Blame - docs/InternalsManual.html - platform/external/clang

blob: 0d029f7e375e7828d5d79b363aed5135c50d0f03 [file] [log] [blame]

Ted Kremenek	17a295d	2008-06-11 06:19:49 +0000	[diff] [blame]	1	<html>
				2	<head>
Chris Lattner	552de0a	2008-11-23 08:16:56 +0000	[diff] [blame]	3	<title>"Clang" CFE Internals Manual</title>
Ted Kremenek	17a295d	2008-06-11 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 Redl	6816856	2008-11-22 22:16:45 +0000	[diff] [blame]	6	<style type="text/css">
				7	td {
				8	vertical-align: top;
				9	}
				10	</style>
Ted Kremenek	17a295d	2008-06-11 06:19:49 +0000	[diff] [blame]	11	</head>
				12	<body>
				13
				14	<!--#include virtual="../menu.html.incl"-->
				15
				16	<div id="content">
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	17
Chris Lattner	552de0a	2008-11-23 08:16:56 +0000	[diff] [blame]	18	<h1>"Clang" CFE Internals Manual</h1>
Chris Lattner	86920d3	2007-07-31 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 Lattner	552de0a	2008-11-23 08:16:56 +0000	[diff] [blame]	23	<li><a href="#libbasic">The Clang 'Basic' Library</a>
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	24	<ul>
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	25	<li><a href="#Diagnostics">The Diagnostics Subsystem</a></li>
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	26	<li><a href="#SourceLocation">The SourceLocation and SourceManager
				27	classes</a></li>
				28	</ul>
				29	</li>
				30	<li><a href="#liblex">The Lexer and Preprocessor Library</a>
				31	<ul>
				32	<li><a href="#Token">The Token class</a></li>
				33	<li><a href="#Lexer">The Lexer class</a></li>
Chris Lattner	3932fe0	2009-01-06 06:02:08 +0000	[diff] [blame]	34	<li><a href="#AnnotationToken">Annotation Tokens</a></li>
Chris Lattner	7928125	2008-03-09 02:27:26 +0000	[diff] [blame]	35	<li><a href="#TokenLexer">The TokenLexer class</a></li>
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	36	<li><a href="#MultipleIncludeOpt">The MultipleIncludeOpt class</a></li>
				37	</ul>
				38	</li>
				39	<li><a href="#libparse">The Parser Library</a>
				40	<ul>
				41	</ul>
				42	</li>
				43	<li><a href="#libast">The AST Library</a>
				44	<ul>
				45	<li><a href="#Type">The Type class and its subclasses</a></li>
				46	<li><a href="#QualType">The QualType class</a></li>
Douglas Gregor	2e1cd42	2008-11-17 14:58:09 +0000	[diff] [blame]	47	<li><a href="#DeclarationName">Declaration names</a></li>
Douglas Gregor	074149e	2009-01-05 19:45:36 +0000	[diff] [blame]	48	<li><a href="#DeclContext">Declaration contexts</a>
				49	<ul>
				50	<li><a href="#Redeclarations">Redeclarations and Overloads</a></li>
				51	<li><a href="#LexicalAndSemanticContexts">Lexical and Semantic
				52	Contexts</a></li>
				53	<li><a href="#TransparentContexts">Transparent Declaration Contexts</a></li>
				54	<li><a href="#MultiDeclContext">Multiply-Defined Declaration Contexts</a></li>
				55	</ul>
				56	</li>
Ted Kremenek	8bc0571	2007-10-10 23:01:43 +0000	[diff] [blame]	57	<li><a href="#CFG">The CFG class</a></li>
Chris Lattner	7bad199	2008-11-16 21:48:07 +0000	[diff] [blame]	58	<li><a href="#Constants">Constant Folding in the Clang AST</a></li>
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	59	</ul>
				60	</li>
				61	</ul>
				62
				63
				64	<!-- ======================================================================= -->
				65	<h2 id="intro">Introduction</h2>
				66	<!-- ======================================================================= -->
				67
				68	<p>This document describes some of the more important APIs and internal design
Chris Lattner	552de0a	2008-11-23 08:16:56 +0000	[diff] [blame]	69	decisions made in the Clang C front-end. The purpose of this document is to
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	70	both capture some of this high level information and also describe some of the
				71	design decisions behind it. This is meant for people interested in hacking on
Chris Lattner	552de0a	2008-11-23 08:16:56 +0000	[diff] [blame]	72	Clang, not for end-users. The description below is categorized by
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	73	libraries, and does not describe any of the clients of the libraries.</p>
				74
				75	<!-- ======================================================================= -->
				76	<h2 id="libsystem">LLVM System and Support Libraries</h2>
				77	<!-- ======================================================================= -->
				78
Chris Lattner	552de0a	2008-11-23 08:16:56 +0000	[diff] [blame]	79	<p>The LLVM libsystem library provides the basic Clang system abstraction layer,
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	80	which is used for file system access. The LLVM libsupport library provides many
				81	underlying libraries and <a
				82	href="http://llvm.org/docs/ProgrammersManual.html">data-structures</a>,
				83	including command line option
				84	processing and various containers.</p>
				85
				86	<!-- ======================================================================= -->
Chris Lattner	552de0a	2008-11-23 08:16:56 +0000	[diff] [blame]	87	<h2 id="libbasic">The Clang 'Basic' Library</h2>
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	88	<!-- ======================================================================= -->
				89
				90	<p>This library certainly needs a better name. The 'basic' library contains a
				91	number of low-level utilities for tracking and manipulating source buffers,
				92	locations within the source buffers, diagnostics, tokens, target abstraction,
				93	and information about the subset of the language being compiled for.</p>
				94
				95	<p>Part of this infrastructure is specific to C (such as the TargetInfo class),
				96	other parts could be reused for other non-C-based languages (SourceLocation,
				97	SourceManager, Diagnostics, FileManager). When and if there is future demand
				98	we can figure out if it makes sense to introduce a new library, move the general
				99	classes somewhere else, or introduce some other solution.</p>
				100
				101	<p>We describe the roles of these classes in order of their dependencies.</p>
				102
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	103
				104	<!-- ======================================================================= -->
				105	<h3 id="Diagnostics">The Diagnostics Subsystem</h3>
				106	<!-- ======================================================================= -->
				107
				108	<p>The Clang Diagnostics subsystem is an important part of how the compiler
				109	communicates with the human. Diagnostics are the warnings and errors produced
				110	when the code is incorrect or dubious. In Clang, each diagnostic produced has
				111	(at the minimum) a unique ID, a <a href="#SourceLocation">SourceLocation</a> to
				112	"put the caret", an English translation associated with it, and a severity (e.g.
