Gitiles
Code Review Sign In
gerrit-public.fairphone.software / platform / external / clang / 4c50b697bbf76c13f5b7860518a5ac0484dbbc90 / . / docs / InternalsManual.html
blob: daec6b019415726e9f84710df10e7ae5d0b503d8 [file] [log] [blame]
Ted Kremenek17a295d2008-06-11 06:19:49 +0000[diff] [blame]1<html>
2<head>
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]3<title>"Clang" CFE Internals Manual</title>
Ted Kremenek17a295d2008-06-11 06:19:49 +0000[diff] [blame]4<link type="text/css" rel="stylesheet" href="../menu.css" />
5<link type="text/css" rel="stylesheet" href="../content.css" />
Sebastian Redl68168562008-11-22 22:16:45 +0000[diff] [blame]6<style type="text/css">
7td {
8 vertical-align: top;
9}
10</style>
Ted Kremenek17a295d2008-06-11 06:19:49 +0000[diff] [blame]11</head>
12<body>
13
14<!--#include virtual="../menu.html.incl"-->
15
16<div id="content">
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]17
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]18<h1>"Clang" CFE Internals Manual</h1>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]19
20<ul>
21<li><a href="#intro">Introduction</a></li>
22<li><a href="#libsystem">LLVM System and Support Libraries</a></li>
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]23<li><a href="#libbasic">The Clang 'Basic' Library</a>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]24 <ul>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]25 <li><a href="#Diagnostics">The Diagnostics Subsystem</a></li>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]26 <li><a href="#SourceLocation">The SourceLocation and SourceManager
27 classes</a></li>
28 </ul>
29</li>
Daniel Dunbar27d9e9f2009-03-30 06:50:01 +0000[diff] [blame]30<li><a href="#libdriver">The Driver Library</a>
31 <ul>
32 </ul>
33</li>
Douglas Gregor32110df2009-05-20 00:16:32 +0000[diff] [blame]34<li><a href="#pch">Precompiled Headers</a>
Daniel Dunbar27d9e9f2009-03-30 06:50:01 +0000[diff] [blame]35<li><a href="#libfrontend">The Frontend Library</a>
36 <ul>
37 </ul>
38</li>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]39<li><a href="#liblex">The Lexer and Preprocessor Library</a>
40 <ul>
41 <li><a href="#Token">The Token class</a></li>
42 <li><a href="#Lexer">The Lexer class</a></li>
Chris Lattner3932fe02009-01-06 06:02:08 +0000[diff] [blame]43 <li><a href="#AnnotationToken">Annotation Tokens</a></li>
Chris Lattner79281252008-03-09 02:27:26 +0000[diff] [blame]44 <li><a href="#TokenLexer">The TokenLexer class</a></li>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]45 <li><a href="#MultipleIncludeOpt">The MultipleIncludeOpt class</a></li>
46 </ul>
47</li>
48<li><a href="#libparse">The Parser Library</a>
49 <ul>
50 </ul>
51</li>
52<li><a href="#libast">The AST Library</a>
53 <ul>
54 <li><a href="#Type">The Type class and its subclasses</a></li>
55 <li><a href="#QualType">The QualType class</a></li>
Douglas Gregor2e1cd422008-11-17 14:58:09 +0000[diff] [blame]56 <li><a href="#DeclarationName">Declaration names</a></li>
Douglas Gregor074149e2009-01-05 19:45:36 +0000[diff] [blame]57 <li><a href="#DeclContext">Declaration contexts</a>
58 <ul>
59 <li><a href="#Redeclarations">Redeclarations and Overloads</a></li>
60 <li><a href="#LexicalAndSemanticContexts">Lexical and Semantic
61 Contexts</a></li>
62 <li><a href="#TransparentContexts">Transparent Declaration Contexts</a></li>
63 <li><a href="#MultiDeclContext">Multiply-Defined Declaration Contexts</a></li>
64 </ul>
65 </li>
Ted Kremenek8bc05712007-10-10 23:01:43 +0000[diff] [blame]66 <li><a href="#CFG">The CFG class</a></li>
Chris Lattner7bad1992008-11-16 21:48:07 +0000[diff] [blame]67 <li><a href="#Constants">Constant Folding in the Clang AST</a></li>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]68 </ul>
69</li>
Argyrios Kyrtzidis7240d772009-07-10 03:41:36 +0000[diff] [blame]70<li><a href="libIndex.html">The Index Library</a></li>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]71</ul>
72
73
74<!-- ======================================================================= -->
75<h2 id="intro">Introduction</h2>
76<!-- ======================================================================= -->
77
78<p>This document describes some of the more important APIs and internal design
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]79decisions made in the Clang C front-end. The purpose of this document is to
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]80both capture some of this high level information and also describe some of the
81design decisions behind it. This is meant for people interested in hacking on
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]82Clang, not for end-users. The description below is categorized by
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]83libraries, and does not describe any of the clients of the libraries.</p>
84
85<!-- ======================================================================= -->
86<h2 id="libsystem">LLVM System and Support Libraries</h2>
87<!-- ======================================================================= -->
88
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]89<p>The LLVM libsystem library provides the basic Clang system abstraction layer,
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]90which is used for file system access. The LLVM libsupport library provides many
91underlying libraries and <a
92href="http://llvm.org/docs/ProgrammersManual.html">data-structures</a>,
93 including command line option
94processing and various containers.</p>
95
96<!-- ======================================================================= -->
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]97<h2 id="libbasic">The Clang 'Basic' Library</h2>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]98<!-- ======================================================================= -->
99
100<p>This library certainly needs a better name. The 'basic' library contains a
101number of low-level utilities for tracking and manipulating source buffers,
102locations within the source buffers, diagnostics, tokens, target abstraction,
103and information about the subset of the language being compiled for.</p>
104
105<p>Part of this infrastructure is specific to C (such as the TargetInfo class),
106other parts could be reused for other non-C-based languages (SourceLocation,
107SourceManager, Diagnostics, FileManager). When and if there is future demand
108we can figure out if it makes sense to introduce a new library, move the general
109classes somewhere else, or introduce some other solution.</p>
110
111<p>We describe the roles of these classes in order of their dependencies.</p>
112
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]113
114<!-- ======================================================================= -->
115<h3 id="Diagnostics">The Diagnostics Subsystem</h3>
116<!-- ======================================================================= -->
117
118<p>The Clang Diagnostics subsystem is an important part of how the compiler
119communicates with the human. Diagnostics are the warnings and errors produced
120when the code is incorrect or dubious. In Clang, each diagnostic produced has
121(at the minimum) a unique ID, a <a href="#SourceLocation">SourceLocation</a> to
122"put the caret", an English translation associated with it, and a severity (e.g.
123<tt>WARNING</tt> or <tt>ERROR</tt>). They can also optionally include a number
124of arguments to the dianostic (which fill in "%0"'s in the string) as well as a
125number of source ranges that related to the diagnostic.</p>
126
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]127<p>In this section, we'll be giving examples produced by the Clang command line
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]128driver, but diagnostics can be <a href="#DiagnosticClient">rendered in many
129different ways</a> depending on how the DiagnosticClient interface is
130implemented. A representative example of a diagonstic is:</p>
131
132<pre>
133t.c:38:15: error: invalid operands to binary expression ('int *' and '_Complex float')
134 <font color="darkgreen">P = (P-42) + Gamma*4;</font>
135 <font color="blue">~~~~~~ ^ ~~~~~~~</font>
136</pre>
137
138<p>In this example, you can see the English translation, the severity (error),
139you can see the source location (the caret ("^") and file/line/column info),
140the source ranges "~~~~", arguments to the diagnostic ("int*" and "_Complex
141float"). You'll have to believe me that there is a unique ID backing the
142diagnostic :).</p>
143
144<p>Getting all of this to happen has several steps and involves many moving
145pieces, this section describes them and talks about best practices when adding
146a new diagnostic.</p>
147
Chris Lattner4c50b692010-05-01 17:35:19 +0000[diff] [blame^]148<!-- ============================== -->
149<h4>The Diagnostic*Kinds.def files</h4>
150<!-- ============================== -->
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]151
Chris Lattner4c50b692010-05-01 17:35:19 +0000[diff] [blame^]152<p>Diagnostics are created by adding an entry to one of the <tt>
153clang/Basic/Diagnostic*Kinds.def</tt> files, depending on what library will
154be using it. This file encodes the unique ID of the
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]155diagnostic (as an enum, the first argument), the severity of the diagnostic
156(second argument) and the English translation + format string.</p>
157
158<p>There is little sanity with the naming of the unique ID's right now. Some
159start with err_, warn_, ext_ to encode the severity into the name. Since the
160enum is referenced in the C++ code that produces the diagnostic, it is somewhat
161useful for it to be reasonably short.</p>
162
163<p>The severity of the diagnostic comes from the set {<tt>NOTE</tt>,
164<tt>WARNING</tt>, <tt>EXTENSION</tt>, <tt>EXTWARN</tt>, <tt>ERROR</tt>}. The
165<tt>ERROR</tt> severity is used for diagnostics indicating the program is never
166acceptable under any circumstances. When an error is emitted, the AST for the
167input code may not be fully built. The <tt>EXTENSION</tt> and <tt>EXTWARN</tt>
168severities are used for extensions to the language that Clang accepts. This
169means that Clang fully understands and can represent them in the AST, but we
170produce diagnostics to tell the user their code is non-portable. The difference
171is that the former are ignored by default, and the later warn by default. The
172<tt>WARNING</tt> severity is used for constructs that are valid in the currently
173selected source language but that are dubious in some way. The <tt>NOTE</tt>
Daniel Dunbar426b8632009-02-17 15:49:03 +0000[diff] [blame]174level is used to staple more information onto previous diagnostics.</p>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]175
176<p>These <em>severities</em> are mapped into a smaller set (the
177Diagnostic::Level enum, {<tt>Ignored</tt>, <tt>Note</tt>, <tt>Warning</tt>,
Chris Lattner0aad2972009-02-05 22:49:08 +0000[diff] [blame]178<tt>Error</tt>, <tt>Fatal</tt> }) of output <em>levels</em> by the diagnostics
Chris Lattnera180fdd2009-02-17 07:07:29 +0000[diff] [blame]179subsystem based on various configuration options. Clang internally supports a
180fully fine grained mapping mechanism that allows you to map almost any
181diagnostic to the output level that you want. The only diagnostics that cannot
182be mapped are <tt>NOTE</tt>s, which always follow the severity of the previously
183emitted diagnostic and <tt>ERROR</tt>s, which can only be mapped to
184<tt>Fatal</tt> (it is not possible to turn an error into a warning,
185for example).</p>
186
187<p>Diagnostic mappings are used in many ways. For example, if the user
188specifies <tt>-pedantic</tt>, <tt>EXTENSION</tt> maps to <tt>Warning</tt>, if
189they specify <tt>-pedantic-errors</tt>, it turns into <tt>Error</tt>. This is
190used to implement options like <tt>-Wunused_macros</tt>, <tt>-Wundef</tt> etc.
191</p>
192
193<p>
194Mapping to <tt>Fatal</tt> should only be used for diagnostics that are
195considered so severe that error recovery won't be able to recover sensibly from
196them (thus spewing a ton of bogus errors). One example of this class of error
197are failure to #include a file.
198</p>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]199
200<!-- ================= -->
201<h4>The Format String</h4>
202<!-- ================= -->
203
204<p>The format string for the diagnostic is very simple, but it has some power.
