href="#has_include">Include File Checking Macros</a></li>
Chris Lattner81edc9f2009-04-13 02:45:46 +0000[diff] [blame]28<li><a href="#builtinmacros">Builtin Macros</a></li>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]29<li><a href="#vectors">Vectors and Extended Vectors</a></li>
John McCall48209082010-11-08 19:48:17 +0000[diff] [blame]30<li><a href="#deprecated">Messages on <tt>deprecated</tt> and <tt>unavailable</tt> attributes</a></li>
31<li><a href="#attributes-on-enumerators">Attributes on enumerators</a></li>
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]32<li><a href="#checking_language_features">Checks for Standard Language Features</a>
Ted Kremenek22c34102009-12-03 02:05:57 +0000[diff] [blame]33 <ul>
34 <li><a href="#cxx_exceptions">C++ exceptions</a></li>
35 <li><a href="#cxx_rtti">C++ RTTI</a></li>
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]36 </ul></li>
37<li><a href="#checking_upcoming_features">Checks for Upcoming Standard Language Features</a>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]38 <ul>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]39 <li><a href="#cxx0x">C++0x</a>
40 <ul>
41 <li><a href="#cxx_decltype">C++0x <tt>decltype()</tt></a></li>
42 <li><a href="#cxx_access_control_sfinae">C++0x SFINAE includes access control</a></li>
43 <li><a href="#cxx_alias_templates">C++0x alias templates</a></li>
44 <li><a href="#cxx_attributes">C++0x attributes</a></li>
45 <li><a href="#cxx_default_function_template_args">C++0x default template arguments in function templates</a></li>
46 <li><a href="#cxx_deleted_functions">C++0x deleted functions</a></li>
47 <li><a href="#cxx_lambdas">C++0x lambdas</a></li>
48 <li><a href="#cxx_nullptr">C++0x nullptr</a></li>
49 <li><a href="#cxx_override_control">C++0x override control</a></li>
50 <li><a href="#cxx_range_for">C++0x range-based for loop</a></li>
51 <li><a href="#cxx_rvalue_references">C++0x rvalue references</a></li>
52 <li><a href="#cxx_reference_qualified_functions">C++0x reference-qualified functions</a></li>
53 <li><a href="#cxx_static_assert">C++0x <tt>static_assert()</tt></a></li>
54 <li><a href="#cxx_auto_type">C++0x type inference</a></li>
55 <li><a href="#cxx_variadic_templates">C++0x variadic templates</a></li>
56 <li><a href="#cxx_inline_namespaces">C++0x inline namespaces</a></li>
57 <li><a href="#cxx_strong_enums">C++0x strongly-typed enumerations</a></li>
58 <li><a href="#cxx_trailing_return">C++0x trailing return type</a></li>
59 <li><a href="#cxx_noexcept">C++0x noexcept specification</a></li>
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]60 </ul></li>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]61 <li><a href="#c1x">C1X</a>
62 <ul>
63 <li><a href="#c_generic_selections">C1X generic selections</a></li>
64 <li><a href="#c_static_assert">C1X <tt>_Static_assert()</tt></a></li>
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]65 </ul></li>
66 </ul> </li>
Douglas Gregorafdf1372011-02-03 21:57:35 +0000[diff] [blame]67<li><a href="#checking_type_traits">Checks for Type Traits</a></li>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]68<li><a href="#blocks">Blocks</a></li>
Douglas Gregor926df6c2011-06-11 01:09:30 +0000[diff] [blame]69<li><a href="#objc_features">Objective-C Features</a>
70 <ul>
71 <li><a href="#objc_instancetype">Related result types</a></li>
John McCallf85e1932011-06-15 23:02:42 +0000[diff] [blame]72 <li><a href="#objc_arc">Automatic reference counting</a></li>
Douglas Gregor926df6c2011-06-11 01:09:30 +0000[diff] [blame]73 </ul>
74</li>
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]75<li><a href="#overloading-in-c">Function Overloading in C</a></li>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]76<li><a href="#builtins">Builtin Functions</a>
77 <ul>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]78 <li><a href="#__builtin_shufflevector">__builtin_shufflevector</a></li>
Chris Lattner21190d52009-09-21 03:09:59 +0000[diff] [blame]79 <li><a href="#__builtin_unreachable">__builtin_unreachable</a></li>
Chris Lattner23aa9c82011-04-09 03:57:26 +0000[diff] [blame]80 <li><a href="#__sync_swap">__sync_swap</a></li>
Douglas Gregorafdf1372011-02-03 21:57:35 +0000[diff] [blame]81 </ul>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]82</li>
Chris Lattner1177f912009-04-09 19:58:15 +0000[diff] [blame]83<li><a href="#targetspecific">Target-Specific Extensions</a>
84 <ul>
85 <li><a href="#x86-specific">X86/X86-64 Language Extensions</a></li>
86 </ul>
87</li>
John McCall87494012011-03-18 03:51:49 +0000[diff] [blame]88<li><a href="#analyzerspecific">Static Analysis-Specific Extensions</a></li>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]89</ul>
90
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]91<!-- ======================================================================= -->
92<h2 id="intro">Introduction</h2>
93<!-- ======================================================================= -->
94
95<p>This document describes the language extensions provided by Clang. In
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]96addition to the language extensions listed here, Clang aims to support a broad
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]97range of GCC extensions. Please see the <a
98href="http://gcc.gnu.org/onlinedocs/gcc/C-Extensions.html">GCC manual</a> for
99more information on these extensions.</p>
100
101<!-- ======================================================================= -->
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]102<h2 id="feature_check">Feature Checking Macros</h2>
103<!-- ======================================================================= -->
104
105<p>Language extensions can be very useful, but only if you know you can depend
Chris Lattnerc70e1932011-03-21 16:25:11 +0000[diff] [blame]106on them. In order to allow fine-grain features checks, we support three builtin
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]107function-like macros. This allows you to directly test for a feature in your
108code without having to resort to something like autoconf or fragile "compiler
109version checks".</p>
110
111<!-- ======================================================================= -->
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]112<h3><a name="__has_builtin">__has_builtin</a></h3>
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]113<!-- ======================================================================= -->
114
115<p>This function-like macro takes a single identifier argument that is the name
116of a builtin function. It evaluates to 1 if the builtin is supported or 0 if
117not. It can be used like this:</p>
118
119<blockquote>
120<pre>
121#ifndef __has_builtin // Optional of course.
122 #define __has_builtin(x) 0 // Compatibility with non-clang compilers.
123#endif
124
125...
126#if __has_builtin(__builtin_trap)
127 __builtin_trap();
128#else
129 abort();
130#endif
131...
132</pre>
133</blockquote>
134
135
136<!-- ======================================================================= -->
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]137<h3><a name="__has_feature_extension"> __has_feature and __has_extension</a></h3>
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]138<!-- ======================================================================= -->
139
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]140<p>These function-like macros take a single identifier argument that is the
141name of a feature. <code>__has_feature</code> evaluates to 1 if the feature
142is both supported by Clang and standardized in the current language standard
143or 0 if not (but see <a href="#has_feature_back_compat">below</a>), while
144<code>__has_extension</code> evaluates to 1 if the feature is supported by
145Clang in the current language (either as a language extension or a standard
146language feature) or 0 if not. They can be used like this:</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]147
148<blockquote>
149<pre>
150#ifndef __has_feature // Optional of course.