				113	<tt>WARNING</tt> or <tt>ERROR</tt>). They can also optionally include a number
				114	of arguments to the dianostic (which fill in "%0"'s in the string) as well as a
				115	number of source ranges that related to the diagnostic.</p>
				116
Chris Lattner	552de0a	2008-11-23 08:16:56 +0000	[diff] [blame]	117	<p>In this section, we'll be giving examples produced by the Clang command line
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	118	driver, but diagnostics can be <a href="#DiagnosticClient">rendered in many
				119	different ways</a> depending on how the DiagnosticClient interface is
				120	implemented. A representative example of a diagonstic is:</p>
				121
				122	<pre>
				123	t.c:38:15: error: invalid operands to binary expression ('int *' and '_Complex float')
				124	<font color="darkgreen">P = (P-42) + Gamma*4;</font>
				125	<font color="blue">~~~~~~ ^ ~~~~~~~</font>
				126	</pre>
				127
				128	<p>In this example, you can see the English translation, the severity (error),
				129	you can see the source location (the caret ("^") and file/line/column info),
				130	the source ranges "~~~~", arguments to the diagnostic ("int*" and "_Complex
				131	float"). You'll have to believe me that there is a unique ID backing the
				132	diagnostic :).</p>
				133
				134	<p>Getting all of this to happen has several steps and involves many moving
				135	pieces, this section describes them and talks about best practices when adding
				136	a new diagnostic.</p>
				137
				138	<!-- ============================ -->
				139	<h4>The DiagnosticKinds.def file</h4>
				140	<!-- ============================ -->
				141
				142	<p>Diagnostics are created by adding an entry to the <tt><a
				143	href="http://llvm.org/svn/llvm-project/cfe/trunk/include/clang/Basic/DiagnosticKinds.def"
				144	>DiagnosticKinds.def</a></tt> file. This file encodes the unique ID of the
				145	diagnostic (as an enum, the first argument), the severity of the diagnostic
				146	(second argument) and the English translation + format string.</p>
				147
				148	<p>There is little sanity with the naming of the unique ID's right now. Some
				149	start with err_, warn_, ext_ to encode the severity into the name. Since the
				150	enum is referenced in the C++ code that produces the diagnostic, it is somewhat
				151	useful for it to be reasonably short.</p>
				152
				153	<p>The severity of the diagnostic comes from the set {<tt>NOTE</tt>,
				154	<tt>WARNING</tt>, <tt>EXTENSION</tt>, <tt>EXTWARN</tt>, <tt>ERROR</tt>}. The
				155	<tt>ERROR</tt> severity is used for diagnostics indicating the program is never
				156	acceptable under any circumstances. When an error is emitted, the AST for the
				157	input code may not be fully built. The <tt>EXTENSION</tt> and <tt>EXTWARN</tt>
				158	severities are used for extensions to the language that Clang accepts. This
				159	means that Clang fully understands and can represent them in the AST, but we
				160	produce diagnostics to tell the user their code is non-portable. The difference
				161	is that the former are ignored by default, and the later warn by default. The
				162	<tt>WARNING</tt> severity is used for constructs that are valid in the currently
				163	selected source language but that are dubious in some way. The <tt>NOTE</tt>
				164	level is used to staple more information onto a previous diagnostics.</p>
				165
				166	<p>These <em>severities</em> are mapped into a smaller set (the
				167	Diagnostic::Level enum, {<tt>Ignored</tt>, <tt>Note</tt>, <tt>Warning</tt>,
				168	<tt>Error</tt> }) of output <em>levels</em> by the diagnostics subsystem based
				169	on various configuration options. For example, if the user specifies
				170	<tt>-pedantic</tt>, <tt>EXTENSION</tt> maps to <tt>Warning</tt>, if they specify
				171	<tt>-pedantic-errors</tt>, it turns into <tt>Error</tt>. Clang also internally
				172	supports a fully fine grained mapping mechanism that allows you to map any
				173	diagnostic that doesn't have <tt>ERRROR</tt> severity to any output level that
				174	you want. This is used to implement options like <tt>-Wunused_macros</tt>,
				175	<tt>-Wundef</tt> etc.</p>
				176
				177	<!-- ================= -->
				178	<h4>The Format String</h4>
				179	<!-- ================= -->
				180
				181	<p>The format string for the diagnostic is very simple, but it has some power.
				182	It takes the form of a string in English with markers that indicate where and
				183	how arguments to the diagnostic are inserted and formatted. For example, here
				184	are some simple format strings:</p>
				185
				186	<pre>
				187	"binary integer literals are an extension"
				188	"format string contains '\\0' within the string body"
				189	"more '<b>%%</b>' conversions than data arguments"
Chris Lattner	545b368	2008-11-23 20:27:13 +0000	[diff] [blame]	190	"invalid operands to binary expression (<b>%0</b> and <b>%1</b>)"
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	191	"overloaded '<b>%0</b>' must be a <b>%select{unary\|binary\|unary or binary}2</b> operator"
				192	" (has <b>%1</b> parameter<b>%s1</b>)"
				193	</pre>
				194
				195	<p>These examples show some important points of format strings. You can use any
				196	plain ASCII character in the diagnostic string except "%" without a problem,
				197	but these are C strings, so you have to use and be aware of all the C escape
				198	sequences (as in the second example). If you want to produce a "%" in the
				199	output, use the "%%" escape sequence, like the third diagnostic. Finally,
Chris Lattner	552de0a	2008-11-23 08:16:56 +0000	[diff] [blame]	200	Clang uses the "%...[digit]" sequences to specify where and how arguments to
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	201	the diagnostic are formatted.</p>
				202
				203	<p>Arguments to the diagnostic are numbered according to how they are specified
				204	by the C++ code that <a href="#producingdiag">produces them</a>, and are
				205	referenced by <tt>%0</tt> .. <tt>%9</tt>. If you have more than 10 arguments
Chris Lattner	552de0a	2008-11-23 08:16:56 +0000	[diff] [blame]	206	to your diagnostic, you are doing something wrong :). Unlike printf, there
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	207	is no requirement that arguments to the diagnostic end up in the output in
				208	the same order as they are specified, you could have a format string with
				209	<tt>"%1 %0"</tt> that swaps them, for example. The text in between the
				210	percent and digit are formatting instructions. If there are no instructions,
				211	the argument is just turned into a string and substituted in.</p>
				212
				213	<p>Here are some "best practices" for writing the English format string:</p>
				214
				215	<ul>
				216	<li>Keep the string short. It should ideally fit in the 80 column limit of the
				217	<tt>DiagnosticKinds.def</tt> file. This avoids the diagnostic wrapping when
				218	printed, and forces you to think about the important point you are conveying
				219	with the diagnostic.</li>
				220	<li>Take advantage of location information. The user will be able to see the
				221	line and location of the caret, so you don't need to tell them that the
				222	problem is with the 4th argument to the function: just point to it.</li>
				223	<li>Do not capitalize the diagnostic string, and do not end it with a
				224	period.</li>
				225	<li>If you need to quote something in the diagnostic string, use single
				226	quotes.</li>
				227	</ul>
				228
				229	<p>Diagnostics should never take random English strings as arguments: you
				230	shouldn't use <tt>"you have a problem with %0"</tt> and pass in things like
				231	<tt>"your argument"</tt> or <tt>"your return value"</tt> as arguments. Doing
				232	this prevents <a href="translation">translating</a> the Clang diagnostics to
				233	other languages (because they'll get random English words in their otherwise
				234	localized diagnostic). The exceptions to this are C/C++ language keywords
				235	(e.g. auto, const, mutable, etc) and C/C++ operators (<tt>/=</tt>). Note
				236	that things like "pointer" and "reference" are not keywords. On the other
				237	hand, you <em>can</em> include anything that comes from the user's source code,
Chris Lattner	552de0a	2008-11-23 08:16:56 +0000	[diff] [blame]	238	including variable names, types, labels, etc. The 'select' format can be
				239	used to achieve this sort of thing in a localizable way, see below.</p>
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	240
				241	<!-- ==================================== -->
				242	<h4>Formatting a Diagnostic Argument</a></h4>
				243	<!-- ==================================== -->
				244
				245	<p>Arguments to diagnostics are fully typed internally, and come from a couple
				246	different classes: integers, types, names, and random strings. Depending on
				247	the class of the argument, it can be optionally formatted in different ways.