205It takes the form of a string in English with markers that indicate where and
206how arguments to the diagnostic are inserted and formatted. For example, here
207are some simple format strings:</p>
208
209<pre>
210 "binary integer literals are an extension"
211 "format string contains '\\0' within the string body"
212 "more '<b>%%</b>' conversions than data arguments"
Chris Lattner545b3682008-11-23 20:27:13 +0000[diff] [blame]213 "invalid operands to binary expression (<b>%0</b> and <b>%1</b>)"
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]214 "overloaded '<b>%0</b>' must be a <b>%select{unary|binary|unary or binary}2</b> operator"
215 " (has <b>%1</b> parameter<b>%s1</b>)"
216</pre>
217
218<p>These examples show some important points of format strings. You can use any
219 plain ASCII character in the diagnostic string except "%" without a problem,
220 but these are C strings, so you have to use and be aware of all the C escape
221 sequences (as in the second example). If you want to produce a "%" in the
222 output, use the "%%" escape sequence, like the third diagnostic. Finally,
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]223 Clang uses the "%...[digit]" sequences to specify where and how arguments to
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]224 the diagnostic are formatted.</p>
225
226<p>Arguments to the diagnostic are numbered according to how they are specified
227 by the C++ code that <a href="#producingdiag">produces them</a>, and are
228 referenced by <tt>%0</tt> .. <tt>%9</tt>. If you have more than 10 arguments
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]229 to your diagnostic, you are doing something wrong :). Unlike printf, there
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]230 is no requirement that arguments to the diagnostic end up in the output in
231 the same order as they are specified, you could have a format string with
232 <tt>"%1 %0"</tt> that swaps them, for example. The text in between the
233 percent and digit are formatting instructions. If there are no instructions,
234 the argument is just turned into a string and substituted in.</p>
235
236<p>Here are some "best practices" for writing the English format string:</p>
237
238<ul>
239<li>Keep the string short. It should ideally fit in the 80 column limit of the
240 <tt>DiagnosticKinds.def</tt> file. This avoids the diagnostic wrapping when
241 printed, and forces you to think about the important point you are conveying
242 with the diagnostic.</li>
243<li>Take advantage of location information. The user will be able to see the
244 line and location of the caret, so you don't need to tell them that the
245 problem is with the 4th argument to the function: just point to it.</li>
246<li>Do not capitalize the diagnostic string, and do not end it with a
247 period.</li>
248<li>If you need to quote something in the diagnostic string, use single
249 quotes.</li>
250</ul>
251
252<p>Diagnostics should never take random English strings as arguments: you
253shouldn't use <tt>"you have a problem with %0"</tt> and pass in things like
254<tt>"your argument"</tt> or <tt>"your return value"</tt> as arguments. Doing
255this prevents <a href="translation">translating</a> the Clang diagnostics to
256other languages (because they'll get random English words in their otherwise
257localized diagnostic). The exceptions to this are C/C++ language keywords
258(e.g. auto, const, mutable, etc) and C/C++ operators (<tt>/=</tt>). Note
259that things like "pointer" and "reference" are not keywords. On the other
260hand, you <em>can</em> include anything that comes from the user's source code,
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]261including variable names, types, labels, etc. The 'select' format can be
262used to achieve this sort of thing in a localizable way, see below.</p>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]263
264<!-- ==================================== -->
265<h4>Formatting a Diagnostic Argument</a></h4>
266<!-- ==================================== -->
267
268<p>Arguments to diagnostics are fully typed internally, and come from a couple
269different classes: integers, types, names, and random strings. Depending on
270the class of the argument, it can be optionally formatted in different ways.
271This gives the DiagnosticClient information about what the argument means
272without requiring it to use a specific presentation (consider this MVC for
273Clang :).</p>
274
275<p>Here are the different diagnostic argument formats currently supported by
276Clang:</p>
277
278<table>
279<tr><td colspan="2"><b>"s" format</b></td></tr>
280<tr><td>Example:</td><td><tt>"requires %1 parameter%s1"</tt></td></tr>
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]281<tr><td>Class:</td><td>Integers</td></tr>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]282<tr><td>Description:</td><td>This is a simple formatter for integers that is
283 useful when producing English diagnostics. When the integer is 1, it prints
284 as nothing. When the integer is not 1, it prints as "s". This allows some
Chris Lattner627b7052008-11-23 00:28:33 +0000[diff] [blame]285 simple grammatical forms to be to be handled correctly, and eliminates the
286 need to use gross things like <tt>"requires %1 parameter(s)"</tt>.</td></tr>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]287
288<tr><td colspan="2"><b>"select" format</b></td></tr>
289<tr><td>Example:</td><td><tt>"must be a %select{unary|binary|unary or binary}2
290 operator"</tt></td></tr>
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]291<tr><td>Class:</td><td>Integers</td></tr>
John McCall3a47e232010-01-14 19:12:17 +0000[diff] [blame]292<tr><td>Description:</td><td><p>This format specifier is used to merge multiple
Chris Lattnercc543342008-11-22 23:50:47 +0000[diff] [blame]293 related diagnostics together into one common one, without requiring the
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]294 difference to be specified as an English string argument. Instead of
Chris Lattnercc543342008-11-22 23:50:47 +0000[diff] [blame]295 specifying the string, the diagnostic gets an integer argument and the
296 format string selects the numbered option. In this case, the "%2" value
297 must be an integer in the range [0..2]. If it is 0, it prints 'unary', if
298 it is 1 it prints 'binary' if it is 2, it prints 'unary or binary'. This
299 allows other language translations to substitute reasonable words (or entire
300 phrases) based on the semantics of the diagnostic instead of having to do
John McCall3a47e232010-01-14 19:12:17 +0000[diff] [blame]301 things textually.</p>
302 <p>The selected string does undergo formatting.</p></td></tr>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]303
304<tr><td colspan="2"><b>"plural" format</b></td></tr>
Sebastian Redl68168562008-11-22 22:16:45 +0000[diff] [blame]305<tr><td>Example:</td><td><tt>"you have %1 %plural{1:mouse|:mice}1 connected to
306 your computer"</tt></td></tr>
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]307<tr><td>Class:</td><td>Integers</td></tr>
Sebastian Redl68168562008-11-22 22:16:45 +0000[diff] [blame]308<tr><td>Description:</td><td><p>This is a formatter for complex plural forms.
309 It is designed to handle even the requirements of languages with very
310 complex plural forms, as many Baltic languages have. The argument consists
311 of a series of expression/form pairs, separated by ':', where the first form
312 whose expression evaluates to true is the result of the modifier.</p>
313 <p>An expression can be empty, in which case it is always true. See the
314 example at the top. Otherwise, it is a series of one or more numeric
315 conditions, separated by ','. If any condition matches, the expression
316 matches. Each numeric condition can take one of three forms.</p>
317 <ul>
318 <li>number: A simple decimal number matches if the argument is the same
Chris Lattner627b7052008-11-23 00:28:33 +0000[diff] [blame]319 as the number. Example: <tt>"%plural{1:mouse|:mice}4"</tt></li>
Sebastian Redl68168562008-11-22 22:16:45 +0000[diff] [blame]320 <li>range: A range in square brackets matches if the argument is within
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]321 the range. Then range is inclusive on both ends. Example:
Chris Lattner627b7052008-11-23 00:28:33 +0000[diff] [blame]322 <tt>"%plural{0:none|1:one|[2,5]:some|:many}2"</tt></li>
323 <li>modulo: A modulo operator is followed by a number, and
324 equals sign and either a number or a range. The tests are the
325 same as for plain
Sebastian Redl68168562008-11-22 22:16:45 +0000[diff] [blame]326 numbers and ranges, but the argument is taken modulo the number first.
Chris Lattner627b7052008-11-23 00:28:33 +0000[diff] [blame]327 Example: <tt>"%plural{%100=0:even hundred|%100=[1,50]:lower half|:everything
328 else}1"</tt></li>
Sebastian Redl68168562008-11-22 22:16:45 +0000[diff] [blame]329 </ul>
330 <p>The parser is very unforgiving. A syntax error, even whitespace, will
331 abort, as will a failure to match the argument against any
332 expression.</p></td></tr>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]333
John McCall3a47e232010-01-14 19:12:17 +0000[diff] [blame]334<tr><td colspan="2"><b>"ordinal" format</b></td></tr>
335<tr><td>Example:</td><td><tt>"ambiguity in %ordinal0 argument"</tt></td></tr>
336<tr><td>Class:</td><td>Integers</td></tr>
337<tr><td>Description:</td><td><p>This is a formatter which represents the
338 argument number as an ordinal: the value <tt>1</tt> becomes <tt>1st</tt>,
339 <tt>3</tt> becomes <tt>3rd</tt>, and so on. Values less than <tt>1</tt>
340 are not supported.</p>
341 <p>This formatter is currently hard-coded to use English ordinals.</p></td></tr>
342
Chris Lattner077bf5e2008-11-24 03:33:13 +0000[diff] [blame]343<tr><td colspan="2"><b>"objcclass" format</b></td></tr>
344<tr><td>Example:</td><td><tt>"method %objcclass0 not found"</tt></td></tr>
345<tr><td>Class:</td><td>DeclarationName</td></tr>
346<tr><td>Description:</td><td><p>This is a simple formatter that indicates the
347 DeclarationName corresponds to an Objective-C class method selector. As
348 such, it prints the selector with a leading '+'.</p></td></tr>
349
350<tr><td colspan="2"><b>"objcinstance" format</b></td></tr>
351<tr><td>Example:</td><td><tt>"method %objcinstance0 not found"</tt></td></tr>
352<tr><td>Class:</td><td>DeclarationName</td></tr>
353<tr><td>Description:</td><td><p>This is a simple formatter that indicates the
354 DeclarationName corresponds to an Objective-C instance method selector. As
355 such, it prints the selector with a leading '-'.</p></td></tr>
356
Douglas Gregor47b9a1c2009-02-04 17:27:36 +0000[diff] [blame]357<tr><td colspan="2"><b>"q" format</b></td></tr>
358<tr><td>Example:</td><td><tt>"candidate found by name lookup is %q0"</tt></td></tr>
359<tr><td>Class:</td><td>NamedDecl*</td></tr>
360<tr><td>Description</td><td><p>This formatter indicates that the fully-qualified name of the declaration should be printed, e.g., "std::vector" rather than "vector".</p></td></tr>
361
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]362</table>
363
Chris Lattnercc543342008-11-22 23:50:47 +0000[diff] [blame]364<p>It is really easy to add format specifiers to the Clang diagnostics system,
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]365but they should be discussed before they are added. If you are creating a lot
366of repetitive diagnostics and/or have an idea for a useful formatter, please
367bring it up on the cfe-dev mailing list.</p>
Chris Lattnercc543342008-11-22 23:50:47 +0000[diff] [blame]368
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]369<!-- ===================================================== -->
370<h4><a name="#producingdiag">Producing the Diagnostic</a></h4>
371<!-- ===================================================== -->
372
Chris Lattner627b7052008-11-23 00:28:33 +0000[diff] [blame]373<p>Now that you've created the diagnostic in the DiagnosticKinds.def file, you
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]374need to write the code that detects the condition in question and emits the
375new diagnostic. Various components of Clang (e.g. the preprocessor, Sema,
Chris Lattner627b7052008-11-23 00:28:33 +0000[diff] [blame]376etc) provide a helper function named "Diag". It creates a diagnostic and
377accepts the arguments, ranges, and other information that goes along with
378it.</p>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]379
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]380<p>For example, the binary expression error comes from code like this:</p>
Chris Lattner627b7052008-11-23 00:28:33 +0000[diff] [blame]381
382<pre>
383 if (various things that are bad)
384 Diag(Loc, diag::err_typecheck_invalid_operands)
385 &lt;&lt; lex-&gt;getType() &lt;&lt; rex-&gt;getType()
386 &lt;&lt; lex-&gt;getSourceRange() &lt;&lt; rex-&gt;getSourceRange();
387</pre>
388
389<p>This shows that use of the Diag method: they take a location (a <a
390href="#SourceLocation">SourceLocation</a> object) and a diagnostic enum value
391(which matches the name from DiagnosticKinds.def). If the diagnostic takes
392arguments, they are specified with the &lt;&lt; operator: the first argument
393becomes %0, the second becomes %1, etc. The diagnostic interface allows you to
Chris Lattnerfd408ea2008-11-23 00:42:53 +0000[diff] [blame]394specify arguments of many different types, including <tt>int</tt> and
395<tt>unsigned</tt> for integer arguments, <tt>const char*</tt> and
396<tt>std::string</tt> for string arguments, <tt>DeclarationName</tt> and
397<tt>const IdentifierInfo*</tt> for names, <tt>QualType</tt> for types, etc.