151 #define __has_feature(x) 0 // Compatibility with non-clang compilers.
152#endif
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]153#ifndef __has_extension
154 #define __has_extension __has_feature // Compatibility with pre-3.0 compilers.
155#endif
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]156
157...
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]158#if __has_feature(cxx_rvalue_references)
159// This code will only be compiled with the -std=c++0x and -std=gnu++0x
160// options, because rvalue references are only standardized in C++0x.
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]161#endif
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]162
163#if __has_extension(cxx_rvalue_references)
164// This code will be compiled with the -std=c++0x, -std=gnu++0x, -std=c++98
165// and -std=gnu++98 options, because rvalue references are supported as a
166// language extension in C++98.
167#endif
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]168</pre>
169</blockquote>
170
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]171<p id="has_feature_back_compat">For backwards compatibility reasons,
172<code>__has_feature</code> can also be used to test for support for
173non-standardized features, i.e. features not prefixed <code>c_</code>,
174<code>cxx_</code> or <code>objc_</code>.</p>
175
176<p>If the <code>-pedantic-errors</code> option is given,
177<code>__has_extension</code> is equivalent to <code>__has_feature</code>.</p>
178
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]179<p>The feature tag is described along with the language feature below.</p>
180
John Thompson92bd8c72009-11-02 22:28:12 +0000[diff] [blame]181<!-- ======================================================================= -->
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]182<h3><a name="__has_attribute">__has_attribute</a></h3>
Anders Carlssoncae50952010-10-20 02:31:43 +0000[diff] [blame]183<!-- ======================================================================= -->
184
185<p>This function-like macro takes a single identifier argument that is the name
186of an attribute. It evaluates to 1 if the attribute is supported or 0 if not. It
187can be used like this:</p>
188
189<blockquote>
190<pre>
191#ifndef __has_attribute // Optional of course.
192 #define __has_attribute(x) 0 // Compatibility with non-clang compilers.
193#endif
194
195...
Anders Carlsson961003d2011-01-24 03:54:51 +0000[diff] [blame]196#if __has_attribute(always_inline)
197#define ALWAYS_INLINE __attribute__((always_inline))
Anders Carlssoncae50952010-10-20 02:31:43 +0000[diff] [blame]198#else
Anders Carlsson961003d2011-01-24 03:54:51 +0000[diff] [blame]199#define ALWAYS_INLINE
Anders Carlssoncae50952010-10-20 02:31:43 +0000[diff] [blame]200#endif
201...
202</pre>
203</blockquote>
204
205<!-- ======================================================================= -->
John Thompson92bd8c72009-11-02 22:28:12 +0000[diff] [blame]206<h2 id="has_include">Include File Checking Macros</h2>
207<!-- ======================================================================= -->
208
209<p>Not all developments systems have the same include files.
210The <a href="#__has_include">__has_include</a> and
211<a href="#__has_include_next">__has_include_next</a> macros allow you to
212check for the existence of an include file before doing
213a possibly failing #include directive.</p>
214
215<!-- ======================================================================= -->
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]216<h3><a name="__has_include">__has_include</a></h3>
John Thompson92bd8c72009-11-02 22:28:12 +0000[diff] [blame]217<!-- ======================================================================= -->
218
219<p>This function-like macro takes a single file name string argument that
220is the name of an include file. It evaluates to 1 if the file can
221be found using the include paths, or 0 otherwise:</p>
222
223<blockquote>
224<pre>
225// Note the two possible file name string formats.
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]226#if __has_include("myinclude.h") &amp;&amp; __has_include(&lt;stdint.h&gt;)
John Thompson92bd8c72009-11-02 22:28:12 +0000[diff] [blame]227# include "myinclude.h"
228#endif
229
230// To avoid problem with non-clang compilers not having this macro.
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]231#if defined(__has_include) &amp;&amp; __has_include("myinclude.h")
John Thompson92bd8c72009-11-02 22:28:12 +0000[diff] [blame]232# include "myinclude.h"
233#endif
234</pre>
235</blockquote>
236
237<p>To test for this feature, use #if defined(__has_include).</p>
238
239<!-- ======================================================================= -->
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]240<h3><a name="__has_include_next">__has_include_next</a></h3>
John Thompson92bd8c72009-11-02 22:28:12 +0000[diff] [blame]241<!-- ======================================================================= -->
242
243<p>This function-like macro takes a single file name string argument that
244is the name of an include file. It is like __has_include except that it
245looks for the second instance of the given file found in the include
246paths. It evaluates to 1 if the second instance of the file can
247be found using the include paths, or 0 otherwise:</p>
248
249<blockquote>
250<pre>
251// Note the two possible file name string formats.
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]252#if __has_include_next("myinclude.h") &amp;&amp; __has_include_next(&lt;stdint.h&gt;)
John Thompson92bd8c72009-11-02 22:28:12 +0000[diff] [blame]253# include_next "myinclude.h"
254#endif
255
256// To avoid problem with non-clang compilers not having this macro.
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]257#if defined(__has_include_next) &amp;&amp; __has_include_next("myinclude.h")
John Thompson92bd8c72009-11-02 22:28:12 +0000[diff] [blame]258# include_next "myinclude.h"
259#endif
260</pre>
261</blockquote>
262
263<p>Note that __has_include_next, like the GNU extension
264#include_next directive, is intended for use in headers only,
265and will issue a warning if used in the top-level compilation
266file. A warning will also be issued if an absolute path
267is used in the file argument.</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]268
269<!-- ======================================================================= -->
Chris Lattner81edc9f2009-04-13 02:45:46 +0000[diff] [blame]270<h2 id="builtinmacros">Builtin Macros</h2>
271<!-- ======================================================================= -->
272
Douglas Gregor4290fbd2010-04-30 02:51:06 +0000[diff] [blame]273<dl>
274 <dt><code>__BASE_FILE__</code></dt>
275 <dd>Defined to a string that contains the name of the main input
276 file passed to Clang.</dd>
277
278 <dt><code>__COUNTER__</code></dt>
279 <dd>Defined to an integer value that starts at zero and is
280 incremented each time the <code>__COUNTER__</code> macro is
281 expanded.</dd>
282
283 <dt><code>__INCLUDE_LEVEL__</code></dt>
284 <dd>Defined to an integral value that is the include depth of the
285 file currently being translated. For the main file, this value is
286 zero.</dd>
287
288 <dt><code>__TIMESTAMP__</code></dt>
289 <dd>Defined to the date and time of the last modification of the
290 current source file.</dd>
291
292 <dt><code>__clang__</code></dt>
293 <dd>Defined when compiling with Clang</dd>
294
295 <dt><code>__clang_major__</code></dt>
296 <dd>Defined to the major version number of Clang (e.g., the 2 in
297 2.0.1).</dd>
298
299 <dt><code>__clang_minor__</code></dt>
300 <dd>Defined to the minor version number of Clang (e.g., the 0 in
301 2.0.1).</dd>
302
303 <dt><code>__clang_patchlevel__</code></dt>
304 <dd>Defined to the patch level of Clang (e.g., the 1 in 2.0.1).</dd>
305
306 <dt><code>__clang_version__</code></dt>
307 <dd>Defined to a string that captures the Clang version, including
308 the Subversion tag or revision number, e.g., "1.5 (trunk
309 102332)".</dd>
310</dl>
Chris Lattner81edc9f2009-04-13 02:45:46 +0000[diff] [blame]311
