Gitiles
Code Review Sign In
gerrit-public.fairphone.software / platform / external / clang / a55e3ff9cbb5ccf338abd15d60970d9655e46d55 / . / www / compatibility.html
blob: de864eda4f7279850a4198cc621ac128b2322552 [file] [log] [blame]
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
3<html>
4<head>
Benjamin Kramer665a8dc2012-01-15 15:26:07 +0000[diff] [blame]5 <META http-equiv="Content-Type" content="text/html; charset=ISO-8859-1">
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]6 <title>Language Compatibility</title>
Benjamin Kramer665a8dc2012-01-15 15:26:07 +0000[diff] [blame]7 <link type="text/css" rel="stylesheet" href="menu.css">
8 <link type="text/css" rel="stylesheet" href="content.css">
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]9 <style type="text/css">
10</style>
11</head>
12<body>
13
14<!--#include virtual="menu.html.incl"-->
15
16<div id="content">
17
18<!-- ======================================================================= -->
19<h1>Language Compatibility</h1>
20<!-- ======================================================================= -->
21
22<p>Clang strives to both conform to current language standards (C99,
23 C++98) and also to implement many widely-used extensions available
24 in other compilers, so that most correct code will "just work" when
25 compiler with Clang. However, Clang is more strict than other
26 popular compilers, and may reject incorrect code that other
27 compilers allow. This page documents common compatibility and
28 portability issues with Clang to help you understand and fix the
29 problem in your code when Clang emits an error message.</p>
30
31<ul>
32 <li><a href="#c">C compatibility</a>
33 <ul>
34 <li><a href="#inline">C99 inline functions</a></li>
Chris Lattnera02d1832010-09-16 18:17:55 +0000[diff] [blame]35 <li><a href="#vector_builtins">"missing" vector __builtin functions</a></li>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]36 <li><a href="#lvalue-cast">Lvalue casts</a></li>
Daniel Dunbar5a410212010-09-02 21:35:16 +0000[diff] [blame]37 <li><a href="#blocks-in-protected-scope">Jumps to within <tt>__block</tt> variable scope</a></li>
Daniel Dunbar15952c92010-11-09 22:45:16 +0000[diff] [blame]38 <li><a href="#block-variable-initialization">Non-initialization of <tt>__block</tt> variables</a></li>
Chris Lattner9b743f42010-11-10 23:51:50 +0000[diff] [blame]39 <li><a href="#inline-asm">Inline assembly</a></li>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]40 </ul>
41 </li>
42 <li><a href="#objective-c">Objective-C compatibility</a>
43 <ul>
44 <li><a href="#super-cast">Cast of super</a></li>
45 <li><a href="#sizeof-interface">Size of interfaces</a></li>
Argyrios Kyrtzidis3b5b92a2010-09-13 17:48:07 +0000[diff] [blame]46 <li><a href="#objc_objs-cast">Internal Objective-C types</a></li>
Fariborz Jahanianddfa6c32010-10-22 22:35:51 +0000[diff] [blame]47 <li><a href="#c_variables-class">C variables in @class or @protocol</a></li>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]48 </ul>
49 </li>
Benjamin Kramer665a8dc2012-01-15 15:26:07 +0000[diff] [blame]50 <li><a href="#cxx">C++ compatibility</a>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]51 <ul>
52 <li><a href="#vla">Variable-length arrays</a></li>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]53 <li><a href="#dep_lookup">Unqualified lookup in templates</a></li>
54 <li><a href="#dep_lookup_bases">Unqualified lookup into dependent bases of class templates</a></li>
55 <li><a href="#undep_incomplete">Incomplete types in templates</a></li>
56 <li><a href="#bad_templates">Templates with no valid instantiations</a></li>
57 <li><a href="#default_init_const">Default initialization of const
58 variable of a class type requires user-defined default
59 constructor</a></li>
Douglas Gregora66d3bb2010-11-10 20:24:21 +0000[diff] [blame]60 <li><a href="#param_name_lookup">Parameter name lookup</a></li>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]61 </ul>
62 </li>
Benjamin Kramer665a8dc2012-01-15 15:26:07 +0000[diff] [blame]63 <li><a href="#cxx11">C++11 compatibility</a>
Douglas Gregorfb2a0c52011-09-27 18:58:27 +0000[diff] [blame]64 <ul>
65 <li><a href="#deleted-special-func">Deleted special member
66 functions</a></li>
67 </ul>
68 </li>
Benjamin Kramer665a8dc2012-01-15 15:26:07 +0000[diff] [blame]69 <li><a href="#objective-cxx">Objective-C++ compatibility</a>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]70 <ul>
71 <li><a href="#implicit-downcasts">Implicit downcasts</a></li>
72 </ul>
Fariborz Jahanian36e738a2010-08-11 18:57:26 +0000[diff] [blame]73 <ul>
John McCall6966c672011-02-03 11:29:18 +0000[diff] [blame]74 <li><a href="#class-as-property-name">Using <code>class</code> as a property name</a></li>