398SourceRanges are also specified with the &lt;&lt; operator, but do not have a
399specific ordering requirement.</p>
Chris Lattner627b7052008-11-23 00:28:33 +0000[diff] [blame]400
401<p>As you can see, adding and producing a diagnostic is pretty straightforward.
402The hard part is deciding exactly what you need to say to help the user, picking
403a suitable wording, and providing the information needed to format it correctly.
Chris Lattnerfd408ea2008-11-23 00:42:53 +0000[diff] [blame]404The good news is that the call site that issues a diagnostic should be
405completely independent of how the diagnostic is formatted and in what language
406it is rendered.
Chris Lattner627b7052008-11-23 00:28:33 +0000[diff] [blame]407</p>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]408
Douglas Gregorb2fb6de2009-02-27 17:53:17 +0000[diff] [blame]409<!-- ==================================================== -->
410<h4 id="code-modification-hints">Code Modification Hints</h4>
411<!-- ==================================================== -->
412
413<p>In some cases, the front end emits diagnostics when it is clear
414that some small change to the source code would fix the problem. For
415example, a missing semicolon at the end of a statement or a use of
Chris Lattner34c05332009-02-27 19:31:12 +0000[diff] [blame]416deprecated syntax that is easily rewritten into a more modern form.
417Clang tries very hard to emit the diagnostic and recover gracefully
418in these and other cases.</p>
Douglas Gregorb2fb6de2009-02-27 17:53:17 +0000[diff] [blame]419
Chris Lattner34c05332009-02-27 19:31:12 +0000[diff] [blame]420<p>However, for these cases where the fix is obvious, the diagnostic
421can be annotated with a code
422modification "hint" that describes how to change the code referenced
Douglas Gregorb2fb6de2009-02-27 17:53:17 +0000[diff] [blame]423by the diagnostic to fix the problem. For example, it might add the
424missing semicolon at the end of the statement or rewrite the use of a
425deprecated construct into something more palatable. Here is one such
426example C++ front end, where we warn about the right-shift operator
427changing meaning from C++98 to C++0x:</p>
428
429<pre>
430test.cpp:3:7: warning: use of right-shift operator ('&gt;&gt;') in template argument will require parentheses in C++0x
431A&lt;100 &gt;&gt; 2&gt; *a;
432 ^
433 ( )
434</pre>
435
436<p>Here, the code modification hint is suggesting that parentheses be
437added, and showing exactly where those parentheses would be inserted
438into the source code. The code modification hints themselves describe
439what changes to make to the source code in an abstract manner, which
440the text diagnostic printer renders as a line of "insertions" below
441the caret line. <a href="#DiagnosticClient">Other diagnostic
442clients</a> might choose to render the code differently (e.g., as
443markup inline) or even give the user the ability to automatically fix
444the problem.</p>
445
446<p>All code modification hints are described by the
447<code>CodeModificationHint</code> class, instances of which should be
448attached to the diagnostic using the &lt;&lt; operator in the same way
449that highlighted source ranges and arguments are passed to the
450diagnostic. Code modification hints can be created with one of three
451constructors:</p>
452
453<dl>
454 <dt><code>CodeModificationHint::CreateInsertion(Loc, Code)</code></dt>
455 <dd>Specifies that the given <code>Code</code> (a string) should be inserted
456 before the source location <code>Loc</code>.</dd>
457
458 <dt><code>CodeModificationHint::CreateRemoval(Range)</code></dt>
459 <dd>Specifies that the code in the given source <code>Range</code>
460 should be removed.</dd>
461
462 <dt><code>CodeModificationHint::CreateReplacement(Range, Code)</code></dt>
463 <dd>Specifies that the code in the given source <code>Range</code>
464 should be removed, and replaced with the given <code>Code</code> string.</dd>
465</dl>
466
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]467<!-- ============================================================= -->
468<h4><a name="DiagnosticClient">The DiagnosticClient Interface</a></h4>
469<!-- ============================================================= -->
470
Chris Lattner627b7052008-11-23 00:28:33 +0000[diff] [blame]471<p>Once code generates a diagnostic with all of the arguments and the rest of
472the relevant information, Clang needs to know what to do with it. As previously
473mentioned, the diagnostic machinery goes through some filtering to map a
474severity onto a diagnostic level, then (assuming the diagnostic is not mapped to
475"<tt>Ignore</tt>") it invokes an object that implements the DiagnosticClient
476interface with the information.</p>
477
478<p>It is possible to implement this interface in many different ways. For
479example, the normal Clang DiagnosticClient (named 'TextDiagnosticPrinter') turns
480the arguments into strings (according to the various formatting rules), prints
481out the file/line/column information and the string, then prints out the line of
482code, the source ranges, and the caret. However, this behavior isn't required.
483</p>
484
485<p>Another implementation of the DiagnosticClient interface is the
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]486'TextDiagnosticBuffer' class, which is used when Clang is in -verify mode.
Chris Lattnerfd408ea2008-11-23 00:42:53 +0000[diff] [blame]487Instead of formatting and printing out the diagnostics, this implementation just
488captures and remembers the diagnostics as they fly by. Then -verify compares
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]489the list of produced diagnostics to the list of expected ones. If they disagree,
Chris Lattnerfd408ea2008-11-23 00:42:53 +0000[diff] [blame]490it prints out its own output.
Chris Lattner627b7052008-11-23 00:28:33 +0000[diff] [blame]491</p>
492
Chris Lattnerfd408ea2008-11-23 00:42:53 +0000[diff] [blame]493<p>There are many other possible implementations of this interface, and this is
494why we prefer diagnostics to pass down rich structured information in arguments.
495For example, an HTML output might want declaration names be linkified to where
496they come from in the source. Another example is that a GUI might let you click
497on typedefs to expand them. This application would want to pass significantly
498more information about types through to the GUI than a simple flat string. The
499interface allows this to happen.</p>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]500
501<!-- ====================================================== -->
502<h4><a name="translation">Adding Translations to Clang</a></h4>
503<!-- ====================================================== -->
504
Chris Lattner627b7052008-11-23 00:28:33 +0000[diff] [blame]505<p>Not possible yet! Diagnostic strings should be written in UTF-8, the client
Chris Lattnerfd408ea2008-11-23 00:42:53 +0000[diff] [blame]506can translate to the relevant code page if needed. Each translation completely
507replaces the format string for the diagnostic.</p>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]508
509
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]510<!-- ======================================================================= -->
511<h3 id="SourceLocation">The SourceLocation and SourceManager classes</h3>
512<!-- ======================================================================= -->
513
514<p>Strangely enough, the SourceLocation class represents a location within the
515source code of the program. Important design points include:</p>
516
517<ol>
518<li>sizeof(SourceLocation) must be extremely small, as these are embedded into
519 many AST nodes and are passed around often. Currently it is 32 bits.</li>
520<li>SourceLocation must be a simple value object that can be efficiently
521 copied.</li>
522<li>We should be able to represent a source location for any byte of any input
523 file. This includes in the middle of tokens, in whitespace, in trigraphs,
524 etc.</li>
525<li>A SourceLocation must encode the current #include stack that was active when
526 the location was processed. For example, if the location corresponds to a
527 token, it should contain the set of #includes active when the token was
528 lexed. This allows us to print the #include stack for a diagnostic.</li>
529<li>SourceLocation must be able to describe macro expansions, capturing both
530 the ultimate instantiation point and the source of the original character
531 data.</li>
532</ol>
533
534<p>In practice, the SourceLocation works together with the SourceManager class
Chris Lattner18376dd2009-01-16 07:00:50 +0000[diff] [blame]535to encode two pieces of information about a location: it's spelling location
Chris Lattner88054de2009-01-16 07:15:35 +0000[diff] [blame]536and it's instantiation location. For most tokens, these will be the same. However,
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]537for a macro expansion (or tokens that came from a _Pragma directive) these will
538describe the location of the characters corresponding to the token and the
539location where the token was used (i.e. the macro instantiation point or the
540location of the _Pragma itself).</p>
541
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]542<p>The Clang front-end inherently depends on the location of a token being
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]543tracked correctly. If it is ever incorrect, the front-end may get confused and
544die. The reason for this is that the notion of the 'spelling' of a Token in
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]545Clang depends on being able to find the original input characters for the token.
Chris Lattner18376dd2009-01-16 07:00:50 +0000[diff] [blame]546This concept maps directly to the "spelling location" for the token.</p>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]547
548<!-- ======================================================================= -->
Daniel Dunbar27d9e9f2009-03-30 06:50:01 +0000[diff] [blame]549<h2 id="libdriver">The Driver Library</h2>
550<!-- ======================================================================= -->
551
Ted Kremenekcfa8d572009-04-09 18:08:18 +0000[diff] [blame]552<p>The clang Driver and library are documented <a
553href="DriverInternals.html">here<a>.<p>
554
555<!-- ======================================================================= -->
Douglas Gregor32110df2009-05-20 00:16:32 +0000[diff] [blame]556<h2 id="pch">Precompiled Headers</h2>
Ted Kremenekcfa8d572009-04-09 18:08:18 +0000[diff] [blame]557<!-- ======================================================================= -->
558
Douglas Gregor32110df2009-05-20 00:16:32 +0000[diff] [blame]559<p>Clang supports two implementations of precompiled headers. The
560 default implementation, precompiled headers (<a
561 href="PCHInternals.html">PCH</a>) uses a serialized representation
562 of Clang's internal data structures, encoded with the <a
563 href="http://llvm.org/docs/BitCodeFormat.html">LLVM bitstream
564 format</a>. Pretokenized headers (<a
565 href="PTHInternals.html">PTH</a>), on the other hand, contain a
566 serialized representation of the tokens encountered when
567 preprocessing a header (and anything that header includes).</p>
568
Daniel Dunbar27d9e9f2009-03-30 06:50:01 +0000[diff] [blame]569
570<!-- ======================================================================= -->
571<h2 id="libfrontend">The Frontend Library</h2>
572<!-- ======================================================================= -->
573
574<p>The Frontend library contains functionality useful for building
575tools on top of the clang libraries, for example several methods for
576outputting diagnostics.</p>
577
578<!-- ======================================================================= -->
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]579<h2 id="liblex">The Lexer and Preprocessor Library</h2>
580<!-- ======================================================================= -->
581
582<p>The Lexer library contains several tightly-connected classes that are involved
583with the nasty process of lexing and preprocessing C source code. The main
584interface to this library for outside clients is the large <a
585href="#Preprocessor">Preprocessor</a> class.
586It contains the various pieces of state that are required to coherently read
587tokens out of a translation unit.</p>
588
589<p>The core interface to the Preprocessor object (once it is set up) is the
590Preprocessor::Lex method, which returns the next <a href="#Token">Token</a> from
591the preprocessor stream. There are two types of token providers that the
592preprocessor is capable of reading from: a buffer lexer (provided by the <a
593href="#Lexer">Lexer</a> class) and a buffered token stream (provided by the <a
Chris Lattner79281252008-03-09 02:27:26 +0000[diff] [blame]594href="#TokenLexer">TokenLexer</a> class).