				248	This gives the DiagnosticClient information about what the argument means
				249	without requiring it to use a specific presentation (consider this MVC for
				250	Clang :).</p>
				251
				252	<p>Here are the different diagnostic argument formats currently supported by
				253	Clang:</p>
				254
				255	<table>
				256	<tr><td colspan="2"><b>"s" format</b></td></tr>
				257	<tr><td>Example:</td><td><tt>"requires %1 parameter%s1"</tt></td></tr>
Chris Lattner	552de0a	2008-11-23 08:16:56 +0000	[diff] [blame]	258	<tr><td>Class:</td><td>Integers</td></tr>
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	259	<tr><td>Description:</td><td>This is a simple formatter for integers that is
				260	useful when producing English diagnostics. When the integer is 1, it prints
				261	as nothing. When the integer is not 1, it prints as "s". This allows some
Chris Lattner	627b705	2008-11-23 00:28:33 +0000	[diff] [blame]	262	simple grammatical forms to be to be handled correctly, and eliminates the
				263	need to use gross things like <tt>"requires %1 parameter(s)"</tt>.</td></tr>
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	264
				265	<tr><td colspan="2"><b>"select" format</b></td></tr>
				266	<tr><td>Example:</td><td><tt>"must be a %select{unary\|binary\|unary or binary}2
				267	operator"</tt></td></tr>
Chris Lattner	552de0a	2008-11-23 08:16:56 +0000	[diff] [blame]	268	<tr><td>Class:</td><td>Integers</td></tr>
Chris Lattner	cc54334	2008-11-22 23:50:47 +0000	[diff] [blame]	269	<tr><td>Description:</td><td>This format specifier is used to merge multiple
				270	related diagnostics together into one common one, without requiring the
Chris Lattner	552de0a	2008-11-23 08:16:56 +0000	[diff] [blame]	271	difference to be specified as an English string argument. Instead of
Chris Lattner	cc54334	2008-11-22 23:50:47 +0000	[diff] [blame]	272	specifying the string, the diagnostic gets an integer argument and the
				273	format string selects the numbered option. In this case, the "%2" value
				274	must be an integer in the range [0..2]. If it is 0, it prints 'unary', if
				275	it is 1 it prints 'binary' if it is 2, it prints 'unary or binary'. This
				276	allows other language translations to substitute reasonable words (or entire
				277	phrases) based on the semantics of the diagnostic instead of having to do
				278	things textually.</td></tr>
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	279
				280	<tr><td colspan="2"><b>"plural" format</b></td></tr>
Sebastian Redl	6816856	2008-11-22 22:16:45 +0000	[diff] [blame]	281	<tr><td>Example:</td><td><tt>"you have %1 %plural{1:mouse\|:mice}1 connected to
				282	your computer"</tt></td></tr>
Chris Lattner	552de0a	2008-11-23 08:16:56 +0000	[diff] [blame]	283	<tr><td>Class:</td><td>Integers</td></tr>
Sebastian Redl	6816856	2008-11-22 22:16:45 +0000	[diff] [blame]	284	<tr><td>Description:</td><td><p>This is a formatter for complex plural forms.
				285	It is designed to handle even the requirements of languages with very
				286	complex plural forms, as many Baltic languages have. The argument consists
				287	of a series of expression/form pairs, separated by ':', where the first form
				288	whose expression evaluates to true is the result of the modifier.</p>
				289	<p>An expression can be empty, in which case it is always true. See the
				290	example at the top. Otherwise, it is a series of one or more numeric
				291	conditions, separated by ','. If any condition matches, the expression
				292	matches. Each numeric condition can take one of three forms.</p>
				293	<ul>
				294	<li>number: A simple decimal number matches if the argument is the same
Chris Lattner	627b705	2008-11-23 00:28:33 +0000	[diff] [blame]	295	as the number. Example: <tt>"%plural{1:mouse\|:mice}4"</tt></li>
Sebastian Redl	6816856	2008-11-22 22:16:45 +0000	[diff] [blame]	296	<li>range: A range in square brackets matches if the argument is within
Chris Lattner	552de0a	2008-11-23 08:16:56 +0000	[diff] [blame]	297	the range. Then range is inclusive on both ends. Example:
Chris Lattner	627b705	2008-11-23 00:28:33 +0000	[diff] [blame]	298	<tt>"%plural{0:none\|1:one\|[2,5]:some\|:many}2"</tt></li>
				299	<li>modulo: A modulo operator is followed by a number, and
				300	equals sign and either a number or a range. The tests are the
				301	same as for plain
Sebastian Redl	6816856	2008-11-22 22:16:45 +0000	[diff] [blame]	302	numbers and ranges, but the argument is taken modulo the number first.