Fariborz Jahanian36e738a2010-08-11 18:57:26 +0000[diff] [blame]75 </ul>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]76 </li>
77</ul>
78
79<!-- ======================================================================= -->
Benjamin Kramer665a8dc2012-01-15 15:26:07 +0000[diff] [blame]80<h2 id="c">C compatibility</h2>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]81<!-- ======================================================================= -->
82
83<!-- ======================================================================= -->
84<h3 id="inline">C99 inline functions</h3>
85<!-- ======================================================================= -->
86<p>By default, Clang builds C code according to the C99 standard,
John McCall5ae84f22011-02-03 10:16:40 +0000[diff] [blame]87which provides different semantics for the <code>inline</code> keyword
88than GCC's default behavior. For example, consider the following
89code:</p>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]90<pre>
91inline int add(int i, int j) { return i + j; }
92
93int main() {
94 int i = add(4, 5);
95 return i;
96}
97</pre>
98
John McCall5ae84f22011-02-03 10:16:40 +0000[diff] [blame]99<p>In C99, <code>inline</code> means that a function's definition is
100provided only for inlining, and that there is another definition
101(without <code>inline</code>) somewhere else in the program. That
102means that this program is incomplete, because if <code>add</code>
103isn't inlined (for example, when compiling without optimization), then
104<code>main</code> will have an unresolved reference to that other
105definition. Therefore we'll get a (correct) link-time error like this:</p>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]106
107<pre>
108Undefined symbols:
109 "_add", referenced from:
110 _main in cc-y1jXIr.o
111</pre>
112
John McCall5ae84f22011-02-03 10:16:40 +0000[diff] [blame]113<p>By contrast, GCC's default behavior follows the GNU89 dialect,
114which is the C89 standard plus a lot of extensions. C89 doesn't have
115an <code>inline</code> keyword, but GCC recognizes it as an extension
116and just treats it as a hint to the optimizer.</p>
117
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]118<p>There are several ways to fix this problem:</p>
119
120<ul>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]121 <li>Change <code>add</code> to a <code>static inline</code>
John McCall5ae84f22011-02-03 10:16:40 +0000[diff] [blame]122 function. This is usually the right solution if only one
123 translation unit needs to use the function. <code>static
124 inline</code> functions are always resolved within the translation
125 unit, so you won't have to add a non-<code>inline</code> definition
126 of the function elsewhere in your program.</li>
127
128 <li>Remove the <code>inline</code> keyword from this definition of
129 <code>add</code>. The <code>inline</code> keyword is not required
130 for a function to be inlined, nor does it guarantee that it will be.
131 Some compilers ignore it completely. Clang treats it as a mild
132 suggestion from the programmer.</li>
133
134 <li>Provide an external (non-<code>inline</code>) definition
135 of <code>add</code> somewhere else in your program. The two
136 definitions must be equivalent!</li>
137
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]138 <li>Compile with the GNU89 dialect by adding
139 <code>-std=gnu89</code> to the set of Clang options. This option is
140 only recommended if the program source cannot be changed or if the
141 program also relies on additional C89-specific behavior that cannot
142 be changed.</li>
143</ul>
144
John McCall5ae84f22011-02-03 10:16:40 +0000[diff] [blame]145<p>All of this only applies to C code; the meaning of <code>inline</code>
146in C++ is very different from its meaning in either GNU89 or C99.</p>
Chris Lattnera02d1832010-09-16 18:17:55 +0000[diff] [blame]147
148<!-- ======================================================================= -->
149<h3 id="vector_builtins">"missing" vector __builtin functions</h3>
150<!-- ======================================================================= -->
151
152<p>The Intel and AMD manuals document a number "<tt>&lt;*mmintrin.h&gt;</tt>"
153header files, which define a standardized API for accessing vector operations
154on X86 CPUs. These functions have names like <tt>_mm_xor_ps</tt> and
155<tt>_mm256_addsub_pd</tt>. Compilers have leeway to implement these functions
156however they want. Since Clang supports an excellent set of <a
157href="../docs/LanguageExtensions.html#vectors">native vector operations</a>,
158the Clang headers implement these interfaces in terms of the native vector
159operations.