312<!-- ======================================================================= -->
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]313<h2 id="vectors">Vectors and Extended Vectors</h2>
314<!-- ======================================================================= -->
315
Owen Andersond2bf0cd2010-01-27 01:22:36 +0000[diff] [blame]316<p>Supports the GCC vector extensions, plus some stuff like V[1].</p>
317
318<p>Also supports <tt>ext_vector</tt>, which additionally support for V.xyzw
319syntax and other tidbits as seen in OpenCL. An example is:</p>
320
321<blockquote>
322<pre>
323typedef float float4 <b>__attribute__((ext_vector_type(4)))</b>;
324typedef float float2 <b>__attribute__((ext_vector_type(2)))</b>;
325
326float4 foo(float2 a, float2 b) {
327 float4 c;
328 c.xz = a;
329 c.yw = b;
330 return c;
331}
John McCall48209082010-11-08 19:48:17 +0000[diff] [blame]332</pre>
Owen Andersond2bf0cd2010-01-27 01:22:36 +0000[diff] [blame]333</blockquote>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]334
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]335<p>Query for this feature with __has_extension(attribute_ext_vector_type).</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]336
Owen Andersond2bf0cd2010-01-27 01:22:36 +0000[diff] [blame]337<p>See also <a href="#__builtin_shufflevector">__builtin_shufflevector</a>.</p>
338
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]339<!-- ======================================================================= -->
John McCall48209082010-11-08 19:48:17 +0000[diff] [blame]340<h2 id="deprecated">Messages on <tt>deprecated</tt> and <tt>unavailable</tt> Attributes</h2>
Fariborz Jahanianc784dc12010-10-06 23:12:32 +0000[diff] [blame]341<!-- ======================================================================= -->
342
John McCall48209082010-11-08 19:48:17 +0000[diff] [blame]343<p>An optional string message can be added to the <tt>deprecated</tt>
344and <tt>unavailable</tt> attributes. For example:</p>
Fariborz Jahanianc784dc12010-10-06 23:12:32 +0000[diff] [blame]345
John McCall48209082010-11-08 19:48:17 +0000[diff] [blame]346<blockquote>
Chris Lattner4836d6a2010-11-09 19:43:35 +0000[diff] [blame]347<pre>void explode(void) __attribute__((deprecated("extremely unsafe, use 'combust' instead!!!")));</pre>
John McCall48209082010-11-08 19:48:17 +0000[diff] [blame]348</blockquote>
349
350<p>If the deprecated or unavailable declaration is used, the message
351will be incorporated into the appropriate diagnostic:</p>
352
353<blockquote>
Chris Lattner4836d6a2010-11-09 19:43:35 +0000[diff] [blame]354<pre>harmless.c:4:3: warning: 'explode' is deprecated: extremely unsafe, use 'combust' instead!!! [-Wdeprecated-declarations]
John McCall48209082010-11-08 19:48:17 +0000[diff] [blame]355 explode();
356 ^</pre>
357</blockquote>
358
359<p>Query for this feature
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]360with <tt>__has_extension(attribute_deprecated_with_message)</tt>
361and <tt>__has_extension(attribute_unavailable_with_message)</tt>.</p>
John McCall48209082010-11-08 19:48:17 +0000[diff] [blame]362
363<!-- ======================================================================= -->
364<h2 id="attributes-on-enumerators">Attributes on Enumerators</h2>
365<!-- ======================================================================= -->
366
367<p>Clang allows attributes to be written on individual enumerators.
368This allows enumerators to be deprecated, made unavailable, etc. The
369attribute must appear after the enumerator name and before any
370initializer, like so:</p>
371
372<blockquote>
373<pre>enum OperationMode {
374 OM_Invalid,
375 OM_Normal,
376 OM_Terrified __attribute__((deprecated)),
377 OM_AbortOnError __attribute__((deprecated)) = 4
378};</pre>
379</blockquote>
380
381<p>Attributes on the <tt>enum</tt> declaration do not apply to
382individual enumerators.</p>
383
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]384<p>Query for this feature with <tt>__has_extension(enumerator_attributes)</tt>.</p>
Fariborz Jahanianc784dc12010-10-06 23:12:32 +0000[diff] [blame]385
386<!-- ======================================================================= -->
Ted Kremenek87774fd2009-12-03 02:04:01 +0000[diff] [blame]387<h2 id="checking_language_features">Checks for Standard Language Features</h2>
388<!-- ======================================================================= -->
389
390<p>The <tt>__has_feature</tt> macro can be used to query if certain standard language features are
391enabled. Those features are listed here.</p>
392
Ted Kremenek22c34102009-12-03 02:05:57 +0000[diff] [blame]393<h3 id="cxx_exceptions">C++ exceptions</h3>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000[diff] [blame]394
Ted Kremenek22c34102009-12-03 02:05:57 +0000[diff] [blame]395<p>Use <tt>__has_feature(cxx_exceptions)</tt> to determine if C++ exceptions have been enabled. For
Sean Hunt647ba1b2011-06-23 00:42:53 +0000[diff] [blame]396example, compiling code with <tt>-fexceptions</tt> enables C++ exceptions.</p>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000[diff] [blame]397
Ted Kremenek22c34102009-12-03 02:05:57 +0000[diff] [blame]398<h3 id="cxx_rtti">C++ RTTI</h3>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000[diff] [blame]399
Ted Kremenek0eb95602009-12-03 02:06:43 +0000[diff] [blame]400<p>Use <tt>__has_feature(cxx_rtti)</tt> to determine if C++ RTTI has been enabled. For example,
Ted Kremenek22c34102009-12-03 02:05:57 +0000[diff] [blame]401compiling code with <tt>-fno-rtti</tt> disables the use of RTTI.</p>
Ted Kremenek87774fd2009-12-03 02:04:01 +0000[diff] [blame]402
403<!-- ======================================================================= -->
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]404<h2 id="checking_upcoming_features">Checks for Upcoming Standard Language Features</h2>
405<!-- ======================================================================= -->
406
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]407<p>The <tt>__has_feature</tt> or <tt>__has_extension</tt> macros can be used
408to query if certain upcoming standard language features are enabled. Those
409features are listed here. Features that are not yet implemented will be
410noted.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]411
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]412<h3 id="cxx0x">C++0x</h3>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]413
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]414<p>The features listed below are slated for inclusion in the upcoming
415C++0x standard. As a result, all these features are enabled
416with the <tt>-std=c++0x</tt> option when compiling C++ code.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]417
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]418<h4 id="cxx_decltype">C++0x <tt>decltype()</tt></h4>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]419
420<p>Use <tt>__has_feature(cxx_decltype)</tt> or
421<tt>__has_extension(cxx_decltype)</tt> to determine if support for the
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]422<tt>decltype()</tt> specifier is enabled.</p>
423
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]424<h4 id="cxx_access_control_sfinae">C++0x SFINAE includes access control</h4>
Douglas Gregor7822ee32011-05-11 23:45:11 +0000[diff] [blame]425
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]426<p>Use <tt>__has_feature(cxx_access_control_sfinae)</tt> or <tt>__has_extension(cxx_access_control_sfinae)</tt> to determine whether access-control errors (e.g., calling a private constructor) are considered to be template argument deduction errors (aka SFINAE errors), per <a href="http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#1170">C++ DR1170</a>.</p>
Douglas Gregor7822ee32011-05-11 23:45:11 +0000[diff] [blame]427
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]428<h4 id="cxx_alias_templates">C++0x alias templates</h4>
Richard Smith3e4c6c42011-05-05 21:57:07 +0000[diff] [blame]429
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]430<p>Use <tt>__has_feature(cxx_alias_templates)</tt> or