Chris Lattner	627b705	2008-11-23 00:28:33 +0000	[diff] [blame]	303	Example: <tt>"%plural{%100=0:even hundred\|%100=[1,50]:lower half\|:everything
				304	else}1"</tt></li>
Sebastian Redl	6816856	2008-11-22 22:16:45 +0000	[diff] [blame]	305	</ul>
				306	<p>The parser is very unforgiving. A syntax error, even whitespace, will
				307	abort, as will a failure to match the argument against any
				308	expression.</p></td></tr>
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	309
Chris Lattner	077bf5e	2008-11-24 03:33:13 +0000	[diff] [blame]	310	<tr><td colspan="2"><b>"objcclass" format</b></td></tr>
				311	<tr><td>Example:</td><td><tt>"method %objcclass0 not found"</tt></td></tr>
				312	<tr><td>Class:</td><td>DeclarationName</td></tr>
				313	<tr><td>Description:</td><td><p>This is a simple formatter that indicates the
				314	DeclarationName corresponds to an Objective-C class method selector. As
				315	such, it prints the selector with a leading '+'.</p></td></tr>
				316
				317	<tr><td colspan="2"><b>"objcinstance" format</b></td></tr>
				318	<tr><td>Example:</td><td><tt>"method %objcinstance0 not found"</tt></td></tr>
				319	<tr><td>Class:</td><td>DeclarationName</td></tr>
				320	<tr><td>Description:</td><td><p>This is a simple formatter that indicates the
				321	DeclarationName corresponds to an Objective-C instance method selector. As
				322	such, it prints the selector with a leading '-'.</p></td></tr>
				323
Douglas Gregor	47b9a1c	2009-02-04 17:27:36 +0000	[diff] [blame^]	324	<tr><td colspan="2"><b>"q" format</b></td></tr>
				325	<tr><td>Example:</td><td><tt>"candidate found by name lookup is %q0"</tt></td></tr>
				326	<tr><td>Class:</td><td>NamedDecl*</td></tr>
				327	<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>
				328
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	329	</table>
				330
Chris Lattner	cc54334	2008-11-22 23:50:47 +0000	[diff] [blame]	331	<p>It is really easy to add format specifiers to the Clang diagnostics system,
Chris Lattner	552de0a	2008-11-23 08:16:56 +0000	[diff] [blame]	332	but they should be discussed before they are added. If you are creating a lot
				333	of repetitive diagnostics and/or have an idea for a useful formatter, please
				334	bring it up on the cfe-dev mailing list.</p>
Chris Lattner	cc54334	2008-11-22 23:50:47 +0000	[diff] [blame]	335
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	336	<!-- ===================================================== -->
				337	<h4><a name="#producingdiag">Producing the Diagnostic</a></h4>
				338	<!-- ===================================================== -->
				339
Chris Lattner	627b705	2008-11-23 00:28:33 +0000	[diff] [blame]	340	<p>Now that you've created the diagnostic in the DiagnosticKinds.def file, you
Chris Lattner	552de0a	2008-11-23 08:16:56 +0000	[diff] [blame]	341	need to write the code that detects the condition in question and emits the
				342	new diagnostic. Various components of Clang (e.g. the preprocessor, Sema,
Chris Lattner	627b705	2008-11-23 00:28:33 +0000	[diff] [blame]	343	etc) provide a helper function named "Diag". It creates a diagnostic and
				344	accepts the arguments, ranges, and other information that goes along with
				345	it.</p>
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	346
Chris Lattner	552de0a	2008-11-23 08:16:56 +0000	[diff] [blame]	347	<p>For example, the binary expression error comes from code like this:</p>
Chris Lattner	627b705	2008-11-23 00:28:33 +0000	[diff] [blame]	348
				349	<pre>
				350	if (various things that are bad)
				351	Diag(Loc, diag::err_typecheck_invalid_operands)
				352	<< lex->getType() << rex->getType()
				353	<< lex->getSourceRange() << rex->getSourceRange();
				354	</pre>
				355
				356	<p>This shows that use of the Diag method: they take a location (a <a
				357	href="#SourceLocation">SourceLocation</a> object) and a diagnostic enum value
				358	(which matches the name from DiagnosticKinds.def). If the diagnostic takes
				359	arguments, they are specified with the << operator: the first argument
				360	becomes %0, the second becomes %1, etc. The diagnostic interface allows you to
Chris Lattner	fd408ea	2008-11-23 00:42:53 +0000	[diff] [blame]	361	specify arguments of many different types, including <tt>int</tt> and
				362	<tt>unsigned</tt> for integer arguments, <tt>const char*</tt> and
				363	<tt>std::string</tt> for string arguments, <tt>DeclarationName</tt> and
				364	<tt>const IdentifierInfo*</tt> for names, <tt>QualType</tt> for types, etc.
				365	SourceRanges are also specified with the << operator, but do not have a
				366	specific ordering requirement.</p>
Chris Lattner	627b705	2008-11-23 00:28:33 +0000	[diff] [blame]	367
				368	<p>As you can see, adding and producing a diagnostic is pretty straightforward.
				369	The hard part is deciding exactly what you need to say to help the user, picking
				370	a suitable wording, and providing the information needed to format it correctly.
Chris Lattner	fd408ea	2008-11-23 00:42:53 +0000	[diff] [blame]	371	The good news is that the call site that issues a diagnostic should be
				372	completely independent of how the diagnostic is formatted and in what language
				373	it is rendered.
Chris Lattner	627b705	2008-11-23 00:28:33 +0000	[diff] [blame]	374	</p>
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	375
				376	<!-- ============================================================= -->
				377	<h4><a name="DiagnosticClient">The DiagnosticClient Interface</a></h4>
				378	<!-- ============================================================= -->
				379
Chris Lattner	627b705	2008-11-23 00:28:33 +0000	[diff] [blame]	380	<p>Once code generates a diagnostic with all of the arguments and the rest of
				381	the relevant information, Clang needs to know what to do with it. As previously
				382	mentioned, the diagnostic machinery goes through some filtering to map a
				383	severity onto a diagnostic level, then (assuming the diagnostic is not mapped to
				384	"<tt>Ignore</tt>") it invokes an object that implements the DiagnosticClient
				385	interface with the information.</p>
				386
				387	<p>It is possible to implement this interface in many different ways. For
				388	example, the normal Clang DiagnosticClient (named 'TextDiagnosticPrinter') turns
				389	the arguments into strings (according to the various formatting rules), prints
				390	out the file/line/column information and the string, then prints out the line of
				391	code, the source ranges, and the caret. However, this behavior isn't required.
				392	</p>
				393
				394	<p>Another implementation of the DiagnosticClient interface is the
Chris Lattner	552de0a	2008-11-23 08:16:56 +0000	[diff] [blame]	395	'TextDiagnosticBuffer' class, which is used when Clang is in -verify mode.
Chris Lattner	fd408ea	2008-11-23 00:42:53 +0000	[diff] [blame]	396	Instead of formatting and printing out the diagnostics, this implementation just
				397	captures and remembers the diagnostics as they fly by. Then -verify compares
Chris Lattner	552de0a	2008-11-23 08:16:56 +0000	[diff] [blame]	398	the list of produced diagnostics to the list of expected ones. If they disagree,
Chris Lattner	fd408ea	2008-11-23 00:42:53 +0000	[diff] [blame]	399	it prints out its own output.
Chris Lattner	627b705	2008-11-23 00:28:33 +0000	[diff] [blame]	400	</p>
				401
Chris Lattner	fd408ea	2008-11-23 00:42:53 +0000	[diff] [blame]	402	<p>There are many other possible implementations of this interface, and this is
				403	why we prefer diagnostics to pass down rich structured information in arguments.