160</p>
161
162<p>In contrast, GCC implements these functions mostly as a 1-to-1 mapping to
163builtin function calls, like <tt>__builtin_ia32_paddw128</tt>. These builtin
164functions are an internal implementation detail of GCC, and are not portable to
165the Intel compiler, the Microsoft compiler, or Clang. If you get build errors
166mentioning these, the fix is simple: switch to the *mmintrin.h functions.</p>
167
168<p>The same issue occurs for NEON and Altivec for the ARM and PowerPC
169architectures respectively. For these, make sure to use the &lt;arm_neon.h&gt;
170and &lt;altivec.h&gt; headers.</p>
171
Eric Christophera473c952010-10-25 21:17:59 +0000[diff] [blame]172<p>For x86 architectures this <a href="builtins.py">script</a> should help with
173the manual migration process. It will rewrite your source files in place to
174use the APIs instead of builtin function calls. Just call it like this:</p>
175
176<pre>
177 builtins.py *.c *.h
178</pre>
179
180<p>and it will rewrite all of the .c and .h files in the current directory to
181use the API calls instead of calls like <tt>__builtin_ia32_paddw128</tt>.</p>
Chris Lattnera02d1832010-09-16 18:17:55 +0000[diff] [blame]182
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]183<!-- ======================================================================= -->
184<h3 id="lvalue-cast">Lvalue casts</h3>
185<!-- ======================================================================= -->
186
Douglas Gregor6f1adba2010-06-30 22:38:37 +0000[diff] [blame]187<p>Old versions of GCC permit casting the left-hand side of an assignment to a
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]188different type. Clang produces an error on similar code, e.g.,</p>
189
190<pre>
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]191<b>lvalue.c:2:3: <span class="error">error:</span> assignment to cast is illegal, lvalue casts are not supported</b>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]192 (int*)addr = val;
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]193<span class="caret"> ^~~~~~~~~~ ~</span>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]194</pre>
195
196<p>To fix this problem, move the cast to the right-hand side. In this
197example, one could use:</p>
198
199<pre>
200 addr = (float *)val;
201</pre>
202
203<!-- ======================================================================= -->
Daniel Dunbar5a410212010-09-02 21:35:16 +0000[diff] [blame]204<h3 id="blocks-in-protected-scope">Jumps to within <tt>__block</tt> variable scope</h3>
205<!-- ======================================================================= -->
206
John McCall504b3692011-02-03 10:56:31 +0000[diff] [blame]207<p>Clang disallows jumps into the scope of a <tt>__block</tt>
208variable. Variables marked with <tt>__block</tt> require special
209runtime initialization. A jump into the scope of a <tt>__block</tt>
210variable bypasses this initialization, leaving the variable's metadata
211in an invalid state. Consider the following code fragment:</p>
Daniel Dunbar5a410212010-09-02 21:35:16 +0000[diff] [blame]212
213<pre>
John McCall504b3692011-02-03 10:56:31 +0000[diff] [blame]214int fetch_object_state(struct MyObject *c) {
215 if (!c->active) goto error;
Daniel Dunbar5a410212010-09-02 21:35:16 +0000[diff] [blame]216
John McCall504b3692011-02-03 10:56:31 +0000[diff] [blame]217 __block int result;
218 run_specially_somehow(^{ result = c->state; });
219 return result;
Daniel Dunbar5a410212010-09-02 21:35:16 +0000[diff] [blame]220
221 error:
John McCall504b3692011-02-03 10:56:31 +0000[diff] [blame]222 fprintf(stderr, "error while fetching object state");
223 return -1;
Daniel Dunbar5a410212010-09-02 21:35:16 +0000[diff] [blame]224}
225</pre>
226
John McCall504b3692011-02-03 10:56:31 +0000[diff] [blame]227<p>GCC accepts this code, but it produces code that will usually crash
228when <code>result</code> goes out of scope if the jump is taken. (It's
229possible for this bug to go undetected because it often won't crash if
230the stack is fresh, i.e. still zeroed.) Therefore, Clang rejects this
231code with a hard error:</p>
Daniel Dunbar5a410212010-09-02 21:35:16 +0000[diff] [blame]232
233<pre>
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]234<b>t.c:3:5: <span class="error">error:</span> goto into protected scope</b>
Daniel Dunbar5a410212010-09-02 21:35:16 +0000[diff] [blame]235 goto error;
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]236<span class="caret"> ^</span>
237<b>t.c:5:15: <span class="note">note:</note></b> jump bypasses setup of __block variable
John McCall504b3692011-02-03 10:56:31 +0000[diff] [blame]238 __block int result;
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]239<span class="caret"> ^</span>
Daniel Dunbar5a410212010-09-02 21:35:16 +0000[diff] [blame]240</pre>
241
John McCall504b3692011-02-03 10:56:31 +0000[diff] [blame]242<p>The fix is to rewrite the code to not require jumping into a
243<tt>__block</tt> variable's scope, e.g. by limiting that scope:</p>
244
245<pre>
246 {
247 __block int result;
248 run_specially_somehow(^{ result = c->state; });
249 return result;
250 }
251</pre>
Daniel Dunbar5a410212010-09-02 21:35:16 +0000[diff] [blame]252
253<!-- ======================================================================= -->
Daniel Dunbar15952c92010-11-09 22:45:16 +0000[diff] [blame]254<h3 id="block-variable-initialization">Non-initialization of <tt>__block</tt>
255variables</h3>
256<!-- ======================================================================= -->
257
258<p>In the following example code, the <tt>x</tt> variable is used before it is
259defined:</p>
260<pre>
261int f0() {
262 __block int x;
263 return ^(){ return x; }();
264}
265</pre>
266
267<p>By an accident of implementation, GCC and llvm-gcc unintentionally always
268zero initialized <tt>__block</tt> variables. However, any program which depends
269on this behavior is relying on unspecified compiler behavior. Programs must
270explicitly initialize all local block variables before they are used, as with
271other local variables.</p>
272
273<p>Clang does not zero initialize local block variables, and programs which rely
274on such behavior will most likely break when built with Clang.</p>
275
Chris Lattner9b743f42010-11-10 23:51:50 +0000[diff] [blame]276