431<tt>__has_extension(cxx_alias_templates)</tt> to determine if support for
Richard Smith3e4c6c42011-05-05 21:57:07 +0000[diff] [blame]432C++0x's alias declarations and alias templates is enabled.</p>
433
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]434<h4 id="cxx_attributes">C++0x attributes</h4>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]435
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]436<p>Use <tt>__has_feature(cxx_attributes)</tt> or
437<tt>__has_extension(cxx_attributes)</tt> to determine if support for attribute
438parsing with C++0x's square bracket notation is enabled.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]439
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]440<h4 id="cxx_default_function_template_args">C++0x default template arguments in function templates</h4>
Douglas Gregor07508002011-02-05 20:35:30 +0000[diff] [blame]441
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]442<p>Use <tt>__has_feature(cxx_default_function_template_args)</tt> or
443<tt>__has_extension(cxx_default_function_template_args)</tt> to determine
444if support for default template arguments in function templates is enabled.</p>
Douglas Gregor07508002011-02-05 20:35:30 +0000[diff] [blame]445
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]446<h4 id="cxx_deleted_functions">C++0x <tt>delete</tt>d functions</h4>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]447
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]448<p>Use <tt>__has_feature(cxx_deleted_functions)</tt> or
449<tt>__has_extension(cxx_deleted_functions)</tt> to determine if support for
Sebastian Redlf6c09772010-08-31 23:28:47 +0000[diff] [blame]450deleted function definitions (with <tt>= delete</tt>) is enabled.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]451
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]452<h4 id="cxx_lambdas">C++0x lambdas</h4>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]453
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]454<p>Use <tt>__has_feature(cxx_lambdas)</tt> or
455<tt>__has_extension(cxx_lambdas)</tt> to determine if support for lambdas
456is enabled. clang does not currently implement this feature.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]457
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]458<h4 id="cxx_nullptr">C++0x <tt>nullptr</tt></h4>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]459
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]460<p>Use <tt>__has_feature(cxx_nullptr)</tt> or
461<tt>__has_extension(cxx_nullptr)</tt> to determine if support for
Douglas Gregor84ee2ee2011-05-21 23:15:46 +0000[diff] [blame]462<tt>nullptr</tt> is enabled.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]463
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]464<h4 id="cxx_override_control">C++0x <tt>override control</tt></h4>
Anders Carlssonc8b9f792011-03-25 15:04:23 +0000[diff] [blame]465
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]466<p>Use <tt>__has_feature(cxx_override_control)</tt> or
467<tt>__has_extension(cxx_override_control)</tt> to determine if support for
Anders Carlssonc8b9f792011-03-25 15:04:23 +0000[diff] [blame]468the override control keywords is enabled.</p>
469
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]470<h4 id="cxx_reference_qualified_functions">C++0x reference-qualified functions</h4>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]471<p>Use <tt>__has_feature(cxx_reference_qualified_functions)</tt> or
472<tt>__has_extension(cxx_reference_qualified_functions)</tt> to determine
473if support for reference-qualified functions (e.g., member functions with
474<code>&amp;</code> or <code>&amp;&amp;</code> applied to <code>*this</code>)
475is enabled.</p>
Douglas Gregor56209ff2011-01-26 21:25:54 +0000[diff] [blame]476
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]477<h4 id="cxx_range_for">C++0x range-based <tt>for</tt> loop</h4>
Richard Smitha391a462011-04-15 15:14:40 +0000[diff] [blame]478
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]479<p>Use <tt>__has_feature(cxx_range_for)</tt> or
480<tt>__has_extension(cxx_range_for)</tt> to determine if support for the
481range-based for loop is enabled. </p>
Richard Smitha391a462011-04-15 15:14:40 +0000[diff] [blame]482
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]483<h4 id="cxx_rvalue_references">C++0x rvalue references</h4>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]484
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]485<p>Use <tt>__has_feature(cxx_rvalue_references)</tt> or
486<tt>__has_extension(cxx_rvalue_references)</tt> to determine if support for
Douglas Gregor56209ff2011-01-26 21:25:54 +0000[diff] [blame]487rvalue references is enabled. </p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]488
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]489<h4 id="cxx_static_assert">C++0x <tt>static_assert()</tt></h4>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]490
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]491<p>Use <tt>__has_feature(cxx_static_assert)</tt> or
492<tt>__has_extension(cxx_static_assert)</tt> to determine if support for
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]493compile-time assertions using <tt>static_assert</tt> is enabled.</p>
494
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]495<h4 id="cxx_auto_type">C++0x type inference</h4>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]496
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]497<p>Use <tt>__has_feature(cxx_auto_type)</tt> or
498<tt>__has_extension(cxx_auto_type)</tt> to determine C++0x type inference is
499supported using the <tt>auto</tt> specifier. If this is disabled, <tt>auto</tt>
500will instead be a storage class specifier, as in C or C++98.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]501
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]502<h4 id="cxx_variadic_templates">C++0x variadic templates</h4>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]503
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]504<p>Use <tt>__has_feature(cxx_variadic_templates)</tt> or
505<tt>__has_extension(cxx_variadic_templates)</tt> to determine if support
Douglas Gregor83d77812011-01-19 23:15:20 +0000[diff] [blame]506for variadic templates is enabled.</p>
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]507
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]508<h4 id="cxx_inline_namespaces">C++0x inline namespaces</h4>
Sebastian Redlf6c09772010-08-31 23:28:47 +0000[diff] [blame]509
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]510<p>Use <tt>__has_feature(cxx_inline_namespaces)</tt> or
511<tt>__has_extension(cxx_inline_namespaces)</tt> to determine if support for
Sebastian Redlf6c09772010-08-31 23:28:47 +0000[diff] [blame]512inline namespaces is enabled.</p>
513
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]514<h4 id="cxx_trailing_return">C++0x trailing return type</h4>
Douglas Gregordab60ad2010-10-01 18:44:50 +0000[diff] [blame]515
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]516<p>Use <tt>__has_feature(cxx_trailing_return)</tt> or
517<tt>__has_extension(cxx_trailing_return)</tt> to determine if support for the
518alternate function declaration syntax with trailing return type is enabled.</p>
Douglas Gregordab60ad2010-10-01 18:44:50 +0000[diff] [blame]519
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]520<h4 id="cxx_noexcept">C++0x noexcept</h4>
Sebastian Redl4561ecd2011-03-15 21:17:12 +0000[diff] [blame]521
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]522<p>Use <tt>__has_feature(cxx_noexcept)</tt> or
523<tt>__has_extension(cxx_noexcept)</tt> to determine if support for noexcept
524exception specifications is enabled.</p>
Sebastian Redl4561ecd2011-03-15 21:17:12 +0000[diff] [blame]525
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]526<h4 id="cxx_strong_enums">C++0x strongly typed enumerations</h4>
Douglas Gregor1274ccd2010-10-08 23:50:27 +0000[diff] [blame]527
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]528<p>Use <tt>__has_feature(cxx_strong_enums)</tt> or