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]595
596
597<!-- ======================================================================= -->
598<h3 id="Token">The Token class</h3>
599<!-- ======================================================================= -->
600
601<p>The Token class is used to represent a single lexed token. Tokens are
602intended to be used by the lexer/preprocess and parser libraries, but are not
603intended to live beyond them (for example, they should not live in the ASTs).<p>
604
605<p>Tokens most often live on the stack (or some other location that is efficient
606to access) as the parser is running, but occasionally do get buffered up. For
607example, macro definitions are stored as a series of tokens, and the C++
Chris Lattner3fcbb892008-11-23 08:32:53 +0000[diff] [blame]608front-end periodically needs to buffer tokens up for tentative parsing and
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]609various pieces of look-ahead. As such, the size of a Token matter. On a 32-bit
610system, sizeof(Token) is currently 16 bytes.</p>
611
Chris Lattner3932fe02009-01-06 06:02:08 +0000[diff] [blame]612<p>Tokens occur in two forms: "<a href="#AnnotationToken">Annotation
613Tokens</a>" and normal tokens. Normal tokens are those returned by the lexer,
614annotation tokens represent semantic information and are produced by the parser,
615replacing normal tokens in the token stream. Normal tokens contain the
616following information:</p>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]617
618<ul>
619<li><b>A SourceLocation</b> - This indicates the location of the start of the
620token.</li>
621
622<li><b>A length</b> - This stores the length of the token as stored in the
623SourceBuffer. For tokens that include them, this length includes trigraphs and
624escaped newlines which are ignored by later phases of the compiler. By pointing
625into the original source buffer, it is always possible to get the original
626spelling of a token completely accurately.</li>
627
628<li><b>IdentifierInfo</b> - If a token takes the form of an identifier, and if
629identifier lookup was enabled when the token was lexed (e.g. the lexer was not
630reading in 'raw' mode) this contains a pointer to the unique hash value for the
631identifier. Because the lookup happens before keyword identification, this
632field is set even for language keywords like 'for'.</li>
633
634<li><b>TokenKind</b> - This indicates the kind of token as classified by the
635lexer. This includes things like <tt>tok::starequal</tt> (for the "*="
636operator), <tt>tok::ampamp</tt> for the "&amp;&amp;" token, and keyword values
637(e.g. <tt>tok::kw_for</tt>) for identifiers that correspond to keywords. Note
638that some tokens can be spelled multiple ways. For example, C++ supports
639"operator keywords", where things like "and" are treated exactly like the
640"&amp;&amp;" operator. In these cases, the kind value is set to
641<tt>tok::ampamp</tt>, which is good for the parser, which doesn't have to
642consider both forms. For something that cares about which form is used (e.g.
643the preprocessor 'stringize' operator) the spelling indicates the original
644form.</li>
645
646<li><b>Flags</b> - There are currently four flags tracked by the
647lexer/preprocessor system on a per-token basis:
648
649 <ol>
650 <li><b>StartOfLine</b> - This was the first token that occurred on its input
651 source line.</li>
652 <li><b>LeadingSpace</b> - There was a space character either immediately
653 before the token or transitively before the token as it was expanded
654 through a macro. The definition of this flag is very closely defined by
655 the stringizing requirements of the preprocessor.</li>
656 <li><b>DisableExpand</b> - This flag is used internally to the preprocessor to
657 represent identifier tokens which have macro expansion disabled. This
658 prevents them from being considered as candidates for macro expansion ever
659 in the future.</li>
660 <li><b>NeedsCleaning</b> - This flag is set if the original spelling for the
661 token includes a trigraph or escaped newline. Since this is uncommon,
662 many pieces of code can fast-path on tokens that did not need cleaning.
663 </p>
664 </ol>
665</li>
666</ul>
667
Chris Lattner3932fe02009-01-06 06:02:08 +0000[diff] [blame]668<p>One interesting (and somewhat unusual) aspect of normal tokens is that they
669don't contain any semantic information about the lexed value. For example, if
670the token was a pp-number token, we do not represent the value of the number
671that was lexed (this is left for later pieces of code to decide). Additionally,
672the lexer library has no notion of typedef names vs variable names: both are
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]673returned as identifiers, and the parser is left to decide whether a specific
674identifier is a typedef or a variable (tracking this requires scope information
Chris Lattner3932fe02009-01-06 06:02:08 +0000[diff] [blame]675among other things). The parser can do this translation by replacing tokens
676returned by the preprocessor with "Annotation Tokens".</p>
677
678<!-- ======================================================================= -->
679<h3 id="AnnotationToken">Annotation Tokens</h3>
680<!-- ======================================================================= -->
681
682<p>Annotation Tokens are tokens that are synthesized by the parser and injected
683into the preprocessor's token stream (replacing existing tokens) to record
684semantic information found by the parser. For example, if "foo" is found to be
685a typedef, the "foo" <tt>tok::identifier</tt> token is replaced with an
686<tt>tok::annot_typename</tt>. This is useful for a couple of reasons: 1) this
687makes it easy to handle qualified type names (e.g. "foo::bar::baz&lt;42&gt;::t")
688in C++ as a single "token" in the parser. 2) if the parser backtracks, the
689reparse does not need to redo semantic analysis to determine whether a token
690sequence is a variable, type, template, etc.</p>
691
692<p>Annotation Tokens are created by the parser and reinjected into the parser's
693token stream (when backtracking is enabled). Because they can only exist in
694tokens that the preprocessor-proper is done with, it doesn't need to keep around
695flags like "start of line" that the preprocessor uses to do its job.
696Additionally, an annotation token may "cover" a sequence of preprocessor tokens
697(e.g. <tt>a::b::c</tt> is five preprocessor tokens). As such, the valid fields
698of an annotation token are different than the fields for a normal token (but
699they are multiplexed into the normal Token fields):</p>
700
701<ul>
702<li><b>SourceLocation "Location"</b> - The SourceLocation for the annotation
703token indicates the first token replaced by the annotation token. In the example
704above, it would be the location of the "a" identifier.</li>
705
706<li><b>SourceLocation "AnnotationEndLoc"</b> - This holds the location of the
707last token replaced with the annotation token. In the example above, it would
708be the location of the "c" identifier.</li>
709
710<li><b>void* "AnnotationValue"</b> - This contains an opaque object that the
711parser gets from Sema through an Actions module, it is passed around and Sema
712intepretes it, based on the type of annotation token.</li>
713
714<li><b>TokenKind "Kind"</b> - This indicates the kind of Annotation token this
715is. See below for the different valid kinds.</li>
716</ul>
717
718<p>Annotation tokens currently come in three kinds:</p>
719
720<ol>
721<li><b>tok::annot_typename</b>: This annotation token represents a
722resolved typename token that is potentially qualified. The AnnotationValue
Steve Naroffb43a50f2009-01-28 19:39:02 +0000[diff] [blame]723field contains a pointer returned by Action::getTypeName(). In the case of the
Chris Lattner3932fe02009-01-06 06:02:08 +0000[diff] [blame]724Sema actions module, this is a <tt>Decl*</tt> for the type.</li>
725
726<li><b>tok::annot_cxxscope</b>: This annotation token represents a C++ scope
727specifier, such as "A::B::". This corresponds to the grammar productions "::"
728and ":: [opt] nested-name-specifier". The AnnotationValue pointer is returned
729by the Action::ActOnCXXGlobalScopeSpecifier and
730Action::ActOnCXXNestedNameSpecifier callbacks. In the case of Sema, this is a
731<tt>DeclContext*</tt>.</li>
732
Douglas Gregor39a8de12009-02-25 19:37:18 +0000[diff] [blame]733<li><b>tok::annot_template_id</b>: This annotation token represents a
734C++ template-id such as "foo&lt;int, 4&gt;", where "foo" is the name
735of a template. The AnnotationValue pointer is a pointer to a malloc'd
736TemplateIdAnnotation object. Depending on the context, a parsed template-id that names a type might become a typename annotation token (if all we care about is the named type, e.g., because it occurs in a type specifier) or might remain a template-id token (if we want to retain more source location information or produce a new type, e.g., in a declaration of a class template specialization). template-id annotation tokens that refer to a type can be "upgraded" to typename annotation tokens by the parser.</li>
Chris Lattner3932fe02009-01-06 06:02:08 +0000[diff] [blame]737
738</ol>
739
Cedric Venetda76b282009-01-06 16:22:54 +0000[diff] [blame]740<p>As mentioned above, annotation tokens are not returned by the preprocessor,
Chris Lattner3932fe02009-01-06 06:02:08 +0000[diff] [blame]741they are formed on demand by the parser. This means that the parser has to be
742aware of cases where an annotation could occur and form it where appropriate.
743This is somewhat similar to how the parser handles Translation Phase 6 of C99:
744String Concatenation (see C99 5.1.1.2). In the case of string concatenation,
745the preprocessor just returns distinct tok::string_literal and
746tok::wide_string_literal tokens and the parser eats a sequence of them wherever
747the grammar indicates that a string literal can occur.</p>
748
749<p>In order to do this, whenever the parser expects a tok::identifier or
750tok::coloncolon, it should call the TryAnnotateTypeOrScopeToken or
751TryAnnotateCXXScopeToken methods to form the annotation token. These methods
752will maximally form the specified annotation tokens and replace the current
753token with them, if applicable. If the current tokens is not valid for an
754annotation token, it will remain an identifier or :: token.</p>
755
756
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]757
758<!-- ======================================================================= -->
759<h3 id="Lexer">The Lexer class</h3>
760<!-- ======================================================================= -->
761
762<p>The Lexer class provides the mechanics of lexing tokens out of a source
763buffer and deciding what they mean. The Lexer is complicated by the fact that
764it operates on raw buffers that have not had spelling eliminated (this is a
765necessity to get decent performance), but this is countered with careful coding
766as well as standard performance techniques (for example, the comment handling
767code is vectorized on X86 and PowerPC hosts).</p>
768
769<p>The lexer has a couple of interesting modal features:</p>
770
771<ul>
772<li>The lexer can operate in 'raw' mode. This mode has several features that
773 make it possible to quickly lex the file (e.g. it stops identifier lookup,
774 doesn't specially handle preprocessor tokens, handles EOF differently, etc).