277<!-- ======================================================================= -->
278<h3 id="inline-asm">Inline assembly</h3>
279<!-- ======================================================================= -->
280
281<p>In general, Clang is highly compatible with the GCC inline assembly
282extensions, allowing the same set of constraints, modifiers and operands as GCC
283inline assembly.</p>
284
285<p>On targets that use the integrated assembler (such as most X86 targets),
286inline assembly is run through the integrated assembler instead of your system
287assembler (which is most commonly "gas", the GNU assembler). The LLVM
288integrated assembler is extremely compatible with GAS, but there are a couple of
289minor places where it is more picky, particularly due to outright GAS bugs.</p>
290
291<p>One specific example is that the assembler rejects ambiguous X86 instructions
292that don't have suffixes. For example:</p>
293
294<pre>
295 asm("add %al, (%rax)");
296 asm("addw $4, (%rax)");
297 asm("add $4, (%rax)");
298</pre>
299
300<p>Both clang and GAS accept the first instruction: because the first
301instruction uses the 8-bit <tt>%al</tt> register as an operand, it is clear that
302it is an 8-bit add. The second instruction is accepted by both because the "w"
303suffix indicates that it is a 16-bit add. The last instruction is accepted by
304GAS even though there is nothing that specifies the size of the instruction (and
305the assembler randomly picks a 32-bit add). Because it is ambiguous, Clang
306rejects the instruction with this error message:
307</p>
308
309<pre>
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]310<b>&lt;inline asm&gt;:3:1: <span class="error">error:</span> ambiguous instructions require an explicit suffix (could be 'addb', 'addw', 'addl', or 'addq')</b>
Chris Lattner9b743f42010-11-10 23:51:50 +0000[diff] [blame]311add $4, (%rax)
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]312<span class="caret">^</span>
Chris Lattner9b743f42010-11-10 23:51:50 +0000[diff] [blame]313</pre>
314
315<p>To fix this compatibility issue, add an explicit suffix to the instruction:
316this makes your code more clear and is compatible with both GCC and Clang.</p>
317
Daniel Dunbar15952c92010-11-09 22:45:16 +0000[diff] [blame]318<!-- ======================================================================= -->
Benjamin Kramer665a8dc2012-01-15 15:26:07 +0000[diff] [blame]319<h2 id="objective-c">Objective-C compatibility</h2>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]320<!-- ======================================================================= -->
321
322<!-- ======================================================================= -->
323<h3 id="super-cast">Cast of super</h3>
324<!-- ======================================================================= -->
325
326<p>GCC treats the <code>super</code> identifier as an expression that
327can, among other things, be cast to a different type. Clang treats
328<code>super</code> as a context-sensitive keyword, and will reject a
329type-cast of <code>super</code>:</p>
330
331<pre>
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]332<b>super.m:11:12: <span class="error">error:</span> cannot cast 'super' (it isn't an expression)</b>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]333 [(Super*)super add:4];
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]334<span class="caret"> ~~~~~~~~^</span>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]335</pre>
336
337<p>To fix this problem, remove the type cast, e.g.</p>
338<pre>
339 [super add:4];
340</pre>
341
342<!-- ======================================================================= -->
343<h3 id="sizeof-interface">Size of interfaces</h3>
344<!-- ======================================================================= -->
345
346<p>When using the "non-fragile" Objective-C ABI in use, the size of an
347Objective-C class may change over time as instance variables are added
348(or removed). For this reason, Clang rejects the application of the
349<code>sizeof</code> operator to an Objective-C class when using this
350ABI:</p>
351
352<pre>
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]353<b>sizeof.m:4:14: <span class="error">error:</span> invalid application of 'sizeof' to interface 'NSArray' in non-fragile ABI</b>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]354 int size = sizeof(NSArray);
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]355<span class="caret"> ^ ~~~~~~~~~</span>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]356</pre>
357
358<p>Code that relies on the size of an Objective-C class is likely to
359be broken anyway, since that size is not actually constant. To address
360this problem, use the Objective-C runtime API function
Benjamin Kramere6617502010-06-30 22:29:56 +0000[diff] [blame]361<code>class_getInstanceSize()</code>:</p>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]362
363<pre>
364 class_getInstanceSize([NSArray class])
365</pre>
366
367<!-- ======================================================================= -->
Argyrios Kyrtzidis3b5b92a2010-09-13 17:48:07 +0000[diff] [blame]368<h3 id="objc_objs-cast">Internal Objective-C types</h3>
369<!-- ======================================================================= -->
370
371<p>GCC allows using pointers to internal Objective-C objects, <tt>struct objc_object*</tt>,
372<tt>struct objc_selector*</tt>, and <tt>struct objc_class*</tt> in place of the types
373<tt>id</tt>, <tt>SEL</tt>, and <tt>Class</tt> respectively. Clang treats the
374internal Objective-C structures as implementation detail and won't do implicit conversions:
375
376<pre>
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]377<b>t.mm:11:2: <span class="error">error:</span> no matching function for call to 'f'</b>
Argyrios Kyrtzidis3b5b92a2010-09-13 17:48:07 +0000[diff] [blame]378 f((struct objc_object *)p);
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]379<span class="caret"> ^</span>
380<b>t.mm:5:6: <span class="note">note:</note></b> candidate function not viable: no known conversion from 'struct objc_object *' to 'id' for 1st argument
Argyrios Kyrtzidis3b5b92a2010-09-13 17:48:07 +0000[diff] [blame]381void f(id x);
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]382<span class="caret"> ^</span>