529<tt>__has_extension(cxx_strong_enums)</tt> to determine if support for
Douglas Gregor1274ccd2010-10-08 23:50:27 +0000[diff] [blame]530strongly typed, scoped enumerations is enabled.</p>
531
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]532<h3 id="c1x">C1X</h3>
533
534<p>The features listed below are slated for inclusion in the upcoming
535C1X standard. As a result, all these features are enabled
536with the <tt>-std=c1x</tt> option when compiling C code.</p>
537
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]538<h4 id="c_generic_selections">C1X generic selections</h4>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]539
540<p>Use <tt>__has_feature(c_generic_selections)</tt> or
541<tt>__has_extension(c_generic_selections)</tt> to determine if support for
542generic selections is enabled.</p>
543
544<p>As an extension, the C1X generic selection expression is available in all
545languages supported by Clang. The syntax is the same as that given in the
546C1X draft standard.</p>
547
548<p>In C, type compatibility is decided according to the rules given in the
549appropriate standard, but in C++, which lacks the type compatibility rules
550used in C, types are considered compatible only if they are equivalent.</p>
551
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]552<h4 id="c_static_assert">C1X <tt>_Static_assert()</tt></h4>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]553
554<p>Use <tt>__has_feature(c_static_assert)</tt> or
555<tt>__has_extension(c_static_assert)</tt> to determine if support for
556compile-time assertions using <tt>_Static_assert</tt> is enabled.</p>
557
Sean Hunt4ef4c6b2010-01-13 08:31:49 +0000[diff] [blame]558<!-- ======================================================================= -->
Douglas Gregorafdf1372011-02-03 21:57:35 +0000[diff] [blame]559<h2 id="checking_type_traits">Checks for Type Traits</h2>
560<!-- ======================================================================= -->
561
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]562<p>Clang supports the <a href="http://gcc.gnu.org/onlinedocs/gcc/Type-Traits.html">GNU C++ type traits</a> and a subset of the <a href="http://msdn.microsoft.com/en-us/library/ms177194(v=VS.100).aspx">Microsoft Visual C++ Type traits</a>. For each supported type trait <code>__X</code>, <code>__has_extension(X)</code> indicates the presence of the type trait. For example:
Douglas Gregorafdf1372011-02-03 21:57:35 +0000[diff] [blame]563<blockquote>
564<pre>
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]565#if __has_extension(is_convertible_to)
Douglas Gregorafdf1372011-02-03 21:57:35 +0000[diff] [blame]566template&lt;typename From, typename To&gt;
567struct is_convertible_to {
568 static const bool value = __is_convertible_to(From, To);
569};
570#else
571// Emulate type trait
572#endif
573</pre>
574</blockquote>
575
576<p>The following type traits are supported by Clang:</p>
577<ul>
578 <li><code>__has_nothrow_assign</code> (GNU, Microsoft)</li>
579 <li><code>__has_nothrow_copy</code> (GNU, Microsoft)</li>
580 <li><code>__has_nothrow_constructor</code> (GNU, Microsoft)</li>
581 <li><code>__has_trivial_assign</code> (GNU, Microsoft)</li>
582 <li><code>__has_trivial_copy</code> (GNU, Microsoft)</li>
583 <li><code>__has_trivial_constructor</code> (GNU, Microsoft)</li>
584 <li><code>__has_trivial_destructor</code> (GNU, Microsoft)</li>
585 <li><code>__has_virtual_destructor</code> (GNU, Microsoft)</li>
586 <li><code>__is_abstract</code> (GNU, Microsoft)</li>
587 <li><code>__is_base_of</code> (GNU, Microsoft)</li>
588 <li><code>__is_class</code> (GNU, Microsoft)</li>
589 <li><code>__is_convertible_to</code> (Microsoft)</li>
590 <li><code>__is_empty</code> (GNU, Microsoft)</li>
591 <li><code>__is_enum</code> (GNU, Microsoft)</li>
592 <li><code>__is_pod</code> (GNU, Microsoft)</li>
593 <li><code>__is_polymorphic</code> (GNU, Microsoft)</li>
594 <li><code>__is_union</code> (GNU, Microsoft)</li>
595 <li><code>__is_literal(type)</code>: Determines whether the given type is a literal type</li>
596</ul>
597
598<!-- ======================================================================= -->
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]599<h2 id="blocks">Blocks</h2>
600<!-- ======================================================================= -->
601
Chris Lattnera7dbdf52009-03-09 07:03:22 +0000[diff] [blame]602<p>The syntax and high level language feature description is in <a
603href="BlockLanguageSpec.txt">BlockLanguageSpec.txt</a>. Implementation and ABI
604details for the clang implementation are in <a
Chris Lattner5d7650b2010-03-16 21:43:03 +0000[diff] [blame]605href="Block-ABI-Apple.txt">Block-ABI-Apple.txt</a>.</p>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]606
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]607
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]608<p>Query for this feature with __has_extension(blocks).</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]609
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]610<!-- ======================================================================= -->
Douglas Gregor926df6c2011-06-11 01:09:30 +0000[diff] [blame]611<h2 id="objc_features">Objective-C Features</h2>
612<!-- ======================================================================= -->
613
614<h3 id="objc_instancetype">Related result types</h3>
615
616<p>According to Cocoa conventions, Objective-C methods with certain names ("init", "alloc", etc.) always return objects that are an instance of the receiving class's type. Such methods are said to have a "related result type", meaning that a message send to one of these methods will have the same static type as an instance of the receiver class. For example, given the following classes:</p>
617
618<blockquote>
619<pre>
620@interface NSObject
621+ (id)alloc;
622- (id)init;
623@end
624
625@interface NSArray : NSObject
626@end
627</pre>
628</blockquote>
629
630<p>and this common initialization pattern</p>
631
632<blockquote>
633<pre>
634NSArray *array = [[NSArray alloc] init];
635</pre>
636</blockquote>
637
638<p>the type of the expression <code>[NSArray alloc]</code> is
639<code>NSArray*</code> because <code>alloc</code> implicitly has a
640related result type. Similarly, the type of the expression
641<code>[[NSArray alloc] init]</code> is <code>NSArray*</code>, since
642<code>init</code> has a related result type and its receiver is known
643to have the type <code>NSArray *</code>. If neither <code>alloc</code> nor <code>init</code> had a related result type, the expressions would have had type <code>id</code>, as declared in the method signature.</p>
644
645<p>To determine whether a method has a related result type, the first
646word in the camel-case selector (e.g., "init" in "initWithObjects") is
647considered, and the method will a related result type if its return
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]648type is compatible with the type of its class and if</p>
Douglas Gregor926df6c2011-06-11 01:09:30 +0000[diff] [blame]649
650<ul>
651
652 <li>the first word is "alloc" or "new", and the method is a class
653 method, or</li>
654
655 <li>the first word is "autorelease", "init", "retain", or "self",
656 and the method is an instance method.</li>
657
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]658</ul>
Douglas Gregor926df6c2011-06-11 01:09:30 +0000[diff] [blame]659
660<p>If a method with a related result type is overridden by a subclass
661method, the subclass method must also return a type that is compatible
662with the subclass type. For example:</p>
663
664<blockquote>
665<pre>
666@interface NSString : NSObject
667- (NSUnrelated *)init; // incorrect usage: NSUnrelated is not NSString or a superclass of NSString
668@end
669</pre>
670</blockquote>
671
672<p>Related result types only affect the type of a message send or
673property access via the given method. In all other respects, a method
674with a related result type is treated the same way as method without a
675related result type.</p>
676
677<!-- ======================================================================= -->