				404	For example, an HTML output might want declaration names be linkified to where
				405	they come from in the source. Another example is that a GUI might let you click
				406	on typedefs to expand them. This application would want to pass significantly
				407	more information about types through to the GUI than a simple flat string. The
				408	interface allows this to happen.</p>
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	409
				410	<!-- ====================================================== -->
				411	<h4><a name="translation">Adding Translations to Clang</a></h4>
				412	<!-- ====================================================== -->
				413
Chris Lattner	627b705	2008-11-23 00:28:33 +0000	[diff] [blame]	414	<p>Not possible yet! Diagnostic strings should be written in UTF-8, the client
Chris Lattner	fd408ea	2008-11-23 00:42:53 +0000	[diff] [blame]	415	can translate to the relevant code page if needed. Each translation completely
				416	replaces the format string for the diagnostic.</p>
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	417
				418
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	419	<!-- ======================================================================= -->
				420	<h3 id="SourceLocation">The SourceLocation and SourceManager classes</h3>
				421	<!-- ======================================================================= -->
				422
				423	<p>Strangely enough, the SourceLocation class represents a location within the
				424	source code of the program. Important design points include:</p>
				425
				426	<ol>
				427	<li>sizeof(SourceLocation) must be extremely small, as these are embedded into
				428	many AST nodes and are passed around often. Currently it is 32 bits.</li>
				429	<li>SourceLocation must be a simple value object that can be efficiently
				430	copied.</li>
				431	<li>We should be able to represent a source location for any byte of any input
				432	file. This includes in the middle of tokens, in whitespace, in trigraphs,
				433	etc.</li>
				434	<li>A SourceLocation must encode the current #include stack that was active when
				435	the location was processed. For example, if the location corresponds to a
				436	token, it should contain the set of #includes active when the token was
				437	lexed. This allows us to print the #include stack for a diagnostic.</li>
				438	<li>SourceLocation must be able to describe macro expansions, capturing both
				439	the ultimate instantiation point and the source of the original character
				440	data.</li>
				441	</ol>
				442
				443	<p>In practice, the SourceLocation works together with the SourceManager class
Chris Lattner	18376dd	2009-01-16 07:00:50 +0000	[diff] [blame]	444	to encode two pieces of information about a location: it's spelling location
Chris Lattner	88054de	2009-01-16 07:15:35 +0000	[diff] [blame]	445	and it's instantiation location. For most tokens, these will be the same. However,
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	446	for a macro expansion (or tokens that came from a _Pragma directive) these will
				447	describe the location of the characters corresponding to the token and the
				448	location where the token was used (i.e. the macro instantiation point or the
				449	location of the _Pragma itself).</p>
				450
Chris Lattner	3fcbb89	2008-11-23 08:32:53 +0000	[diff] [blame]	451	<p>For efficiency, we only track one level of macro instantiations: if a token was
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	452	produced by multiple instantiations, we only track the source and ultimate
				453	destination. Though we could track the intermediate instantiation points, this
				454	would require extra bookkeeping and no known client would benefit substantially
				455	from this.</p>
				456
Chris Lattner	552de0a	2008-11-23 08:16:56 +0000	[diff] [blame]	457	<p>The Clang front-end inherently depends on the location of a token being
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	458	tracked correctly. If it is ever incorrect, the front-end may get confused and
				459	die. The reason for this is that the notion of the 'spelling' of a Token in
Chris Lattner	552de0a	2008-11-23 08:16:56 +0000	[diff] [blame]	460	Clang depends on being able to find the original input characters for the token.
Chris Lattner	18376dd	2009-01-16 07:00:50 +0000	[diff] [blame]	461	This concept maps directly to the "spelling location" for the token.</p>
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	462
				463	<!-- ======================================================================= -->
				464	<h2 id="liblex">The Lexer and Preprocessor Library</h2>
				465	<!-- ======================================================================= -->
				466
				467	<p>The Lexer library contains several tightly-connected classes that are involved
				468	with the nasty process of lexing and preprocessing C source code. The main
				469	interface to this library for outside clients is the large <a
				470	href="#Preprocessor">Preprocessor</a> class.
				471	It contains the various pieces of state that are required to coherently read
				472	tokens out of a translation unit.</p>
				473
				474	<p>The core interface to the Preprocessor object (once it is set up) is the
				475	Preprocessor::Lex method, which returns the next <a href="#Token">Token</a> from
				476	the preprocessor stream. There are two types of token providers that the
				477	preprocessor is capable of reading from: a buffer lexer (provided by the <a
				478	href="#Lexer">Lexer</a> class) and a buffered token stream (provided by the <a
Chris Lattner	7928125	2008-03-09 02:27:26 +0000	[diff] [blame]	479	href="#TokenLexer">TokenLexer</a> class).
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	480
				481
				482	<!-- ======================================================================= -->
				483	<h3 id="Token">The Token class</h3>
				484	<!-- ======================================================================= -->
				485
				486	<p>The Token class is used to represent a single lexed token. Tokens are
				487	intended to be used by the lexer/preprocess and parser libraries, but are not
				488	intended to live beyond them (for example, they should not live in the ASTs).<p>
				489
				490	<p>Tokens most often live on the stack (or some other location that is efficient
				491	to access) as the parser is running, but occasionally do get buffered up. For
				492	example, macro definitions are stored as a series of tokens, and the C++
Chris Lattner	3fcbb89	2008-11-23 08:32:53 +0000	[diff] [blame]	493	front-end periodically needs to buffer tokens up for tentative parsing and
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	494	various pieces of look-ahead. As such, the size of a Token matter. On a 32-bit
				495	system, sizeof(Token) is currently 16 bytes.</p>
				496
Chris Lattner	3932fe0	2009-01-06 06:02:08 +0000	[diff] [blame]	497	<p>Tokens occur in two forms: "<a href="#AnnotationToken">Annotation
				498	Tokens</a>" and normal tokens. Normal tokens are those returned by the lexer,
				499	annotation tokens represent semantic information and are produced by the parser,
				500	replacing normal tokens in the token stream. Normal tokens contain the
				501	following information:</p>
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	502
				503	<ul>
				504	<li><b>A SourceLocation</b> - This indicates the location of the start of the
				505	token.</li>
				506
				507	<li><b>A length</b> - This stores the length of the token as stored in the
				508	SourceBuffer. For tokens that include them, this length includes trigraphs and
				509	escaped newlines which are ignored by later phases of the compiler. By pointing
				510	into the original source buffer, it is always possible to get the original
				511	spelling of a token completely accurately.</li>
				512
				513	<li><b>IdentifierInfo</b> - If a token takes the form of an identifier, and if
				514	identifier lookup was enabled when the token was lexed (e.g. the lexer was not
				515	reading in 'raw' mode) this contains a pointer to the unique hash value for the
				516	identifier. Because the lookup happens before keyword identification, this
				517	field is set even for language keywords like 'for'.</li>
				518
				519	<li><b>TokenKind</b> - This indicates the kind of token as classified by the
				520	lexer. This includes things like <tt>tok::starequal</tt> (for the "*="
				521	operator), <tt>tok::ampamp</tt> for the "&&" token, and keyword values
				522	(e.g. <tt>tok::kw_for</tt>) for identifiers that correspond to keywords. Note
				523	that some tokens can be spelled multiple ways. For example, C++ supports
				524	"operator keywords", where things like "and" are treated exactly like the
				525	"&&" operator. In these cases, the kind value is set to
				526	<tt>tok::ampamp</tt>, which is good for the parser, which doesn't have to
				527	consider both forms. For something that cares about which form is used (e.g.