Argyrios Kyrtzidis3b5b92a2010-09-13 17:48:07 +0000[diff] [blame]383</pre>
384
385<p>Code should use types <tt>id</tt>, <tt>SEL</tt>, and <tt>Class</tt>
386instead of the internal types.</p>
387
388<!-- ======================================================================= -->
John McCall3cdbe442011-02-03 11:05:04 +0000[diff] [blame]389<h3 id="c_variables-class">C variables in @interface or @protocol</h3>
Fariborz Jahanianddfa6c32010-10-22 22:35:51 +0000[diff] [blame]390<!-- ======================================================================= -->
391
John McCall3cdbe442011-02-03 11:05:04 +0000[diff] [blame]392<p>GCC allows the declaration of C variables in
393an <code>@interface</code> or <code>@protocol</code>
394declaration. Clang does not allow variable declarations to appear
395within these declarations unless they are marked <code>extern</code>.</p>
396
397<p>Variables may still be declared in an @implementation.</p>
Fariborz Jahanianddfa6c32010-10-22 22:35:51 +0000[diff] [blame]398
399<pre>
400@interface XX
John McCall3cdbe442011-02-03 11:05:04 +0000[diff] [blame]401int a; // not allowed in clang
402int b = 1; // not allowed in clang
403extern int c; // allowed
Fariborz Jahanianddfa6c32010-10-22 22:35:51 +0000[diff] [blame]404@end
405
406</pre>
407
408<!-- ======================================================================= -->
Benjamin Kramer665a8dc2012-01-15 15:26:07 +0000[diff] [blame]409<h2 id="cxx">C++ compatibility</h2>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]410<!-- ======================================================================= -->
411
412<!-- ======================================================================= -->
413<h3 id="vla">Variable-length arrays</h3>
414<!-- ======================================================================= -->
415
416<p>GCC and C99 allow an array's size to be determined at run
417time. This extension is not permitted in standard C++. However, Clang
418supports such variable length arrays in very limited circumstances for
419compatibility with GNU C and C99 programs:</p>
420
421<ul>
422 <li>The element type of a variable length array must be a POD
423 ("plain old data") type, which means that it cannot have any
John McCall6966c672011-02-03 11:29:18 +0000[diff] [blame]424 user-declared constructors or destructors, any base classes, or any
425 members of non-POD type. All C types are POD types.</li>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]426
427 <li>Variable length arrays cannot be used as the type of a non-type
428template parameter.</li> </ul>
429
430<p>If your code uses variable length arrays in a manner that Clang doesn't support, there are several ways to fix your code:
431
432<ol>
433<li>replace the variable length array with a fixed-size array if you can
John McCall6966c672011-02-03 11:29:18 +0000[diff] [blame]434 determine a reasonable upper bound at compile time; sometimes this is as
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]435 simple as changing <tt>int size = ...;</tt> to <tt>const int size
John McCall6966c672011-02-03 11:29:18 +0000[diff] [blame]436 = ...;</tt> (if the initializer is a compile-time constant);</li>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]437<li>use <tt>std::vector</tt> or some other suitable container type;
438 or</li>
439<li>allocate the array on the heap instead using <tt>new Type[]</tt> -
440 just remember to <tt>delete[]</tt> it.</li>
441</ol>
442
443<!-- ======================================================================= -->
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]444<h3 id="dep_lookup">Unqualified lookup in templates</h3>
445<!-- ======================================================================= -->
446
447<p>Some versions of GCC accept the following invalid code:
448
449<pre>
450template &lt;typename T&gt; T Squared(T x) {
451 return Multiply(x, x);
452}
453
454int Multiply(int x, int y) {
455 return x * y;
456}
457
458int main() {
459 Squared(5);
460}
461</pre>
462
463<p>Clang complains:
464
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]465<pre>
466<b>my_file.cpp:2:10: <span class="error">error:</span> call to function 'Multiply' that is neither visible in the template definition nor found by argument-dependent lookup</b>
467 return Multiply(x, x);
468<span class="caret"> ^</span>
469<b>my_file.cpp:10:3: <span class="note">note:</span></b> in instantiation of function template specialization 'Squared&lt;int&gt;' requested here
470 Squared(5);
471<span class="caret"> ^</span>
472<b>my_file.cpp:5:5: <span class="note">note:</span></b> 'Multiply' should be declared prior to the call site
473int Multiply(int x, int y) {
474<span class="caret"> ^</span>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]475</pre>
476
477<p>The C++ standard says that unqualified names like <q>Multiply</q>
478are looked up in two ways.
479
480<p>First, the compiler does <i>unqualified lookup</i> in the scope
481where the name was written. For a template, this means the lookup is
482done at the point where the template is defined, not where it's
483instantiated. Since <tt>Multiply</tt> hasn't been declared yet at
484this point, unqualified lookup won't find it.
485
486<p>Second, if the name is called like a function, then the compiler
487also does <i>argument-dependent lookup</i> (ADL). (Sometimes
488unqualified lookup can suppress ADL; see [basic.lookup.argdep]p3 for
489more information.) In ADL, the compiler looks at the types of all the
490arguments to the call. When it finds a class type, it looks up the
491name in that class's namespace; the result is all the declarations it
492finds in those namespaces, plus the declarations from unqualified
493lookup. However, the compiler doesn't do ADL until it knows all the
494argument types.
495
496<p>In our example, <tt>Multiply</tt> is called with dependent
497arguments, so ADL isn't done until the template is instantiated. At
498that point, the arguments both have type <tt>int</tt>, which doesn't
499contain any class types, and so ADL doesn't look in any namespaces.
500Since neither form of lookup found the declaration
501of <tt>Multiply</tt>, the code doesn't compile.
502
503<p>Here's another example, this time using overloaded operators,
504which obey very similar rules.