John McCallf85e1932011-06-15 23:02:42 +0000[diff] [blame]678<h2 id="objc_arc">Automatic reference counting </h2>
679<!-- ======================================================================= -->
680
681<p>Clang provides support for <a href="AutomaticReferenceCounting.html">automated reference counting</a> in Objective-C, which eliminates the need for manual retain/release/autorelease message sends. There are two feature macros associated with automatic reference counting: <code>__has_feature(objc_arc)</code> indicates the availability of automated reference counting in general, while <code>__has_feature(objc_arc_weak)</code> indicates that automated reference counting also includes support for <code>__weak</code> pointers to Objective-C objects.</p>
682
683<!-- ======================================================================= -->
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]684<h2 id="overloading-in-c">Function Overloading in C</h2>
685<!-- ======================================================================= -->
686
Chris Lattnerf161d412009-02-13 21:51:45 +0000[diff] [blame]687<p>Clang provides support for C++ function overloading in C. Function
688overloading in C is introduced using the <tt>overloadable</tt> attribute. For
689example, one might provide several overloaded versions of a <tt>tgsin</tt>
690function that invokes the appropriate standard function computing the sine of a
691value with <tt>float</tt>, <tt>double</tt>, or <tt>long double</tt>
692precision:</p>
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]693
694<blockquote>
695<pre>
696#include &lt;math.h&gt;
697float <b>__attribute__((overloadable))</b> tgsin(float x) { return sinf(x); }
698double <b>__attribute__((overloadable))</b> tgsin(double x) { return sin(x); }
699long double <b>__attribute__((overloadable))</b> tgsin(long double x) { return sinl(x); }
700</pre>
701</blockquote>
702
703<p>Given these declarations, one can call <tt>tgsin</tt> with a
704<tt>float</tt> value to receive a <tt>float</tt> result, with a
705<tt>double</tt> to receive a <tt>double</tt> result, etc. Function
706overloading in C follows the rules of C++ function overloading to pick
707the best overload given the call arguments, with a few C-specific
708semantics:</p>
709<ul>
710 <li>Conversion from <tt>float</tt> or <tt>double</tt> to <tt>long
711 double</tt> is ranked as a floating-point promotion (per C99) rather
712 than as a floating-point conversion (as in C++).</li>
713
714 <li>A conversion from a pointer of type <tt>T*</tt> to a pointer of type
715 <tt>U*</tt> is considered a pointer conversion (with conversion
716 rank) if <tt>T</tt> and <tt>U</tt> are compatible types.</li>
717
718 <li>A conversion from type <tt>T</tt> to a value of type <tt>U</tt>
719 is permitted if <tt>T</tt> and <tt>U</tt> are compatible types. This
720 conversion is given "conversion" rank.</li>
721</ul>
722
723<p>The declaration of <tt>overloadable</tt> functions is restricted to
724function declarations and definitions. Most importantly, if any
725function with a given name is given the <tt>overloadable</tt>
726attribute, then all function declarations and definitions with that
727name (and in that scope) must have the <tt>overloadable</tt>
Chris Lattnerf161d412009-02-13 21:51:45 +0000[diff] [blame]728attribute. This rule even applies to redeclarations of functions whose original
729declaration had the <tt>overloadable</tt> attribute, e.g.,</p>
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]730
731<blockquote>
732<pre>
733int f(int) __attribute__((overloadable));
734float f(float); <i>// error: declaration of "f" must have the "overloadable" attribute</i>
735
736int g(int) __attribute__((overloadable));
737int g(int) { } <i>// error: redeclaration of "g" must also have the "overloadable" attribute</i>
738</pre>
739</blockquote>
740
Douglas Gregor965acbb2009-02-18 07:07:28 +0000[diff] [blame]741<p>Functions marked <tt>overloadable</tt> must have
742prototypes. Therefore, the following code is ill-formed:</p>
743
744<blockquote>
745<pre>
746int h() __attribute__((overloadable)); <i>// error: h does not have a prototype</i>
747</pre>
748</blockquote>
749
750<p>However, <tt>overloadable</tt> functions are allowed to use a
751ellipsis even if there are no named parameters (as is permitted in C++). This feature is particularly useful when combined with the <tt>unavailable</tt> attribute:</p>
752
753<blockquote>
754<pre>
Chris Lattner02246802009-02-18 22:27:46 +0000[diff] [blame]755void honeypot(...) __attribute__((overloadable, unavailable)); <i>// calling me is an error</i>
Douglas Gregor965acbb2009-02-18 07:07:28 +0000[diff] [blame]756</pre>
757</blockquote>
758
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]759<p>Functions declared with the <tt>overloadable</tt> attribute have
760their names mangled according to the same rules as C++ function
761names. For example, the three <tt>tgsin</tt> functions in our
762motivating example get the mangled names <tt>_Z5tgsinf</tt>,
Chris Lattner71b48d62010-11-28 18:19:13 +0000[diff] [blame]763<tt>_Z5tgsind</tt>, and <tt>_Z5tgsine</tt>, respectively. There are two
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]764caveats to this use of name mangling:</p>
765
766<ul>
767
768 <li>Future versions of Clang may change the name mangling of
769 functions overloaded in C, so you should not depend on an specific
770 mangling. To be completely safe, we strongly urge the use of
771 <tt>static inline</tt> with <tt>overloadable</tt> functions.</li>
772
773 <li>The <tt>overloadable</tt> attribute has almost no meaning when
774 used in C++, because names will already be mangled and functions are
775 already overloadable. However, when an <tt>overloadable</tt>
776 function occurs within an <tt>extern "C"</tt> linkage specification,
777 it's name <i>will</i> be mangled in the same way as it would in
778 C.</li>
779</ul>
780
Peter Collingbournec1b5fa42011-05-13 20:54:45 +0000[diff] [blame]781<p>Query for this feature with __has_extension(attribute_overloadable).</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]782
783
Douglas Gregorcb54d432009-02-13 00:57:04 +0000[diff] [blame]784<!-- ======================================================================= -->
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]785<h2 id="builtins">Builtin Functions</h2>
786<!-- ======================================================================= -->
787
788<p>Clang supports a number of builtin library functions with the same syntax as
789GCC, including things like <tt>__builtin_nan</tt>,
790<tt>__builtin_constant_p</tt>, <tt>__builtin_choose_expr</tt>,
791<tt>__builtin_types_compatible_p</tt>, <tt>__sync_fetch_and_add</tt>, etc. In
792addition to the GCC builtins, Clang supports a number of builtins that GCC does
793not, which are listed here.</p>
794
795<p>Please note that Clang does not and will not support all of the GCC builtins
796for vector operations. Instead of using builtins, you should use the functions
797defined in target-specific header files like <tt>&lt;xmmintrin.h&gt;</tt>, which
798define portable wrappers for these. Many of the Clang versions of these
799functions are implemented directly in terms of <a href="#vectors">extended
800vector support</a> instead of builtins, in order to reduce the number of
801builtins that we need to implement.</p>
802
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]803<!-- ======================================================================= -->
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]804<h3><a name="__builtin_shufflevector">__builtin_shufflevector</a></h3>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]805<!-- ======================================================================= -->
806
Chris Lattneraad826b2009-09-16 18:56:12 +0000[diff] [blame]807<p><tt>__builtin_shufflevector</tt> is used to express generic vector
Chris Lattner6f72da52009-02-13 20:00:20 +0000[diff] [blame]808permutation/shuffle/swizzle operations. This builtin is also very important for
809the implementation of various target-specific header files like
810<tt>&lt;xmmintrin.h&gt;</tt>.