775 This mode is used for lexing within an "<tt>#if 0</tt>" block, for
776 example.</li>
777<li>The lexer can capture and return comments as tokens. This is required to
778 support the -C preprocessor mode, which passes comments through, and is
779 used by the diagnostic checker to identifier expect-error annotations.</li>
780<li>The lexer can be in ParsingFilename mode, which happens when preprocessing
Chris Lattner84386242007-09-16 19:25:23 +0000[diff] [blame]781 after reading a #include directive. This mode changes the parsing of '&lt;'
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]782 to return an "angled string" instead of a bunch of tokens for each thing
783 within the filename.</li>
784<li>When parsing a preprocessor directive (after "<tt>#</tt>") the
785 ParsingPreprocessorDirective mode is entered. This changes the parser to
786 return EOM at a newline.</li>
787<li>The Lexer uses a LangOptions object to know whether trigraphs are enabled,
788 whether C++ or ObjC keywords are recognized, etc.</li>
789</ul>
790
791<p>In addition to these modes, the lexer keeps track of a couple of other
792 features that are local to a lexed buffer, which change as the buffer is
793 lexed:</p>
794
795<ul>
796<li>The Lexer uses BufferPtr to keep track of the current character being
797 lexed.</li>
798<li>The Lexer uses IsAtStartOfLine to keep track of whether the next lexed token
799 will start with its "start of line" bit set.</li>
800<li>The Lexer keeps track of the current #if directives that are active (which
801 can be nested).</li>
802<li>The Lexer keeps track of an <a href="#MultipleIncludeOpt">
803 MultipleIncludeOpt</a> object, which is used to
804 detect whether the buffer uses the standard "<tt>#ifndef XX</tt> /
805 <tt>#define XX</tt>" idiom to prevent multiple inclusion. If a buffer does,
806 subsequent includes can be ignored if the XX macro is defined.</li>
807</ul>
808
809<!-- ======================================================================= -->
Chris Lattner79281252008-03-09 02:27:26 +0000[diff] [blame]810<h3 id="TokenLexer">The TokenLexer class</h3>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]811<!-- ======================================================================= -->
812
Chris Lattner79281252008-03-09 02:27:26 +0000[diff] [blame]813<p>The TokenLexer class is a token provider that returns tokens from a list
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]814of tokens that came from somewhere else. It typically used for two things: 1)
815returning tokens from a macro definition as it is being expanded 2) returning
816tokens from an arbitrary buffer of tokens. The later use is used by _Pragma and
817will most likely be used to handle unbounded look-ahead for the C++ parser.</p>
818
819<!-- ======================================================================= -->
820<h3 id="MultipleIncludeOpt">The MultipleIncludeOpt class</h3>
821<!-- ======================================================================= -->
822
823<p>The MultipleIncludeOpt class implements a really simple little state machine
824that is used to detect the standard "<tt>#ifndef XX</tt> / <tt>#define XX</tt>"
825idiom that people typically use to prevent multiple inclusion of headers. If a
826buffer uses this idiom and is subsequently #include'd, the preprocessor can
827simply check to see whether the guarding condition is defined or not. If so,
828the preprocessor can completely ignore the include of the header.</p>
829
830
831
832<!-- ======================================================================= -->
833<h2 id="libparse">The Parser Library</h2>
834<!-- ======================================================================= -->
835
836<!-- ======================================================================= -->
837<h2 id="libast">The AST Library</h2>
838<!-- ======================================================================= -->
839
840<!-- ======================================================================= -->
841<h3 id="Type">The Type class and its subclasses</h3>
842<!-- ======================================================================= -->
843
844<p>The Type class (and its subclasses) are an important part of the AST. Types
845are accessed through the ASTContext class, which implicitly creates and uniques
846them as they are needed. Types have a couple of non-obvious features: 1) they
847do not capture type qualifiers like const or volatile (See
848<a href="#QualType">QualType</a>), and 2) they implicitly capture typedef
Chris Lattner8a2bc622007-07-31 06:37:39 +0000[diff] [blame]849information. Once created, types are immutable (unlike decls).</p>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]850
851<p>Typedefs in C make semantic analysis a bit more complex than it would
852be without them. The issue is that we want to capture typedef information
853and represent it in the AST perfectly, but the semantics of operations need to
854"see through" typedefs. For example, consider this code:</p>
855
856<code>
857void func() {<br>
Bill Wendling30d17752007-10-06 01:56:01 +0000[diff] [blame]858&nbsp;&nbsp;typedef int foo;<br>
859&nbsp;&nbsp;foo X, *Y;<br>
860&nbsp;&nbsp;typedef foo* bar;<br>
861&nbsp;&nbsp;bar Z;<br>
862&nbsp;&nbsp;*X; <i>// error</i><br>
863&nbsp;&nbsp;**Y; <i>// error</i><br>
864&nbsp;&nbsp;**Z; <i>// error</i><br>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]865}<br>
866</code>
867
868<p>The code above is illegal, and thus we expect there to be diagnostics emitted
869on the annotated lines. In this example, we expect to get:</p>
870
871<pre>
Chris Lattner8a2bc622007-07-31 06:37:39 +0000[diff] [blame]872<b>test.c:6:1: error: indirection requires pointer operand ('foo' invalid)</b>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]873*X; // error
874<font color="blue">^~</font>
Chris Lattner8a2bc622007-07-31 06:37:39 +0000[diff] [blame]875<b>test.c:7:1: error: indirection requires pointer operand ('foo' invalid)</b>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]876**Y; // error
877<font color="blue">^~~</font>
Chris Lattner8a2bc622007-07-31 06:37:39 +0000[diff] [blame]878<b>test.c:8:1: error: indirection requires pointer operand ('foo' invalid)</b>
879**Z; // error
880<font color="blue">^~~</font>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]881</pre>
882
883<p>While this example is somewhat silly, it illustrates the point: we want to
884retain typedef information where possible, so that we can emit errors about
885"<tt>std::string</tt>" instead of "<tt>std::basic_string&lt;char, std:...</tt>".
886Doing this requires properly keeping typedef information (for example, the type
887of "X" is "foo", not "int"), and requires properly propagating it through the
Chris Lattner8a2bc622007-07-31 06:37:39 +0000[diff] [blame]888various operators (for example, the type of *Y is "foo", not "int"). In order
889to retain this information, the type of these expressions is an instance of the
890TypedefType class, which indicates that the type of these expressions is a
891typedef for foo.
892</p>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]893
Chris Lattner8a2bc622007-07-31 06:37:39 +0000[diff] [blame]894<p>Representing types like this is great for diagnostics, because the
895user-specified type is always immediately available. There are two problems
896with this: first, various semantic checks need to make judgements about the
Chris Lattner33fc68a2007-07-31 18:54:50 +0000[diff] [blame]897<em>actual structure</em> of a type, ignoring typdefs. Second, we need an
898efficient way to query whether two types are structurally identical to each
899other, ignoring typedefs. The solution to both of these problems is the idea of
Chris Lattner8a2bc622007-07-31 06:37:39 +0000[diff] [blame]900canonical types.</p>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]901
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]902<!-- =============== -->
Chris Lattner8a2bc622007-07-31 06:37:39 +0000[diff] [blame]903<h4>Canonical Types</h4>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]904<!-- =============== -->
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]905
Chris Lattner8a2bc622007-07-31 06:37:39 +0000[diff] [blame]906<p>Every instance of the Type class contains a canonical type pointer. For
907simple types with no typedefs involved (e.g. "<tt>int</tt>", "<tt>int*</tt>",
908"<tt>int**</tt>"), the type just points to itself. For types that have a
909typedef somewhere in their structure (e.g. "<tt>foo</tt>", "<tt>foo*</tt>",
910"<tt>foo**</tt>", "<tt>bar</tt>"), the canonical type pointer points to their
911structurally equivalent type without any typedefs (e.g. "<tt>int</tt>",
912"<tt>int*</tt>", "<tt>int**</tt>", and "<tt>int*</tt>" respectively).</p>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]913
Chris Lattner8a2bc622007-07-31 06:37:39 +0000[diff] [blame]914<p>This design provides a constant time operation (dereferencing the canonical
915type pointer) that gives us access to the structure of types. For example,
916we can trivially tell that "bar" and "foo*" are the same type by dereferencing
917their canonical type pointers and doing a pointer comparison (they both point
918to the single "<tt>int*</tt>" type).</p>
919
920<p>Canonical types and typedef types bring up some complexities that must be
921carefully managed. Specifically, the "isa/cast/dyncast" operators generally
922shouldn't be used in code that is inspecting the AST. For example, when type
923checking the indirection operator (unary '*' on a pointer), the type checker
924must verify that the operand has a pointer type. It would not be correct to
925check that with "<tt>isa&lt;PointerType&gt;(SubExpr-&gt;getType())</tt>",
926because this predicate would fail if the subexpression had a typedef type.</p>
927
928<p>The solution to this problem are a set of helper methods on Type, used to
929check their properties. In this case, it would be correct to use
930"<tt>SubExpr-&gt;getType()-&gt;isPointerType()</tt>" to do the check. This
931predicate will return true if the <em>canonical type is a pointer</em>, which is
932true any time the type is structurally a pointer type. The only hard part here
933is remembering not to use the <tt>isa/cast/dyncast</tt> operations.</p>
934
935<p>The second problem we face is how to get access to the pointer type once we
936know it exists. To continue the example, the result type of the indirection
937operator is the pointee type of the subexpression. In order to determine the
938type, we need to get the instance of PointerType that best captures the typedef
939information in the program. If the type of the expression is literally a
940PointerType, we can return that, otherwise we have to dig through the
941typedefs to find the pointer type. For example, if the subexpression had type
942"<tt>foo*</tt>", we could return that type as the result. If the subexpression
943had type "<tt>bar</tt>", we want to return "<tt>foo*</tt>" (note that we do
944<em>not</em> want "<tt>int*</tt>"). In order to provide all of this, Type has
Chris Lattner11406c12007-07-31 16:50:51 +0000[diff] [blame]945a getAsPointerType() method that checks whether the type is structurally a
Chris Lattner8a2bc622007-07-31 06:37:39 +0000[diff] [blame]946PointerType and, if so, returns the best one. If not, it returns a null
947pointer.</p>
948
949<p>This structure is somewhat mystical, but after meditating on it, it will
950make sense to you :).</p>
Chris Lattner86920d32007-07-31 05:42:17 +0000[diff] [blame]951
952<!-- ======================================================================= -->
953<h3 id="QualType">The QualType class</h3>
954<!-- ======================================================================= -->
955
956<p>The QualType class is designed as a trivial value class that is small,
957passed by-value and is efficient to query. The idea of QualType is that it
958stores the type qualifiers (const, volatile, restrict) separately from the types
959themselves: QualType is conceptually a pair of "Type*" and bits for the type
960qualifiers.</p>
961
962<p>By storing the type qualifiers as bits in the conceptual pair, it is
963extremely efficient to get the set of qualifiers on a QualType (just return the
964field of the pair), add a type qualifier (which is a trivial constant-time