505
506<pre>#include &lt;iostream&gt;
507
508template&lt;typename T&gt;
509void Dump(const T&amp; value) {
510 std::cout &lt;&lt; value &lt;&lt; "\n";
511}
512
513namespace ns {
514 struct Data {};
515}
516
517std::ostream&amp; operator&lt;&lt;(std::ostream&amp; out, ns::Data data) {
518 return out &lt;&lt; "Some data";
519}
520
521void Use() {
522 Dump(ns::Data());
523}</pre>
524
Richard Smithf50e88a2011-06-05 22:42:48 +0000[diff] [blame]525<p>Again, Clang complains:</p>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]526
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]527<pre>
528<b>my_file2.cpp:5:13: <span class="error">error:</span> call to function 'operator&lt;&lt;' that is neither visible in the template definition nor found by argument-dependent lookup</b>
529 std::cout &lt;&lt; value &lt;&lt; "\n";
530<span class="caret"> ^</span>
531<b>my_file2.cpp:17:3: <span class="note">note:</span></b> in instantiation of function template specialization 'Dump&lt;ns::Data&gt;' requested here
532 Dump(ns::Data());
533<span class="caret"> ^</span>
534<b>my_file2.cpp:12:15: <span class="note">note:</span></b> 'operator&lt;&lt;' should be declared prior to the call site or in namespace 'ns'
535std::ostream&amp; operator&lt;&lt;(std::ostream&amp; out, ns::Data data) {
536<span class="caret"> ^</span>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]537</pre>
538
539<p>Just like before, unqualified lookup didn't find any declarations
540with the name <tt>operator&lt;&lt;</tt>. Unlike before, the argument
541types both contain class types: one of them is an instance of the
542class template type <tt>std::basic_ostream</tt>, and the other is the
543type <tt>ns::Data</tt> that we declared above. Therefore, ADL will
544look in the namespaces <tt>std</tt> and <tt>ns</tt> for
545an <tt>operator&lt;&lt;</tt>. Since one of the argument types was
546still dependent during the template definition, ADL isn't done until
547the template is instantiated during <tt>Use</tt>, which means that
548the <tt>operator&lt;&lt;</tt> we want it to find has already been
549declared. Unfortunately, it was declared in the global namespace, not
550in either of the namespaces that ADL will look in!
551
552<p>There are two ways to fix this problem:</p>
553<ol><li>Make sure the function you want to call is declared before the
554template that might call it. This is the only option if none of its
555argument types contain classes. You can do this either by moving the
556template definition, or by moving the function definition, or by
557adding a forward declaration of the function before the template.</li>
558<li>Move the function into the same namespace as one of its arguments
559so that ADL applies.</li></ol>
560
561<p>For more information about argument-dependent lookup, see
562[basic.lookup.argdep]. For more information about the ordering of
563lookup in templates, see [temp.dep.candidate].
564
565<!-- ======================================================================= -->
566<h3 id="dep_lookup_bases">Unqualified lookup into dependent bases of class templates</h3>
567<!-- ======================================================================= -->
568
569Some versions of GCC accept the following invalid code:
570
571<pre>
572template &lt;typename T&gt; struct Base {
573 void DoThis(T x) {}
574 static void DoThat(T x) {}
575};
576
577template &lt;typename T&gt; struct Derived : public Base&lt;T&gt; {
578 void Work(T x) {
579 DoThis(x); // Invalid!
580 DoThat(x); // Invalid!
581 }
582};
583</pre>
584
585Clang correctly rejects it with the following errors
586(when <tt>Derived</tt> is eventually instantiated):
587
588<pre>
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]589<b>my_file.cpp:8:5: <span class="error">error:</span> use of undeclared identifier 'DoThis'</b>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]590 DoThis(x);
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]591<span class="caret"> ^</span>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]592 this-&gt;
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]593<b>my_file.cpp:2:8: <span class="note">note:</note></b> must qualify identifier to find this declaration in dependent base class
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]594 void DoThis(T x) {}
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]595<span class="caret"> ^</span>
596<b>my_file.cpp:9:5: <span class="error">error:</span> use of undeclared identifier 'DoThat'</b>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]597 DoThat(x);
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]598<span class="caret"> ^</span>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]599 this-&gt;
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]600<b>my_file.cpp:3:15: <span class="note">note:</note></b> must qualify identifier to find this declaration in dependent base class
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]601 static void DoThat(T x) {}
602</pre>
603
604Like we said <a href="#dep_lookup">above</a>, unqualified names like
605<tt>DoThis</tt> and <tt>DoThat</tt> are looked up when the template
606<tt>Derived</tt> is defined, not when it's instantiated. When we look
607up a name used in a class, we usually look into the base classes.
608However, we can't look into the base class <tt>Base&lt;T&gt;</tt>
609because its type depends on the template argument <tt>T</tt>, so the
610standard says we should just ignore it. See [temp.dep]p3 for details.
611
612<p>The fix, as Clang tells you, is to tell the compiler that we want a
613class member by prefixing the calls with <tt>this-&gt;</tt>:
614
615<pre>
616 void Work(T x) {
617 <b>this-&gt;</b>DoThis(x);
618 <b>this-&gt;</b>DoThat(x);
619 }
620</pre>
621
622Alternatively, you can tell the compiler exactly where to look:
623
624<pre>
625 void Work(T x) {
626 <b>Base&lt;T&gt;</b>::DoThis(x);
627 <b>Base&lt;T&gt;</b>::DoThat(x);
628 }
629</pre>
630
631This works whether the methods are static or not, but be careful:
632if <tt>DoThis</tt> is virtual, calling it this way will bypass virtual
633dispatch!