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]811</p>
812
813<p><b>Syntax:</b></p>
814
815<pre>
Chris Lattner6f72da52009-02-13 20:00:20 +0000[diff] [blame]816__builtin_shufflevector(vec1, vec2, index1, index2, ...)
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]817</pre>
818
819<p><b>Examples:</b></p>
820
821<pre>
Chris Lattner6f72da52009-02-13 20:00:20 +0000[diff] [blame]822 // Identity operation - return 4-element vector V1.
823 __builtin_shufflevector(V1, V1, 0, 1, 2, 3)
824
825 // "Splat" element 0 of V1 into a 4-element result.
826 __builtin_shufflevector(V1, V1, 0, 0, 0, 0)
827
828 // Reverse 4-element vector V1.
829 __builtin_shufflevector(V1, V1, 3, 2, 1, 0)
830
831 // Concatenate every other element of 4-element vectors V1 and V2.
832 __builtin_shufflevector(V1, V2, 0, 2, 4, 6)
833
834 // Concatenate every other element of 8-element vectors V1 and V2.
835 __builtin_shufflevector(V1, V2, 0, 2, 4, 6, 8, 10, 12, 14)
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]836</pre>
837
838<p><b>Description:</b></p>
839
Chris Lattner6f72da52009-02-13 20:00:20 +0000[diff] [blame]840<p>The first two arguments to __builtin_shufflevector are vectors that have the
841same element type. The remaining arguments are a list of integers that specify
842the elements indices of the first two vectors that should be extracted and
843returned in a new vector. These element indices are numbered sequentially
844starting with the first vector, continuing into the second vector. Thus, if
845vec1 is a 4-element vector, index 5 would refer to the second element of vec2.
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]846</p>
847
Chris Lattner6f72da52009-02-13 20:00:20 +0000[diff] [blame]848<p>The result of __builtin_shufflevector is a vector
849with the same element type as vec1/vec2 but that has an element count equal to
850the number of indices specified.
851</p>
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]852
Chris Lattner21190d52009-09-21 03:09:59 +0000[diff] [blame]853<p>Query for this feature with __has_builtin(__builtin_shufflevector).</p>
854
855<!-- ======================================================================= -->
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]856<h3><a name="__builtin_unreachable">__builtin_unreachable</a></h3>
Chris Lattner21190d52009-09-21 03:09:59 +0000[diff] [blame]857<!-- ======================================================================= -->
858
859<p><tt>__builtin_unreachable</tt> is used to indicate that a specific point in
860the program cannot be reached, even if the compiler might otherwise think it
861can. This is useful to improve optimization and eliminates certain warnings.
862For example, without the <tt>__builtin_unreachable</tt> in the example below,
863the compiler assumes that the inline asm can fall through and prints a "function
864declared 'noreturn' should not return" warning.
865</p>
866
867<p><b>Syntax:</b></p>
868
869<pre>
870__builtin_unreachable()
871</pre>
872
873<p><b>Example of Use:</b></p>
874
875<pre>
876void myabort(void) __attribute__((noreturn));
877void myabort(void) {
878 asm("int3");
879 __builtin_unreachable();
880}
881</pre>
882
883<p><b>Description:</b></p>
884
885<p>The __builtin_unreachable() builtin has completely undefined behavior. Since
886it has undefined behavior, it is a statement that it is never reached and the
887optimizer can take advantage of this to produce better code. This builtin takes
888no arguments and produces a void result.
889</p>
890
891<p>Query for this feature with __has_builtin(__builtin_unreachable).</p>
892
Chris Lattner23aa9c82011-04-09 03:57:26 +0000[diff] [blame]893<!-- ======================================================================= -->
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]894<h3><a name="__sync_swap">__sync_swap</a></h3>
Chris Lattner23aa9c82011-04-09 03:57:26 +0000[diff] [blame]895<!-- ======================================================================= -->
896
897<p><tt>__sync_swap</tt> is used to atomically swap integers or pointers in
898memory.
899</p>
900
901<p><b>Syntax:</b></p>
902
903<pre>
904<i>type</i> __sync_swap(<i>type</i> *ptr, <i>type</i> value, ...)
905</pre>
906
907<p><b>Example of Use:</b></p>
908
909<pre>
Sean Hunt7e98b472011-06-23 01:21:01 +0000[diff] [blame]910int old_value = __sync_swap(&amp;value, new_value);
Chris Lattner23aa9c82011-04-09 03:57:26 +0000[diff] [blame]911</pre>
912
913<p><b>Description:</b></p>
914
915<p>The __sync_swap() builtin extends the existing __sync_*() family of atomic
916intrinsics to allow code to atomically swap the current value with the new
917value. More importantly, it helps developers write more efficient and correct
918code by avoiding expensive loops around __sync_bool_compare_and_swap() or
919relying on the platform specific implementation details of
920__sync_lock_test_and_set(). The __sync_swap() builtin is a full barrier.
921</p>
922
Chris Lattner21190d52009-09-21 03:09:59 +0000[diff] [blame]923
Chris Lattner1177f912009-04-09 19:58:15 +0000[diff] [blame]924<!-- ======================================================================= -->
925<h2 id="targetspecific">Target-Specific Extensions</h2>
926<!-- ======================================================================= -->
927
928<p>Clang supports some language features conditionally on some targets.</p>
929
930<!-- ======================================================================= -->
931<h3 id="x86-specific">X86/X86-64 Language Extensions</h3>
932<!-- ======================================================================= -->
933
934<p>The X86 backend has these language extensions:</p>
935
936<!-- ======================================================================= -->
937<h4 id="x86-gs-segment">Memory references off the GS segment</h4>
938<!-- ======================================================================= -->
939
940<p>Annotating a pointer with address space #256 causes it to be code generated
Chris Lattnera021e7c2009-05-05 18:54:47 +0000[diff] [blame]941relative to the X86 GS segment register, and address space #257 causes it to be
942relative to the X86 FS segment. Note that this is a very very low-level
943feature that should only be used if you know what you're doing (for example in
944an OS kernel).</p>
Chris Lattner1177f912009-04-09 19:58:15 +0000[diff] [blame]945
946<p>Here is an example:</p>
947
948<pre>
949#define GS_RELATIVE __attribute__((address_space(256)))
950int foo(int GS_RELATIVE *P) {
951 return *P;
952}
953</pre>
954
955<p>Which compiles to (on X86-32):</p>
956
957<pre>
958_foo:
959 movl 4(%esp), %eax
960 movl %gs:(%eax), %eax
961 ret
962</pre>
963
Ted Kremeneked869312009-04-10 05:03:33 +0000[diff] [blame]964<!-- ======================================================================= -->
965<h2 id="analyzerspecific">Static Analysis-Specific Extensions</h2>
966<!-- ======================================================================= -->
967
968<p>Clang supports additional attributes that are useful for documenting program
969invariants and rules for static analysis tools. The extensions documented here
970are used by the <a
971href="http://clang.llvm.org/StaticAnalysis.html">path-sensitive static analyzer
972engine</a> that is part of Clang's Analysis library.</p>
973
John McCall87494012011-03-18 03:51:49 +0000[diff] [blame]974<h3 id="attr_analyzer_noreturn">The <tt>analyzer_noreturn</tt> attribute</h3>
Ted Kremeneked869312009-04-10 05:03:33 +0000[diff] [blame]975
976<p>Clang's static analysis engine understands the standard <tt>noreturn</tt>
Ted Kremenek4df21142009-04-10 05:04:22 +0000[diff] [blame]977attribute. This attribute, which is typically affixed to a function prototype,
978indicates that a call to a given function never returns. Function prototypes for
979common functions like <tt>exit</tt> are typically annotated with this attribute,
980as well as a variety of common assertion handlers. Users can educate the static
981analyzer about their own custom assertion handles (thus cutting down on false
982positives due to false paths) by marking their own &quot;panic&quot; functions
983with this attribute.</p>
Ted Kremeneked869312009-04-10 05:03:33 +0000[diff] [blame]984
985<p>While useful, <tt>noreturn</tt> is not applicable in all cases. Sometimes
Nick Lewycky625b5862009-06-14 04:08:08 +0000[diff] [blame]986there are special functions that for all intents and purposes should be
987considered panic functions (i.e., they are only called when an internal program
988error occurs) but may actually return so that the program can fail gracefully.