965operation that sets a bit), and remove one or more type qualifiers (just return
966a QualType with the bitfield set to empty).</p>
967
968<p>Further, because the bits are stored outside of the type itself, we do not
969need to create duplicates of types with different sets of qualifiers (i.e. there
970is only a single heap allocated "int" type: "const int" and "volatile const int"
971both point to the same heap allocated "int" type). This reduces the heap size
972used to represent bits and also means we do not have to consider qualifiers when
973uniquing types (<a href="#Type">Type</a> does not even contain qualifiers).</p>
974
975<p>In practice, on hosts where it is safe, the 3 type qualifiers are stored in
976the low bit of the pointer to the Type object. This means that QualType is
977exactly the same size as a pointer, and this works fine on any system where
978malloc'd objects are at least 8 byte aligned.</p>
Ted Kremenek8bc05712007-10-10 23:01:43 +0000[diff] [blame]979
980<!-- ======================================================================= -->
Douglas Gregor2e1cd422008-11-17 14:58:09 +0000[diff] [blame]981<h3 id="DeclarationName">Declaration names</h3>
982<!-- ======================================================================= -->
983
984<p>The <tt>DeclarationName</tt> class represents the name of a
985 declaration in Clang. Declarations in the C family of languages can
Chris Lattner3fcbb892008-11-23 08:32:53 +0000[diff] [blame]986 take several different forms. Most declarations are named by
Douglas Gregor2e1cd422008-11-17 14:58:09 +0000[diff] [blame]987 simple identifiers, e.g., "<code>f</code>" and "<code>x</code>" in
988 the function declaration <code>f(int x)</code>. In C++, declaration
989 names can also name class constructors ("<code>Class</code>"
990 in <code>struct Class { Class(); }</code>), class destructors
991 ("<code>~Class</code>"), overloaded operator names ("operator+"),
992 and conversion functions ("<code>operator void const *</code>"). In
993 Objective-C, declaration names can refer to the names of Objective-C
994 methods, which involve the method name and the parameters,
Chris Lattner3fcbb892008-11-23 08:32:53 +0000[diff] [blame]995 collectively called a <i>selector</i>, e.g.,
Douglas Gregor2e1cd422008-11-17 14:58:09 +0000[diff] [blame]996 "<code>setWidth:height:</code>". Since all of these kinds of
Chris Lattner3fcbb892008-11-23 08:32:53 +0000[diff] [blame]997 entities - variables, functions, Objective-C methods, C++
998 constructors, destructors, and operators - are represented as
Douglas Gregor2e1cd422008-11-17 14:58:09 +0000[diff] [blame]999 subclasses of Clang's common <code>NamedDecl</code>
1000 class, <code>DeclarationName</code> is designed to efficiently
1001 represent any kind of name.</p>
1002
1003<p>Given
1004 a <code>DeclarationName</code> <code>N</code>, <code>N.getNameKind()</code>
Douglas Gregor2def4832008-11-17 20:34:05 +0000[diff] [blame]1005 will produce a value that describes what kind of name <code>N</code>
Douglas Gregore94ca9e42008-11-18 14:39:36 +0000[diff] [blame]1006 stores. There are 8 options (all of the names are inside
Douglas Gregor2e1cd422008-11-17 14:58:09 +0000[diff] [blame]1007 the <code>DeclarationName</code> class)</p>
1008<dl>
1009 <dt>Identifier</dt>
1010 <dd>The name is a simple
1011 identifier. Use <code>N.getAsIdentifierInfo()</code> to retrieve the
1012 corresponding <code>IdentifierInfo*</code> pointing to the actual
1013 identifier. Note that C++ overloaded operators (e.g.,
1014 "<code>operator+</code>") are represented as special kinds of
1015 identifiers. Use <code>IdentifierInfo</code>'s <code>getOverloadedOperatorID</code>
1016 function to determine whether an identifier is an overloaded
1017 operator name.</dd>
1018
1019 <dt>ObjCZeroArgSelector, ObjCOneArgSelector,
1020 ObjCMultiArgSelector</dt>
1021 <dd>The name is an Objective-C selector, which can be retrieved as a
1022 <code>Selector</code> instance
1023 via <code>N.getObjCSelector()</code>. The three possible name
1024 kinds for Objective-C reflect an optimization within
1025 the <code>DeclarationName</code> class: both zero- and
1026 one-argument selectors are stored as a
1027 masked <code>IdentifierInfo</code> pointer, and therefore require
1028 very little space, since zero- and one-argument selectors are far
1029 more common than multi-argument selectors (which use a different
1030 structure).</dd>
1031
1032 <dt>CXXConstructorName</dt>
1033 <dd>The name is a C++ constructor
1034 name. Use <code>N.getCXXNameType()</code> to retrieve
1035 the <a href="#QualType">type</a> that this constructor is meant to
1036 construct. The type is always the canonical type, since all
1037 constructors for a given type have the same name.</dd>
1038
1039 <dt>CXXDestructorName</dt>
1040 <dd>The name is a C++ destructor
1041 name. Use <code>N.getCXXNameType()</code> to retrieve
1042 the <a href="#QualType">type</a> whose destructor is being
1043 named. This type is always a canonical type.</dd>
1044
1045 <dt>CXXConversionFunctionName</dt>
1046 <dd>The name is a C++ conversion function. Conversion functions are
1047 named according to the type they convert to, e.g., "<code>operator void
1048 const *</code>". Use <code>N.getCXXNameType()</code> to retrieve
1049 the type that this conversion function converts to. This type is
1050 always a canonical type.</dd>
Douglas Gregore94ca9e42008-11-18 14:39:36 +0000[diff] [blame]1051
1052 <dt>CXXOperatorName</dt>
1053 <dd>The name is a C++ overloaded operator name. Overloaded operators
1054 are named according to their spelling, e.g.,
1055 "<code>operator+</code>" or "<code>operator new
1056 []</code>". Use <code>N.getCXXOverloadedOperator()</code> to
1057 retrieve the overloaded operator (a value of
1058 type <code>OverloadedOperatorKind</code>).</dd>
Douglas Gregor2e1cd422008-11-17 14:58:09 +0000[diff] [blame]1059</dl>
1060
1061<p><code>DeclarationName</code>s are cheap to create, copy, and
1062 compare. They require only a single pointer's worth of storage in
Douglas Gregore94ca9e42008-11-18 14:39:36 +0000[diff] [blame]1063 the common cases (identifiers, zero-
Douglas Gregor2e1cd422008-11-17 14:58:09 +0000[diff] [blame]1064 and one-argument Objective-C selectors) and use dense, uniqued
1065 storage for the other kinds of
1066 names. Two <code>DeclarationName</code>s can be compared for
1067 equality (<code>==</code>, <code>!=</code>) using a simple bitwise
1068 comparison, can be ordered
1069 with <code>&lt;</code>, <code>&gt;</code>, <code>&lt;=</code>,
1070 and <code>&gt;=</code> (which provide a lexicographical ordering for
1071 normal identifiers but an unspecified ordering for other kinds of
1072 names), and can be placed into LLVM <code>DenseMap</code>s
1073 and <code>DenseSet</code>s.</p>
1074
1075<p><code>DeclarationName</code> instances can be created in different
1076 ways depending on what kind of name the instance will store. Normal
Douglas Gregore94ca9e42008-11-18 14:39:36 +0000[diff] [blame]1077 identifiers (<code>IdentifierInfo</code> pointers) and Objective-C selectors
Douglas Gregor2e1cd422008-11-17 14:58:09 +0000[diff] [blame]1078 (<code>Selector</code>) can be implicitly converted
1079 to <code>DeclarationName</code>s. Names for C++ constructors,
Douglas Gregore94ca9e42008-11-18 14:39:36 +0000[diff] [blame]1080 destructors, conversion functions, and overloaded operators can be retrieved from
Douglas Gregor2e1cd422008-11-17 14:58:09 +0000[diff] [blame]1081 the <code>DeclarationNameTable</code>, an instance of which is
1082 available as <code>ASTContext::DeclarationNames</code>. The member
1083 functions <code>getCXXConstructorName</code>, <code>getCXXDestructorName</code>,
Douglas Gregore94ca9e42008-11-18 14:39:36 +0000[diff] [blame]1084 <code>getCXXConversionFunctionName</code>, and <code>getCXXOperatorName</code>, respectively,
1085 return <code>DeclarationName</code> instances for the four kinds of
Douglas Gregor2e1cd422008-11-17 14:58:09 +0000[diff] [blame]1086 C++ special function names.</p>
1087
1088<!-- ======================================================================= -->
Douglas Gregor074149e2009-01-05 19:45:36 +0000[diff] [blame]1089<h3 id="DeclContext">Declaration contexts</h3>
1090<!-- ======================================================================= -->
1091<p>Every declaration in a program exists within some <i>declaration
1092 context</i>, such as a translation unit, namespace, class, or
1093 function. Declaration contexts in Clang are represented by
1094 the <code>DeclContext</code> class, from which the various
1095 declaration-context AST nodes
1096 (<code>TranslationUnitDecl</code>, <code>NamespaceDecl</code>, <code>RecordDecl</code>, <code>FunctionDecl</code>,
1097 etc.) will derive. The <code>DeclContext</code> class provides
1098 several facilities common to each declaration context:</p>
1099<dl>
1100 <dt>Source-centric vs. Semantics-centric View of Declarations</dt>
1101 <dd><code>DeclContext</code> provides two views of the declarations
1102 stored within a declaration context. The source-centric view
1103 accurately represents the program source code as written, including
1104 multiple declarations of entities where present (see the
1105 section <a href="#Redeclarations">Redeclarations and
1106 Overloads</a>), while the semantics-centric view represents the
1107 program semantics. The two views are kept synchronized by semantic
1108 analysis while the ASTs are being constructed.</dd>
1109
1110 <dt>Storage of declarations within that context</dt>
1111 <dd>Every declaration context can contain some number of
1112 declarations. For example, a C++ class (represented
1113 by <code>RecordDecl</code>) contains various member functions,
1114 fields, nested types, and so on. All of these declarations will be
1115 stored within the <code>DeclContext</code>, and one can iterate
1116 over the declarations via
1117 [<code>DeclContext::decls_begin()</code>,
1118 <code>DeclContext::decls_end()</code>). This mechanism provides
1119 the source-centric view of declarations in the context.</dd>
1120
1121 <dt>Lookup of declarations within that context</dt>
1122 <dd>The <code>DeclContext</code> structure provides efficient name
1123 lookup for names within that declaration context. For example,
1124 if <code>N</code> is a namespace we can look for the
1125 name <code>N::f</code>
1126 using <code>DeclContext::lookup</code>. The lookup itself is
1127 based on a lazily-constructed array (for declaration contexts
1128 with a small number of declarations) or hash table (for
1129 declaration contexts with more declarations). The lookup
1130 operation provides the semantics-centric view of the declarations
1131 in the context.</dd>
1132
1133 <dt>Ownership of declarations</dt>
1134 <dd>The <code>DeclContext</code> owns all of the declarations that
1135 were declared within its declaration context, and is responsible
1136 for the management of their memory as well as their
1137 (de-)serialization.</dd>
1138</dl>
1139
Douglas Gregor4afa39d2009-01-20 01:17:11 +0000[diff] [blame]1140<p>All declarations are stored within a declaration context, and one
1141 can query
1142 information about the context in which each declaration lives. One
Douglas Gregor074149e2009-01-05 19:45:36 +0000[diff] [blame]1143 can retrieve the <code>DeclContext</code> that contains a
Douglas Gregor4afa39d2009-01-20 01:17:11 +0000[diff] [blame]1144 particular <code>Decl</code>
1145 using <code>Decl::getDeclContext</code>. However, see the
Douglas Gregor074149e2009-01-05 19:45:36 +0000[diff] [blame]1146 section <a href="#LexicalAndSemanticContexts">Lexical and Semantic
1147 Contexts</a> for more information about how to interpret this
1148 context information.</p>
1149
1150<h4 id="Redeclarations">Redeclarations and Overloads</h4>
1151<p>Within a translation unit, it is common for an entity to be
1152declared several times. For example, we might declare a function "f"
1153 and then later re-declare it as part of an inlined definition:</p>
1154
1155<pre>
1156void f(int x, int y, int z = 1);
1157
1158inline void f(int x, int y, int z) { /* ... */ }
1159</pre>
1160
1161<p>The representation of "f" differs in the source-centric and
1162 semantics-centric views of a declaration context. In the
1163 source-centric view, all redeclarations will be present, in the
1164 order they occurred in the source code, making