634
635<!-- ======================================================================= -->
636<h3 id="undep_incomplete">Incomplete types in templates</h3>
637<!-- ======================================================================= -->
638
639The following code is invalid, but compilers are allowed to accept it:
640
641<pre>
642 class IOOptions;
643 template &lt;class T&gt; bool read(T &amp;value) {
644 IOOptions opts;
645 return read(opts, value);
646 }
647
648 class IOOptions { bool ForceReads; };
649 bool read(const IOOptions &amp;opts, int &amp;x);
650 template bool read&lt;&gt;(int &amp;);
651</pre>
652
653The standard says that types which don't depend on template parameters
654must be complete when a template is defined if they affect the
655program's behavior. However, the standard also says that compilers
656are free to not enforce this rule. Most compilers enforce it to some
657extent; for example, it would be an error in GCC to
658write <tt>opts.ForceReads</tt> in the code above. In Clang, we feel
659that enforcing the rule consistently lets us provide a better
660experience, but unfortunately it also means we reject some code that
661other compilers accept.
662
663<p>We've explained the rule here in very imprecise terms; see
664[temp.res]p8 for details.
665
666<!-- ======================================================================= -->
667<h3 id="bad_templates">Templates with no valid instantiations</h3>
668<!-- ======================================================================= -->
669
670The following code contains a typo: the programmer
671meant <tt>init()</tt> but wrote <tt>innit()</tt> instead.
672
673<pre>
674 template &lt;class T&gt; class Processor {
675 ...
676 void init();
677 ...
678 };
679 ...
680 template &lt;class T&gt; void process() {
681 Processor&lt;T&gt; processor;
682 processor.innit(); // <-- should be 'init()'
683 ...
684 }
685</pre>
686
687Unfortunately, we can't flag this mistake as soon as we see it: inside
688a template, we're not allowed to make assumptions about "dependent
689types" like <tt>Processor&lt;T&gt;</tt>. Suppose that later on in
690this file the programmer adds an explicit specialization
691of <tt>Processor</tt>, like so:
692
693<pre>
694 template &lt;&gt; class Processor&lt;char*&gt; {
695 void innit();
696 };
697</pre>
698
699Now the program will work &mdash; as long as the programmer only ever
700instantiates <tt>process()</tt> with <tt>T = char*</tt>! This is why
701it's hard, and sometimes impossible, to diagnose mistakes in a
702template definition before it's instantiated.
703
704<p>The standard says that a template with no valid instantiations is
705ill-formed. Clang tries to do as much checking as possible at
706definition-time instead of instantiation-time: not only does this
707produce clearer diagnostics, but it also substantially improves
708compile times when using pre-compiled headers. The downside to this
709philosophy is that Clang sometimes fails to process files because they
710contain broken templates that are no longer used. The solution is
711simple: since the code is unused, just remove it.
712
713<!-- ======================================================================= -->
714<h3 id="default_init_const">Default initialization of const variable of a class type requires user-defined default constructor</h3>
715<!-- ======================================================================= -->
716
717If a <tt>class</tt> or <tt>struct</tt> has no user-defined default
718constructor, C++ doesn't allow you to default construct a <tt>const</tt>
719instance of it like this ([dcl.init], p9):
720
721<pre>
722class Foo {
723 public:
724 // The compiler-supplied default constructor works fine, so we
725 // don't bother with defining one.
726 ...
727};
728
729void Bar() {
730 const Foo foo; // Error!
731 ...
732}
733</pre>
734
735To fix this, you can define a default constructor for the class:
736
737<pre>
738class Foo {
739 public:
740 Foo() {}
741 ...
742};
743
744void Bar() {
745 const Foo foo; // Now the compiler is happy.
746 ...