				528	the preprocessor 'stringize' operator) the spelling indicates the original
				529	form.</li>
				530
				531	<li><b>Flags</b> - There are currently four flags tracked by the
				532	lexer/preprocessor system on a per-token basis:
				533
				534	<ol>
				535	<li><b>StartOfLine</b> - This was the first token that occurred on its input
				536	source line.</li>
				537	<li><b>LeadingSpace</b> - There was a space character either immediately
				538	before the token or transitively before the token as it was expanded
				539	through a macro. The definition of this flag is very closely defined by
				540	the stringizing requirements of the preprocessor.</li>
				541	<li><b>DisableExpand</b> - This flag is used internally to the preprocessor to
				542	represent identifier tokens which have macro expansion disabled. This
				543	prevents them from being considered as candidates for macro expansion ever
				544	in the future.</li>
				545	<li><b>NeedsCleaning</b> - This flag is set if the original spelling for the
				546	token includes a trigraph or escaped newline. Since this is uncommon,
				547	many pieces of code can fast-path on tokens that did not need cleaning.
				548	</p>
				549	</ol>
				550	</li>
				551	</ul>
				552
Chris Lattner	3932fe0	2009-01-06 06:02:08 +0000	[diff] [blame]	553	<p>One interesting (and somewhat unusual) aspect of normal tokens is that they
				554	don't contain any semantic information about the lexed value. For example, if
				555	the token was a pp-number token, we do not represent the value of the number
				556	that was lexed (this is left for later pieces of code to decide). Additionally,
				557	the lexer library has no notion of typedef names vs variable names: both are
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	558	returned as identifiers, and the parser is left to decide whether a specific
				559	identifier is a typedef or a variable (tracking this requires scope information
Chris Lattner	3932fe0	2009-01-06 06:02:08 +0000	[diff] [blame]	560	among other things). The parser can do this translation by replacing tokens
				561	returned by the preprocessor with "Annotation Tokens".</p>
				562
				563	<!-- ======================================================================= -->
				564	<h3 id="AnnotationToken">Annotation Tokens</h3>
				565	<!-- ======================================================================= -->
				566
				567	<p>Annotation Tokens are tokens that are synthesized by the parser and injected
				568	into the preprocessor's token stream (replacing existing tokens) to record
				569	semantic information found by the parser. For example, if "foo" is found to be
				570	a typedef, the "foo" <tt>tok::identifier</tt> token is replaced with an
				571	<tt>tok::annot_typename</tt>. This is useful for a couple of reasons: 1) this
				572	makes it easy to handle qualified type names (e.g. "foo::bar::baz<42>::t")
				573	in C++ as a single "token" in the parser. 2) if the parser backtracks, the
				574	reparse does not need to redo semantic analysis to determine whether a token
				575	sequence is a variable, type, template, etc.</p>
				576
				577	<p>Annotation Tokens are created by the parser and reinjected into the parser's
				578	token stream (when backtracking is enabled). Because they can only exist in
				579	tokens that the preprocessor-proper is done with, it doesn't need to keep around
				580	flags like "start of line" that the preprocessor uses to do its job.
				581	Additionally, an annotation token may "cover" a sequence of preprocessor tokens
				582	(e.g. <tt>a::b::c</tt> is five preprocessor tokens). As such, the valid fields
				583	of an annotation token are different than the fields for a normal token (but
				584	they are multiplexed into the normal Token fields):</p>
				585
				586	<ul>
				587	<li><b>SourceLocation "Location"</b> - The SourceLocation for the annotation
				588	token indicates the first token replaced by the annotation token. In the example
				589	above, it would be the location of the "a" identifier.</li>
				590
				591	<li><b>SourceLocation "AnnotationEndLoc"</b> - This holds the location of the
				592	last token replaced with the annotation token. In the example above, it would
				593	be the location of the "c" identifier.</li>
				594
				595	<li><b>void* "AnnotationValue"</b> - This contains an opaque object that the
				596	parser gets from Sema through an Actions module, it is passed around and Sema
				597	intepretes it, based on the type of annotation token.</li>
				598
				599	<li><b>TokenKind "Kind"</b> - This indicates the kind of Annotation token this
				600	is. See below for the different valid kinds.</li>
				601	</ul>
				602
				603	<p>Annotation tokens currently come in three kinds:</p>
				604
				605	<ol>
				606	<li><b>tok::annot_typename</b>: This annotation token represents a
				607	resolved typename token that is potentially qualified. The AnnotationValue
Steve Naroff	b43a50f	2009-01-28 19:39:02 +0000	[diff] [blame]	608	field contains a pointer returned by Action::getTypeName(). In the case of the
Chris Lattner	3932fe0	2009-01-06 06:02:08 +0000	[diff] [blame]	609	Sema actions module, this is a <tt>Decl*</tt> for the type.</li>
				610
				611	<li><b>tok::annot_cxxscope</b>: This annotation token represents a C++ scope
				612	specifier, such as "A::B::". This corresponds to the grammar productions "::"
				613	and ":: [opt] nested-name-specifier". The AnnotationValue pointer is returned
				614	by the Action::ActOnCXXGlobalScopeSpecifier and
				615	Action::ActOnCXXNestedNameSpecifier callbacks. In the case of Sema, this is a
				616	<tt>DeclContext*</tt>.</li>
				617
				618	<li><b>tok::annot_template_id</b>: This annotation token represents a C++
				619	template-id such as "foo<int, 4>", which may refer to a function or type
				620	depending on whether foo is a function template or class template. The
				621	AnnotationValue pointer is a pointer to a malloc'd TemplateIdAnnotation object.
				622	FIXME: I don't think the parsing logic is right for this. Shouldn't type
				623	templates be turned into annot_typename??</li>
				624
				625	</ol>
				626
Cedric Venet	da76b28	2009-01-06 16:22:54 +0000	[diff] [blame]	627	<p>As mentioned above, annotation tokens are not returned by the preprocessor,
Chris Lattner	3932fe0	2009-01-06 06:02:08 +0000	[diff] [blame]	628	they are formed on demand by the parser. This means that the parser has to be
				629	aware of cases where an annotation could occur and form it where appropriate.