1165 this view suitable for clients that wish to see the structure of
1166 the source code. In the semantics-centric view, only the most recent "f"
1167 will be found by the lookup, since it effectively replaces the first
1168 declaration of "f".</p>
1169
1170<p>In the semantics-centric view, overloading of functions is
1171 represented explicitly. For example, given two declarations of a
1172 function "g" that are overloaded, e.g.,</p>
1173<pre>
1174void g();
1175void g(int);
1176</pre>
1177<p>the <code>DeclContext::lookup</code> operation will return
1178 an <code>OverloadedFunctionDecl</code> that contains both
1179 declarations of "g". Clients that perform semantic analysis on a
1180 program that is not concerned with the actual source code will
1181 primarily use this semantics-centric view.</p>
1182
1183<h4 id="LexicalAndSemanticContexts">Lexical and Semantic Contexts</h4>
Douglas Gregor4afa39d2009-01-20 01:17:11 +0000[diff] [blame]1184<p>Each declaration has two potentially different
Douglas Gregor074149e2009-01-05 19:45:36 +0000[diff] [blame]1185 declaration contexts: a <i>lexical</i> context, which corresponds to
1186 the source-centric view of the declaration context, and
1187 a <i>semantic</i> context, which corresponds to the
1188 semantics-centric view. The lexical context is accessible
Douglas Gregor4afa39d2009-01-20 01:17:11 +0000[diff] [blame]1189 via <code>Decl::getLexicalDeclContext</code> while the
Douglas Gregor074149e2009-01-05 19:45:36 +0000[diff] [blame]1190 semantic context is accessible
Douglas Gregor4afa39d2009-01-20 01:17:11 +0000[diff] [blame]1191 via <code>Decl::getDeclContext</code>, both of which return
Douglas Gregor074149e2009-01-05 19:45:36 +0000[diff] [blame]1192 <code>DeclContext</code> pointers. For most declarations, the two
1193 contexts are identical. For example:</p>
1194
1195<pre>
1196class X {
1197public:
1198 void f(int x);
1199};
1200</pre>
1201
1202<p>Here, the semantic and lexical contexts of <code>X::f</code> are
1203 the <code>DeclContext</code> associated with the
1204 class <code>X</code> (itself stored as a <code>RecordDecl</code> AST
1205 node). However, we can now define <code>X::f</code> out-of-line:</p>
1206
1207<pre>
1208void X::f(int x = 17) { /* ... */ }
1209</pre>
1210
1211<p>This definition of has different lexical and semantic
1212 contexts. The lexical context corresponds to the declaration
1213 context in which the actual declaration occurred in the source
1214 code, e.g., the translation unit containing <code>X</code>. Thus,
1215 this declaration of <code>X::f</code> can be found by traversing
1216 the declarations provided by
1217 [<code>decls_begin()</code>, <code>decls_end()</code>) in the
1218 translation unit.</p>
1219
1220<p>The semantic context of <code>X::f</code> corresponds to the
1221 class <code>X</code>, since this member function is (semantically) a
1222 member of <code>X</code>. Lookup of the name <code>f</code> into
1223 the <code>DeclContext</code> associated with <code>X</code> will
1224 then return the definition of <code>X::f</code> (including
1225 information about the default argument).</p>
1226
1227<h4 id="TransparentContexts">Transparent Declaration Contexts</h4>
1228<p>In C and C++, there are several contexts in which names that are
1229 logically declared inside another declaration will actually "leak"
1230 out into the enclosing scope from the perspective of name
1231 lookup. The most obvious instance of this behavior is in
1232 enumeration types, e.g.,</p>
1233<pre>
1234enum Color {
1235 Red,
1236 Green,
1237 Blue
1238};
1239</pre>
1240
1241<p>Here, <code>Color</code> is an enumeration, which is a declaration
1242 context that contains the
1243 enumerators <code>Red</code>, <code>Green</code>,
1244 and <code>Blue</code>. Thus, traversing the list of declarations
1245 contained in the enumeration <code>Color</code> will
1246 yield <code>Red</code>, <code>Green</code>,
1247 and <code>Blue</code>. However, outside of the scope
1248 of <code>Color</code> one can name the enumerator <code>Red</code>
1249 without qualifying the name, e.g.,</p>
1250
1251<pre>
1252Color c = Red;
1253</pre>
1254
1255<p>There are other entities in C++ that provide similar behavior. For
1256 example, linkage specifications that use curly braces:</p>
1257
1258<pre>
1259extern "C" {
1260 void f(int);
1261 void g(int);
1262}
1263// f and g are visible here
1264</pre>
1265
1266<p>For source-level accuracy, we treat the linkage specification and
1267 enumeration type as a
1268 declaration context in which its enclosed declarations ("Red",
1269 "Green", and "Blue"; "f" and "g")
1270 are declared. However, these declarations are visible outside of the
1271 scope of the declaration context.</p>
1272
1273<p>These language features (and several others, described below) have
1274 roughly the same set of
1275 requirements: declarations are declared within a particular lexical
1276 context, but the declarations are also found via name lookup in
1277 scopes enclosing the declaration itself. This feature is implemented
1278 via <i>transparent</i> declaration contexts
1279 (see <code>DeclContext::isTransparentContext()</code>), whose
1280 declarations are visible in the nearest enclosing non-transparent
1281 declaration context. This means that the lexical context of the
1282 declaration (e.g., an enumerator) will be the
1283 transparent <code>DeclContext</code> itself, as will the semantic
1284 context, but the declaration will be visible in every outer context
1285 up to and including the first non-transparent declaration context (since
1286 transparent declaration contexts can be nested).</p>
1287
1288<p>The transparent <code>DeclContexts</code> are:</p>
1289<ul>
1290 <li>Enumerations (but not C++0x "scoped enumerations"):
1291 <pre>
1292enum Color {
1293 Red,
1294 Green,
1295 Blue
1296};
1297// Red, Green, and Blue are in scope
1298 </pre></li>
1299 <li>C++ linkage specifications:
1300 <pre>
1301extern "C" {
1302 void f(int);
1303 void g(int);
1304}
1305// f and g are in scope
1306 </pre></li>
1307 <li>Anonymous unions and structs:
1308 <pre>
1309struct LookupTable {
1310 bool IsVector;
1311 union {
1312 std::vector&lt;Item&gt; *Vector;
1313 std::set&lt;Item&gt; *Set;
1314 };
1315};
1316
1317LookupTable LT;
1318LT.Vector = 0; // Okay: finds Vector inside the unnamed union
1319 </pre>
1320 </li>
1321 <li>C++0x inline namespaces:
1322<pre>
1323namespace mylib {
1324 inline namespace debug {
1325 class X;
1326 }
1327}
1328mylib::X *xp; // okay: mylib::X refers to mylib::debug::X
1329</pre>
1330</li>
1331</ul>
1332
1333
1334<h4 id="MultiDeclContext">Multiply-Defined Declaration Contexts</h4>
1335<p>C++ namespaces have the interesting--and, so far, unique--property that
1336the namespace can be defined multiple times, and the declarations
1337provided by each namespace definition are effectively merged (from
1338the semantic point of view). For example, the following two code
1339snippets are semantically indistinguishable:</p>
1340<pre>
1341// Snippet #1:
1342namespace N {
1343 void f();
1344}
1345namespace N {
1346 void f(int);
1347}
1348
1349// Snippet #2:
1350namespace N {
1351 void f();
1352 void f(int);
1353}
1354</pre>
1355
1356<p>In Clang's representation, the source-centric view of declaration
1357 contexts will actually have two separate <code>NamespaceDecl</code>
1358 nodes in Snippet #1, each of which is a declaration context that
1359 contains a single declaration of "f". However, the semantics-centric
1360 view provided by name lookup into the namespace <code>N</code> for
1361 "f" will return an <code>OverloadedFunctionDecl</code> that contains
1362 both declarations of "f".</p>
1363
1364<p><code>DeclContext</code> manages multiply-defined declaration
1365 contexts internally. The
1366 function <code>DeclContext::getPrimaryContext</code> retrieves the
1367 "primary" context for a given <code>DeclContext</code> instance,
1368 which is the <code>DeclContext</code> responsible for maintaining
1369 the lookup table used for the semantics-centric view. Given the
1370 primary context, one can follow the chain
1371 of <code>DeclContext</code> nodes that define additional
1372 declarations via <code>DeclContext::getNextContext</code>. Note that
1373 these functions are used internally within the lookup and insertion
1374 methods of the <code>DeclContext</code>, so the vast majority of
1375 clients can ignore them.</p>
1376
1377<!-- ======================================================================= -->
Ted Kremenek8bc05712007-10-10 23:01:43 +0000[diff] [blame]1378<h3 id="CFG">The <tt>CFG</tt> class</h3>
1379<!-- ======================================================================= -->
1380
1381<p>The <tt>CFG</tt> class is designed to represent a source-level
1382control-flow graph for a single statement (<tt>Stmt*</tt>). Typically
1383instances of <tt>CFG</tt> are constructed for function bodies (usually
1384an instance of <tt>CompoundStmt</tt>), but can also be instantiated to
1385represent the control-flow of any class that subclasses <tt>Stmt</tt>,
1386which includes simple expressions. Control-flow graphs are especially
1387useful for performing
1388<a href="http://en.wikipedia.org/wiki/Data_flow_analysis#Sensitivities">flow-
1389or path-sensitive</a> program analyses on a given function.</p>
1390
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]1391<!-- ============ -->
Ted Kremenek8bc05712007-10-10 23:01:43 +0000[diff] [blame]1392<h4>Basic Blocks</h4>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]1393<!-- ============ -->
Ted Kremenek8bc05712007-10-10 23:01:43 +0000[diff] [blame]1394
1395<p>Concretely, an instance of <tt>CFG</tt> is a collection of basic
1396blocks. Each basic block is an instance of <tt>CFGBlock</tt>, which
1397simply contains an ordered sequence of <tt>Stmt*</tt> (each referring
1398to statements in the AST). The ordering of statements within a block
1399indicates unconditional flow of control from one statement to the
1400next. <a href="#ConditionalControlFlow">Conditional control-flow</a>
1401is represented using edges between basic blocks. The statements
1402within a given <tt>CFGBlock</tt> can be traversed using
1403the <tt>CFGBlock::*iterator</tt> interface.</p>
1404
1405<p>
Ted Kremenek18e17e72007-10-18 22:50:52 +0000[diff] [blame]1406A <tt>CFG</tt> object owns the instances of <tt>CFGBlock</tt> within
Ted Kremenek8bc05712007-10-10 23:01:43 +0000[diff] [blame]1407the control-flow graph it represents. Each <tt>CFGBlock</tt> within a
1408CFG is also uniquely numbered (accessible
1409via <tt>CFGBlock::getBlockID()</tt>). Currently the number is
1410based on the ordering the blocks were created, but no assumptions
1411should be made on how <tt>CFGBlock</tt>s are numbered other than their
1412numbers are unique and that they are numbered from 0..N-1 (where N is
1413the number of basic blocks in the CFG).</p>
1414
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]1415<!-- ===================== -->
Ted Kremenek8bc05712007-10-10 23:01:43 +0000[diff] [blame]1416<h4>Entry and Exit Blocks</h4>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]1417<!-- ===================== -->
Ted Kremenek8bc05712007-10-10 23:01:43 +0000[diff] [blame]1418
1419Each instance of <tt>CFG</tt> contains two special blocks:
1420an <i>entry</i> block (accessible via <tt>CFG::getEntry()</tt>), which
1421has no incoming edges, and an <i>exit</i> block (accessible
1422via <tt>CFG::getExit()</tt>), which has no outgoing edges. Neither
1423block contains any statements, and they serve the role of providing a
1424clear entrance and exit for a body of code such as a function body.
1425The presence of these empty blocks greatly simplifies the
1426implementation of many analyses built on top of CFGs.
1427
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]1428<!-- ===================================================== -->
Ted Kremenek8bc05712007-10-10 23:01:43 +0000[diff] [blame]1429<h4 id ="ConditionalControlFlow">Conditional Control-Flow</h4>
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]1430<!-- ===================================================== -->
Ted Kremenek8bc05712007-10-10 23:01:43 +0000[diff] [blame]1431
1432<p>Conditional control-flow (such as those induced by if-statements
1433and loops) is represented as edges between <tt>CFGBlock</tt>s.