989The <tt>analyzer_noreturn</tt> attribute allows one to annotate such functions
990as being interpreted as &quot;no return&quot; functions by the analyzer (thus
Chris Lattner28935892009-04-10 05:54:56 +0000[diff] [blame]991pruning bogus paths) but will not affect compilation (as in the case of
Ted Kremeneked869312009-04-10 05:03:33 +0000[diff] [blame]992<tt>noreturn</tt>).</p>
993
994<p><b>Usage</b>: The <tt>analyzer_noreturn</tt> attribute can be placed in the
Chris Lattner28935892009-04-10 05:54:56 +0000[diff] [blame]995same places where the <tt>noreturn</tt> attribute can be placed. It is commonly
Ted Kremeneked869312009-04-10 05:03:33 +0000[diff] [blame]996placed at the end of function prototypes:</p>
997
998<pre>
999 void foo() <b>__attribute__((analyzer_noreturn))</b>;
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]1000</pre>
1001
John McCall87494012011-03-18 03:51:49 +0000[diff] [blame]1002<p>Query for this feature with
1003<tt>__has_attribute(analyzer_noreturn)</tt>.</p>
Chris Lattner148772a2009-06-13 07:13:28 +0000[diff] [blame]1004
John McCall87494012011-03-18 03:51:49 +0000[diff] [blame]1005<h3 id="attr_method_family">The <tt>objc_method_family</tt> attribute</h3>
1006
1007<p>Many methods in Objective-C have conventional meanings determined
1008by their selectors. For the purposes of static analysis, it is
1009sometimes useful to be able to mark a method as having a particular
1010conventional meaning despite not having the right selector, or as not
1011having the conventional meaning that its selector would suggest.
1012For these use cases, we provide an attribute to specifically describe
1013the <q>method family</q> that a method belongs to.</p>
1014
1015<p><b>Usage</b>: <tt>__attribute__((objc_method_family(X)))</tt>,
1016where <tt>X</tt> is one of <tt>none</tt>, <tt>alloc</tt>, <tt>copy</tt>,
1017<tt>init</tt>, <tt>mutableCopy</tt>, or <tt>new</tt>. This attribute
1018can only be placed at the end of a method declaration:</p>
1019
1020<pre>
1021 - (NSString*) initMyStringValue <b>__attribute__((objc_method_family(none)))</b>;
1022</pre>
1023
1024<p>Users who do not wish to change the conventional meaning of a
1025method, and who merely want to document its non-standard retain and
1026release semantics, should use the
1027<a href="#attr_retain_release">retaining behavior attributes</a>
1028described below.</p>
1029
1030<p>Query for this feature with
1031<tt>__has_attribute(objc_method_family)</tt>.</p>
1032
1033<h3 id="attr_retain_release">Objective-C retaining behavior attributes</h3>
John McCall630b7ae2011-01-25 04:26:21 +0000[diff] [blame]1034
1035<p>In Objective-C, functions and methods are generally assumed to take
1036and return objects with +0 retain counts, with some exceptions for
1037special methods like <tt>+alloc</tt> and <tt>init</tt>. However,
1038there are exceptions, and so Clang provides attributes to allow these
1039exceptions to be documented, which helps the analyzer find leaks (and
John McCall87494012011-03-18 03:51:49 +0000[diff] [blame]1040ignore non-leaks). Some exceptions may be better described using
1041the <a href="#attr_method_family"><tt>objc_method_family</tt></a>
1042attribute instead.</p>
John McCall630b7ae2011-01-25 04:26:21 +0000[diff] [blame]1043
1044<p><b>Usage</b>: The <tt>ns_returns_retained</tt>, <tt>ns_returns_not_retained</tt>,
1045<tt>ns_returns_autoreleased</tt>, <tt>cf_returns_retained</tt>,
1046and <tt>cf_returns_not_retained</tt> attributes can be placed on
1047methods and functions that return Objective-C or CoreFoundation
1048objects. They are commonly placed at the end of a function prototype
1049or method declaration:</p>
1050
1051<pre>
1052 id foo() <b>__attribute__((ns_returns_retained))</b>;
1053
1054 - (NSString*) bar: (int) x <b>__attribute__((ns_returns_retained))</b>;
1055</pre>
1056
1057<p>The <tt>*_returns_retained</tt> attributes specify that the
1058returned object has a +1 retain count.
1059The <tt>*_returns_not_retained</tt> attributes specify that the return
1060object has a +0 retain count, even if the normal convention for its
1061selector would be +1. <tt>ns_returns_autoreleased</tt> specifies that the
1062returned object is +0, but is guaranteed to live at least as long as the
1063next flush of an autorelease pool.</p>
1064
1065<p><b>Usage</b>: The <tt>ns_consumed</tt> and <tt>cf_consumed</tt>
1066attributes can be placed on an parameter declaration; they specify
1067that the argument is expected to have a +1 retain count, which will be
1068balanced in some way by the function or method.
1069The <tt>ns_consumes_self</tt> attribute can only be placed on an
1070Objective-C method; it specifies that the method expects
1071its <tt>self</tt> parameter to have a +1 retain count, which it will
1072balance in some way.</p>
1073
1074<pre>
1075 void <b>foo(__attribute__((ns_consumed))</b> NSString *string);
1076
1077 - (void) bar <b>__attribute__((ns_consumes_self))</b>;
1078 - (void) baz: (id) <b>__attribute__((ns_consumed))</b> x;
1079</pre>
Ted Kremeneked869312009-04-10 05:03:33 +0000[diff] [blame]1080
John McCall87494012011-03-18 03:51:49 +0000[diff] [blame]1081<p>Query for these features with <tt>__has_attribute(ns_consumed)</tt>,
1082<tt>__has_attribute(ns_returns_retained)</tt>, etc.</p>
1083
Chris Lattner5ce933f2009-02-09 08:46:11 +0000[diff] [blame]1084</div>
1085</body>
1086</html>