				630	This is somewhat similar to how the parser handles Translation Phase 6 of C99:
				631	String Concatenation (see C99 5.1.1.2). In the case of string concatenation,
				632	the preprocessor just returns distinct tok::string_literal and
				633	tok::wide_string_literal tokens and the parser eats a sequence of them wherever
				634	the grammar indicates that a string literal can occur.</p>
				635
				636	<p>In order to do this, whenever the parser expects a tok::identifier or
				637	tok::coloncolon, it should call the TryAnnotateTypeOrScopeToken or
				638	TryAnnotateCXXScopeToken methods to form the annotation token. These methods
				639	will maximally form the specified annotation tokens and replace the current
				640	token with them, if applicable. If the current tokens is not valid for an
				641	annotation token, it will remain an identifier or :: token.</p>
				642
				643
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	644
				645	<!-- ======================================================================= -->
				646	<h3 id="Lexer">The Lexer class</h3>
				647	<!-- ======================================================================= -->
				648
				649	<p>The Lexer class provides the mechanics of lexing tokens out of a source
				650	buffer and deciding what they mean. The Lexer is complicated by the fact that
				651	it operates on raw buffers that have not had spelling eliminated (this is a
				652	necessity to get decent performance), but this is countered with careful coding
				653	as well as standard performance techniques (for example, the comment handling
				654	code is vectorized on X86 and PowerPC hosts).</p>
				655
				656	<p>The lexer has a couple of interesting modal features:</p>
				657
				658	<ul>
				659	<li>The lexer can operate in 'raw' mode. This mode has several features that
				660	make it possible to quickly lex the file (e.g. it stops identifier lookup,
				661	doesn't specially handle preprocessor tokens, handles EOF differently, etc).
				662	This mode is used for lexing within an "<tt>#if 0</tt>" block, for
				663	example.</li>
				664	<li>The lexer can capture and return comments as tokens. This is required to
				665	support the -C preprocessor mode, which passes comments through, and is
				666	used by the diagnostic checker to identifier expect-error annotations.</li>
				667	<li>The lexer can be in ParsingFilename mode, which happens when preprocessing
Chris Lattner	8438624	2007-09-16 19:25:23 +0000	[diff] [blame]	668	after reading a #include directive. This mode changes the parsing of '<'
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	669	to return an "angled string" instead of a bunch of tokens for each thing
				670	within the filename.</li>
				671	<li>When parsing a preprocessor directive (after "<tt>#</tt>") the
				672	ParsingPreprocessorDirective mode is entered. This changes the parser to
				673	return EOM at a newline.</li>
				674	<li>The Lexer uses a LangOptions object to know whether trigraphs are enabled,
				675	whether C++ or ObjC keywords are recognized, etc.</li>
				676	</ul>
				677
				678	<p>In addition to these modes, the lexer keeps track of a couple of other
				679	features that are local to a lexed buffer, which change as the buffer is
				680	lexed:</p>
				681
				682	<ul>
				683	<li>The Lexer uses BufferPtr to keep track of the current character being
				684	lexed.</li>
				685	<li>The Lexer uses IsAtStartOfLine to keep track of whether the next lexed token
				686	will start with its "start of line" bit set.</li>
				687	<li>The Lexer keeps track of the current #if directives that are active (which
				688	can be nested).</li>
				689	<li>The Lexer keeps track of an <a href="#MultipleIncludeOpt">
				690	MultipleIncludeOpt</a> object, which is used to
				691	detect whether the buffer uses the standard "<tt>#ifndef XX</tt> /
				692	<tt>#define XX</tt>" idiom to prevent multiple inclusion. If a buffer does,
				693	subsequent includes can be ignored if the XX macro is defined.</li>
				694	</ul>
				695
				696	<!-- ======================================================================= -->
Chris Lattner	7928125	2008-03-09 02:27:26 +0000	[diff] [blame]	697	<h3 id="TokenLexer">The TokenLexer class</h3>
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	698	<!-- ======================================================================= -->
				699
Chris Lattner	7928125	2008-03-09 02:27:26 +0000	[diff] [blame]	700	<p>The TokenLexer class is a token provider that returns tokens from a list
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	701	of tokens that came from somewhere else. It typically used for two things: 1)
				702	returning tokens from a macro definition as it is being expanded 2) returning
				703	tokens from an arbitrary buffer of tokens. The later use is used by _Pragma and
				704	will most likely be used to handle unbounded look-ahead for the C++ parser.</p>
				705
				706	<!-- ======================================================================= -->
				707	<h3 id="MultipleIncludeOpt">The MultipleIncludeOpt class</h3>
				708	<!-- ======================================================================= -->
				709
				710	<p>The MultipleIncludeOpt class implements a really simple little state machine
				711	that is used to detect the standard "<tt>#ifndef XX</tt> / <tt>#define XX</tt>"
				712	idiom that people typically use to prevent multiple inclusion of headers. If a
				713	buffer uses this idiom and is subsequently #include'd, the preprocessor can
				714	simply check to see whether the guarding condition is defined or not. If so,
				715	the preprocessor can completely ignore the include of the header.</p>
				716
				717
				718
				719	<!-- ======================================================================= -->
				720	<h2 id="libparse">The Parser Library</h2>
				721	<!-- ======================================================================= -->
				722
				723	<!-- ======================================================================= -->
				724	<h2 id="libast">The AST Library</h2>
				725	<!-- ======================================================================= -->
				726
				727	<!-- ======================================================================= -->
				728	<h3 id="Type">The Type class and its subclasses</h3>
				729	<!-- ======================================================================= -->
				730
				731	<p>The Type class (and its subclasses) are an important part of the AST. Types
				732	are accessed through the ASTContext class, which implicitly creates and uniques
				733	them as they are needed. Types have a couple of non-obvious features: 1) they
				734	do not capture type qualifiers like const or volatile (See
				735	<a href="#QualType">QualType</a>), and 2) they implicitly capture typedef
Chris Lattner	8a2bc62	2007-07-31 06:37:39 +0000	[diff] [blame]	736	information. Once created, types are immutable (unlike decls).</p>
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	737
				738	<p>Typedefs in C make semantic analysis a bit more complex than it would
				739	be without them. The issue is that we want to capture typedef information
				740	and represent it in the AST perfectly, but the semantics of operations need to
				741	"see through" typedefs. For example, consider this code:</p>
				742
				743	<code>
				744	void func() {<br>
Bill Wendling	30d1775	2007-10-06 01:56:01 +0000	[diff] [blame]	745	typedef int foo;<br>
				746	foo X, *Y;<br>
				747	typedef foo* bar;<br>
				748	bar Z;<br>
				749	*X; <i>// error</i><br>
				750	**Y; <i>// error</i><br>
				751	**Z; <i>// error</i><br>
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	752	}<br>
				753	</code>
				754
				755	<p>The code above is illegal, and thus we expect there to be diagnostics emitted
				756	on the annotated lines. In this example, we expect to get:</p>
				757
				758	<pre>
Chris Lattner	8a2bc62	2007-07-31 06:37:39 +0000	[diff] [blame]	759	<b>test.c:6:1: error: indirection requires pointer operand ('foo' invalid)</b>
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	760	*X; // error
				761	<font color="blue">^~</font>
Chris Lattner	8a2bc62	2007-07-31 06:37:39 +0000	[diff] [blame]	762	<b>test.c:7:1: error: indirection requires pointer operand ('foo' invalid)</b>
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	763	**Y; // error
				764	<font color="blue">^~~</font>
Chris Lattner	8a2bc62	2007-07-31 06:37:39 +0000	[diff] [blame]	765	<b>test.c:8:1: error: indirection requires pointer operand ('foo' invalid)</b>
				766	**Z; // error
				767	<font color="blue">^~~</font>
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	768	</pre>
				769
				770	<p>While this example is somewhat silly, it illustrates the point: we want to
				771	retain typedef information where possible, so that we can emit errors about
				772	"<tt>std::string</tt>" instead of "<tt>std::basic_string<char, std:...</tt>".