1434Because different C language constructs can induce control-flow,
1435each <tt>CFGBlock</tt> also records an extra <tt>Stmt*</tt> that
1436represents the <i>terminator</i> of the block. A terminator is simply
1437the statement that caused the control-flow, and is used to identify
1438the nature of the conditional control-flow between blocks. For
1439example, in the case of an if-statement, the terminator refers to
1440the <tt>IfStmt</tt> object in the AST that represented the given
1441branch.</p>
1442
1443<p>To illustrate, consider the following code example:</p>
1444
1445<code>
1446int foo(int x) {<br>
1447&nbsp;&nbsp;x = x + 1;<br>
1448<br>
1449&nbsp;&nbsp;if (x > 2) x++;<br>
1450&nbsp;&nbsp;else {<br>
1451&nbsp;&nbsp;&nbsp;&nbsp;x += 2;<br>
1452&nbsp;&nbsp;&nbsp;&nbsp;x *= 2;<br>
1453&nbsp;&nbsp;}<br>
1454<br>
1455&nbsp;&nbsp;return x;<br>
1456}
1457</code>
1458
1459<p>After invoking the parser+semantic analyzer on this code fragment,
1460the AST of the body of <tt>foo</tt> is referenced by a
1461single <tt>Stmt*</tt>. We can then construct an instance
1462of <tt>CFG</tt> representing the control-flow graph of this function
1463body by single call to a static class method:</p>
1464
1465<code>
1466&nbsp;&nbsp;Stmt* FooBody = ...<br>
1467&nbsp;&nbsp;CFG* FooCFG = <b>CFG::buildCFG</b>(FooBody);
1468</code>
1469
1470<p>It is the responsibility of the caller of <tt>CFG::buildCFG</tt>
1471to <tt>delete</tt> the returned <tt>CFG*</tt> when the CFG is no
1472longer needed.</p>
1473
1474<p>Along with providing an interface to iterate over
1475its <tt>CFGBlock</tt>s, the <tt>CFG</tt> class also provides methods
1476that are useful for debugging and visualizing CFGs. For example, the
1477method
1478<tt>CFG::dump()</tt> dumps a pretty-printed version of the CFG to
1479standard error. This is especially useful when one is using a
1480debugger such as gdb. For example, here is the output
1481of <tt>FooCFG->dump()</tt>:</p>
1482
1483<code>
1484&nbsp;[ B5 (ENTRY) ]<br>
1485&nbsp;&nbsp;&nbsp;&nbsp;Predecessors (0):<br>
1486&nbsp;&nbsp;&nbsp;&nbsp;Successors (1): B4<br>
1487<br>
1488&nbsp;[ B4 ]<br>
1489&nbsp;&nbsp;&nbsp;&nbsp;1: x = x + 1<br>
1490&nbsp;&nbsp;&nbsp;&nbsp;2: (x > 2)<br>
1491&nbsp;&nbsp;&nbsp;&nbsp;<b>T: if [B4.2]</b><br>
1492&nbsp;&nbsp;&nbsp;&nbsp;Predecessors (1): B5<br>
1493&nbsp;&nbsp;&nbsp;&nbsp;Successors (2): B3 B2<br>
1494<br>
1495&nbsp;[ B3 ]<br>
1496&nbsp;&nbsp;&nbsp;&nbsp;1: x++<br>
1497&nbsp;&nbsp;&nbsp;&nbsp;Predecessors (1): B4<br>
1498&nbsp;&nbsp;&nbsp;&nbsp;Successors (1): B1<br>
1499<br>
1500&nbsp;[ B2 ]<br>
1501&nbsp;&nbsp;&nbsp;&nbsp;1: x += 2<br>
1502&nbsp;&nbsp;&nbsp;&nbsp;2: x *= 2<br>
1503&nbsp;&nbsp;&nbsp;&nbsp;Predecessors (1): B4<br>
1504&nbsp;&nbsp;&nbsp;&nbsp;Successors (1): B1<br>
1505<br>
1506&nbsp;[ B1 ]<br>
1507&nbsp;&nbsp;&nbsp;&nbsp;1: return x;<br>
1508&nbsp;&nbsp;&nbsp;&nbsp;Predecessors (2): B2 B3<br>
1509&nbsp;&nbsp;&nbsp;&nbsp;Successors (1): B0<br>
1510<br>
1511&nbsp;[ B0 (EXIT) ]<br>
1512&nbsp;&nbsp;&nbsp;&nbsp;Predecessors (1): B1<br>
1513&nbsp;&nbsp;&nbsp;&nbsp;Successors (0):
1514</code>
1515
1516<p>For each block, the pretty-printed output displays for each block
1517the number of <i>predecessor</i> blocks (blocks that have outgoing
1518control-flow to the given block) and <i>successor</i> blocks (blocks
1519that have control-flow that have incoming control-flow from the given
1520block). We can also clearly see the special entry and exit blocks at
1521the beginning and end of the pretty-printed output. For the entry
1522block (block B5), the number of predecessor blocks is 0, while for the
1523exit block (block B0) the number of successor blocks is 0.</p>
1524
1525<p>The most interesting block here is B4, whose outgoing control-flow
1526represents the branching caused by the sole if-statement
1527in <tt>foo</tt>. Of particular interest is the second statement in
1528the block, <b><tt>(x > 2)</tt></b>, and the terminator, printed
1529as <b><tt>if [B4.2]</tt></b>. The second statement represents the
1530evaluation of the condition of the if-statement, which occurs before
1531the actual branching of control-flow. Within the <tt>CFGBlock</tt>
1532for B4, the <tt>Stmt*</tt> for the second statement refers to the
1533actual expression in the AST for <b><tt>(x > 2)</tt></b>. Thus
1534pointers to subclasses of <tt>Expr</tt> can appear in the list of
1535statements in a block, and not just subclasses of <tt>Stmt</tt> that
1536refer to proper C statements.</p>
1537
1538<p>The terminator of block B4 is a pointer to the <tt>IfStmt</tt>
1539object in the AST. The pretty-printer outputs <b><tt>if
1540[B4.2]</tt></b> because the condition expression of the if-statement
1541has an actual place in the basic block, and thus the terminator is
1542essentially
1543<i>referring</i> to the expression that is the second statement of
1544block B4 (i.e., B4.2). In this manner, conditions for control-flow
1545(which also includes conditions for loops and switch statements) are
1546hoisted into the actual basic block.</p>
1547
Chris Lattner62fd2782008-11-22 21:41:31 +0000[diff] [blame]1548<!-- ===================== -->
1549<!-- <h4>Implicit Control-Flow</h4> -->
1550<!-- ===================== -->
Ted Kremenek8bc05712007-10-10 23:01:43 +0000[diff] [blame]1551
1552<!--
1553<p>A key design principle of the <tt>CFG</tt> class was to not require
1554any transformations to the AST in order to represent control-flow.
1555Thus the <tt>CFG</tt> does not perform any "lowering" of the
1556statements in an AST: loops are not transformed into guarded gotos,
1557short-circuit operations are not converted to a set of if-statements,
1558and so on.</p>
1559-->
Ted Kremenek17a295d2008-06-11 06:19:49 +0000[diff] [blame]1560
Chris Lattner7bad1992008-11-16 21:48:07 +0000[diff] [blame]1561
1562<!-- ======================================================================= -->
1563<h3 id="Constants">Constant Folding in the Clang AST</h3>
1564<!-- ======================================================================= -->
1565
1566<p>There are several places where constants and constant folding matter a lot to
1567the Clang front-end. First, in general, we prefer the AST to retain the source
1568code as close to how the user wrote it as possible. This means that if they
1569wrote "5+4", we want to keep the addition and two constants in the AST, we don't
1570want to fold to "9". This means that constant folding in various ways turns
1571into a tree walk that needs to handle the various cases.</p>
1572
1573<p>However, there are places in both C and C++ that require constants to be
1574folded. For example, the C standard defines what an "integer constant
1575expression" (i-c-e) is with very precise and specific requirements. The
1576language then requires i-c-e's in a lot of places (for example, the size of a
1577bitfield, the value for a case statement, etc). For these, we have to be able
1578to constant fold the constants, to do semantic checks (e.g. verify bitfield size
1579is non-negative and that case statements aren't duplicated). We aim for Clang
1580to be very pedantic about this, diagnosing cases when the code does not use an
1581i-c-e where one is required, but accepting the code unless running with
1582<tt>-pedantic-errors</tt>.</p>
1583
1584<p>Things get a little bit more tricky when it comes to compatibility with
1585real-world source code. Specifically, GCC has historically accepted a huge
1586superset of expressions as i-c-e's, and a lot of real world code depends on this
1587unfortuate accident of history (including, e.g., the glibc system headers). GCC
1588accepts anything its "fold" optimizer is capable of reducing to an integer
1589constant, which means that the definition of what it accepts changes as its
1590optimizer does. One example is that GCC accepts things like "case X-X:" even
1591when X is a variable, because it can fold this to 0.</p>
1592
1593<p>Another issue are how constants interact with the extensions we support, such
1594as __builtin_constant_p, __builtin_inf, __extension__ and many others. C99
1595obviously does not specify the semantics of any of these extensions, and the
1596definition of i-c-e does not include them. However, these extensions are often
1597used in real code, and we have to have a way to reason about them.</p>
1598
1599<p>Finally, this is not just a problem for semantic analysis. The code
1600generator and other clients have to be able to fold constants (e.g. to
1601initialize global variables) and has to handle a superset of what C99 allows.
1602Further, these clients can benefit from extended information. For example, we
1603know that "foo()||1" always evaluates to true, but we can't replace the
1604expression with true because it has side effects.</p>
1605
1606<!-- ======================= -->
1607<h4>Implementation Approach</h4>
1608<!-- ======================= -->
1609
1610<p>After trying several different approaches, we've finally converged on a
1611design (Note, at the time of this writing, not all of this has been implemented,
1612consider this a design goal!). Our basic approach is to define a single
1613recursive method evaluation method (<tt>Expr::Evaluate</tt>), which is
1614implemented in <tt>AST/ExprConstant.cpp</tt>. Given an expression with 'scalar'
1615type (integer, fp, complex, or pointer) this method returns the following
1616information:</p>
1617
1618<ul>
1619<li>Whether the expression is an integer constant expression, a general
1620 constant that was folded but has no side effects, a general constant that
1621 was folded but that does have side effects, or an uncomputable/unfoldable
1622 value.
1623</li>
1624<li>If the expression was computable in any way, this method returns the APValue
1625 for the result of the expression.</li>
1626<li>If the expression is not evaluatable at all, this method returns
1627 information on one of the problems with the expression. This includes a
1628 SourceLocation for where the problem is, and a diagnostic ID that explains
1629 the problem. The diagnostic should be have ERROR type.</li>
1630<li>If the expression is not an integer constant expression, this method returns
1631 information on one of the problems with the expression. This includes a
1632 SourceLocation for where the problem is, and a diagnostic ID that explains
1633 the problem. The diagnostic should be have EXTENSION type.</li>
1634</ul>
1635
1636<p>This information gives various clients the flexibility that they want, and we
1637will eventually have some helper methods for various extensions. For example,
1638Sema should have a <tt>Sema::VerifyIntegerConstantExpression</tt> method, which
1639calls Evaluate on the expression. If the expression is not foldable, the error
1640is emitted, and it would return true. If the expression is not an i-c-e, the
1641EXTENSION diagnostic is emitted. Finally it would return false to indicate that
1642the AST is ok.</p>
1643
1644<p>Other clients can use the information in other ways, for example, codegen can
1645just use expressions that are foldable in any way.</p>
1646
1647<!-- ========== -->
1648<h4>Extensions</h4>
1649<!-- ========== -->
1650
Chris Lattner552de0a2008-11-23 08:16:56 +0000[diff] [blame]1651<p>This section describes how some of the various extensions Clang supports
Chris Lattner7bad1992008-11-16 21:48:07 +0000[diff] [blame]1652interacts with constant evaluation:</p>
1653
1654<ul>
1655<li><b><tt>__extension__</tt></b>: The expression form of this extension causes
1656 any evaluatable subexpression to be accepted as an integer constant
1657 expression.</li>
1658<li><b><tt>__builtin_constant_p</tt></b>: This returns true (as a integer
Chris Lattner28daa532008-12-12 06:55:44 +0000[diff] [blame]1659 constant expression) if the operand is any evaluatable constant. As a
1660 special case, if <tt>__builtin_constant_p</tt> is the (potentially
1661 parenthesized) condition of a conditional operator expression ("?:"), only
Chris Lattner42b83dd2008-12-12 18:00:51 +0000[diff] [blame]1662 the true side of the conditional operator is considered, and it is evaluated
1663 with full constant folding.</li>
Chris Lattner7bad1992008-11-16 21:48:07 +0000[diff] [blame]1664<li><b><tt>__builtin_choose_expr</tt></b>: The condition is required to be an
1665 integer constant expression, but we accept any constant as an "extension of
1666 an extension". This only evaluates one operand depending on which way the
1667 condition evaluates.</li>
1668<li><b><tt>__builtin_classify_type</tt></b>: This always returns an integer
1669 constant expression.</li>
1670<li><b><tt>__builtin_inf,nan,..</tt></b>: These are treated just like a
1671 floating-point literal.</li>
1672<li><b><tt>__builtin_abs,copysign,..</tt></b>: These are constant folded as
1673 general constant expressions.</li>
1674</ul>
1675
1676
1677
1678
Ted Kremenek17a295d2008-06-11 06:19:49 +0000[diff] [blame]1679</div>
1680</body>
Douglas Gregor2e1cd422008-11-17 14:58:09 +0000[diff] [blame]1681</html>