747}
748</pre>
749
750<!-- ======================================================================= -->
Douglas Gregora66d3bb2010-11-10 20:24:21 +0000[diff] [blame]751<h3 id="param_name_lookup">Parameter name lookup</h3>
752<!-- ======================================================================= -->
753
754<p>Due to a bug in its implementation, GCC allows the redeclaration of function parameter names within a function prototype in C++ code, e.g.</p>
755<blockquote>
756<pre>
757void f(int a, int a);
758</pre>
759</blockquote>
760<p>Clang diagnoses this error (where the parameter name has been redeclared). To fix this problem, rename one of the parameters.</p>
761
762<!-- ======================================================================= -->
Benjamin Kramer665a8dc2012-01-15 15:26:07 +0000[diff] [blame]763<h2 id="cxx11">C++11 compatibility</h2>
Douglas Gregorfb2a0c52011-09-27 18:58:27 +0000[diff] [blame]764<!-- ======================================================================= -->
765
766<!-- ======================================================================= -->
767<h3 id="deleted-special-func">Deleted special member functions</h3>
768<!-- ======================================================================= -->
769
770<p>In C++11, the explicit declaration of a move constructor or a move
Sean Hunt04bea932011-12-29 18:40:13 +0000[diff] [blame]771assignment operator within a class deletes the implicit declaration
Douglas Gregorfb2a0c52011-09-27 18:58:27 +0000[diff] [blame]772of the copy constructor and copy assignment operator. This change came
773fairly late in the C++11 standardization process, so early
774implementations of C++11 (including Clang before 3.0, GCC before 4.7,
775and Visual Studio 2010) do not implement this rule, leading them to
776accept this ill-formed code:</p>
777
778<pre>
779struct X {
Sean Hunt04bea932011-12-29 18:40:13 +0000[diff] [blame]780 X(X&amp;&amp;); <i>// deletes implicit copy constructor:</i>
781 <i>// X(const X&amp;) = delete;</i>
Douglas Gregorfb2a0c52011-09-27 18:58:27 +0000[diff] [blame]782};
783
784void f(X x);
785void g(X x) {
Sean Hunt04bea932011-12-29 18:40:13 +0000[diff] [blame]786 f(x); <i>// error: X has a deleted copy constructor</i>
Douglas Gregorfb2a0c52011-09-27 18:58:27 +0000[diff] [blame]787}
788</pre>
789
Sean Hunt04bea932011-12-29 18:40:13 +0000[diff] [blame]790<p>This affects some early C++11 code, including Boost's popular <a
Douglas Gregorfb2a0c52011-09-27 18:58:27 +0000[diff] [blame]791href="http://www.boost.org/doc/libs/release/libs/smart_ptr/shared_ptr.htm"><tt>shared_ptr</tt></a>
792up to version 1.47.0. The fix for Boost's <tt>shared_ptr</tt> is
793<a href="https://svn.boost.org/trac/boost/changeset/73202">available here</a>.</p>
794
795<!-- ======================================================================= -->
Benjamin Kramer665a8dc2012-01-15 15:26:07 +0000[diff] [blame]796<h2 id="objective-cxx">Objective-C++ compatibility</h2>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]797<!-- ======================================================================= -->
798
799<!-- ======================================================================= -->
800<h3 id="implicit-downcasts">Implicit downcasts</h3>
801<!-- ======================================================================= -->
802
803<p>Due to a bug in its implementation, GCC allows implicit downcasts
John McCall6966c672011-02-03 11:29:18 +0000[diff] [blame]804of Objective-C pointers (from a base class to a derived class) when
805calling functions. Such code is inherently unsafe, since the object
806might not actually be an instance of the derived class, and is
807rejected by Clang. For example, given this code:</p>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]808
809<pre>
810@interface Base @end
811@interface Derived : Base @end
812
John McCall6966c672011-02-03 11:29:18 +0000[diff] [blame]813void f(Derived *p);
814void g(Base *p) {
815 f(p);
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]816}
817</pre>
818
819<p>Clang produces the following error:</p>
820
821<pre>
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]822<b>downcast.mm:6:3: <span class="error">error:</span> no matching function for call to 'f'</b>
John McCall6966c672011-02-03 11:29:18 +0000[diff] [blame]823 f(p);
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]824<span class="caret"> ^</span>
825<b>downcast.mm:4:6: <span class="note">note:</note></b> candidate function not viable: cannot convert from
Douglas Gregor92bc0272010-07-01 03:50:01 +0000[diff] [blame]826 superclass 'Base *' to subclass 'Derived *' for 1st argument
John McCall6966c672011-02-03 11:29:18 +0000[diff] [blame]827void f(Derived *p);
Richard Smitha55e3ff2013-02-01 00:45:12 +0000[diff] [blame^]828<span class="caret"> ^</span>
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]829</pre>
830
831<p>If the downcast is actually correct (e.g., because the code has
832already checked that the object has the appropriate type), add an
833explicit cast:</p>
834
835<pre>
836 f((Derived *)base);
837</pre>
838
Fariborz Jahanian36e738a2010-08-11 18:57:26 +0000[diff] [blame]839<!-- ======================================================================= -->
John McCall6966c672011-02-03 11:29:18 +0000[diff] [blame]840<h3 id="class-as-property-name">Using <code>class</code> as a property name</h3>
Fariborz Jahanian36e738a2010-08-11 18:57:26 +0000[diff] [blame]841<!-- ======================================================================= -->
842
John McCall6966c672011-02-03 11:29:18 +0000[diff] [blame]843<p>In C and Objective-C, <code>class</code> is a normal identifier and
844can be used to name fields, ivars, methods, and so on. In
845C++, <code>class</code> is a keyword. For compatibility with existing
846code, Clang permits <code>class</code> to be used as part of a method
847selector in Objective-C++, but this does not extend to any other part
848of the language. In particular, it is impossible to use property dot
849syntax in Objective-C++ with the property name <code>class</code>, so
850the following code will fail to parse:</p>
Fariborz Jahanian36e738a2010-08-11 18:57:26 +0000[diff] [blame]851
852<pre>
853@interface I {
854int cls;
855}
856+ (int)class;
857@end
858
859@implementation I
860- (int) Meth { return I.class; }
861@end
Benjamin Kramer665a8dc2012-01-15 15:26:07 +0000[diff] [blame]862</pre>
Fariborz Jahanian36e738a2010-08-11 18:57:26 +0000[diff] [blame]863
John McCall6966c672011-02-03 11:29:18 +0000[diff] [blame]864<p>Use explicit message-send syntax instead, i.e. <code>[I class]</code>.</p>
Fariborz Jahanian36e738a2010-08-11 18:57:26 +0000[diff] [blame]865
Douglas Gregorc41b6ff2010-06-30 22:01:08 +0000[diff] [blame]866</div>
867</body>
868</html>