				773	Doing this requires properly keeping typedef information (for example, the type
				774	of "X" is "foo", not "int"), and requires properly propagating it through the
Chris Lattner	8a2bc62	2007-07-31 06:37:39 +0000	[diff] [blame]	775	various operators (for example, the type of *Y is "foo", not "int"). In order
				776	to retain this information, the type of these expressions is an instance of the
				777	TypedefType class, which indicates that the type of these expressions is a
				778	typedef for foo.
				779	</p>
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	780
Chris Lattner	8a2bc62	2007-07-31 06:37:39 +0000	[diff] [blame]	781	<p>Representing types like this is great for diagnostics, because the
				782	user-specified type is always immediately available. There are two problems
				783	with this: first, various semantic checks need to make judgements about the
Chris Lattner	33fc68a	2007-07-31 18:54:50 +0000	[diff] [blame]	784	<em>actual structure</em> of a type, ignoring typdefs. Second, we need an
				785	efficient way to query whether two types are structurally identical to each
				786	other, ignoring typedefs. The solution to both of these problems is the idea of
Chris Lattner	8a2bc62	2007-07-31 06:37:39 +0000	[diff] [blame]	787	canonical types.</p>
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	788
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	789	<!-- =============== -->
Chris Lattner	8a2bc62	2007-07-31 06:37:39 +0000	[diff] [blame]	790	<h4>Canonical Types</h4>
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	791	<!-- =============== -->
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	792
Chris Lattner	8a2bc62	2007-07-31 06:37:39 +0000	[diff] [blame]	793	<p>Every instance of the Type class contains a canonical type pointer. For
				794	simple types with no typedefs involved (e.g. "<tt>int</tt>", "<tt>int*</tt>",
				795	"<tt>int**</tt>"), the type just points to itself. For types that have a
				796	typedef somewhere in their structure (e.g. "<tt>foo</tt>", "<tt>foo*</tt>",
				797	"<tt>foo**</tt>", "<tt>bar</tt>"), the canonical type pointer points to their
				798	structurally equivalent type without any typedefs (e.g. "<tt>int</tt>",
				799	"<tt>int</tt>", "<tt>int</tt>", and "<tt>int</tt>" respectively).</p>
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	800
Chris Lattner	8a2bc62	2007-07-31 06:37:39 +0000	[diff] [blame]	801	<p>This design provides a constant time operation (dereferencing the canonical
				802	type pointer) that gives us access to the structure of types. For example,
				803	we can trivially tell that "bar" and "foo*" are the same type by dereferencing
				804	their canonical type pointers and doing a pointer comparison (they both point
				805	to the single "<tt>int*</tt>" type).</p>
				806
				807	<p>Canonical types and typedef types bring up some complexities that must be
				808	carefully managed. Specifically, the "isa/cast/dyncast" operators generally
				809	shouldn't be used in code that is inspecting the AST. For example, when type
				810	checking the indirection operator (unary '*' on a pointer), the type checker
				811	must verify that the operand has a pointer type. It would not be correct to
				812	check that with "<tt>isa<PointerType>(SubExpr->getType())</tt>",
				813	because this predicate would fail if the subexpression had a typedef type.</p>
				814
				815	<p>The solution to this problem are a set of helper methods on Type, used to
				816	check their properties. In this case, it would be correct to use
				817	"<tt>SubExpr->getType()->isPointerType()</tt>" to do the check. This
				818	predicate will return true if the <em>canonical type is a pointer</em>, which is
				819	true any time the type is structurally a pointer type. The only hard part here
				820	is remembering not to use the <tt>isa/cast/dyncast</tt> operations.</p>
				821
				822	<p>The second problem we face is how to get access to the pointer type once we
				823	know it exists. To continue the example, the result type of the indirection
				824	operator is the pointee type of the subexpression. In order to determine the
				825	type, we need to get the instance of PointerType that best captures the typedef
				826	information in the program. If the type of the expression is literally a
				827	PointerType, we can return that, otherwise we have to dig through the
				828	typedefs to find the pointer type. For example, if the subexpression had type
				829	"<tt>foo*</tt>", we could return that type as the result. If the subexpression
				830	had type "<tt>bar</tt>", we want to return "<tt>foo*</tt>" (note that we do
				831	<em>not</em> want "<tt>int*</tt>"). In order to provide all of this, Type has
Chris Lattner	11406c1	2007-07-31 16:50:51 +0000	[diff] [blame]	832	a getAsPointerType() method that checks whether the type is structurally a
Chris Lattner	8a2bc62	2007-07-31 06:37:39 +0000	[diff] [blame]	833	PointerType and, if so, returns the best one. If not, it returns a null
				834	pointer.</p>
				835
				836	<p>This structure is somewhat mystical, but after meditating on it, it will
				837	make sense to you :).</p>
Chris Lattner	86920d3	2007-07-31 05:42:17 +0000	[diff] [blame]	838
				839	<!-- ======================================================================= -->
				840	<h3 id="QualType">The QualType class</h3>
				841	<!-- ======================================================================= -->
				842
				843	<p>The QualType class is designed as a trivial value class that is small,
				844	passed by-value and is efficient to query. The idea of QualType is that it
				845	stores the type qualifiers (const, volatile, restrict) separately from the types
				846	themselves: QualType is conceptually a pair of "Type*" and bits for the type
				847	qualifiers.</p>
				848
				849	<p>By storing the type qualifiers as bits in the conceptual pair, it is
				850	extremely efficient to get the set of qualifiers on a QualType (just return the
				851	field of the pair), add a type qualifier (which is a trivial constant-time
				852	operation that sets a bit), and remove one or more type qualifiers (just return
				853	a QualType with the bitfield set to empty).</p>
				854
				855	<p>Further, because the bits are stored outside of the type itself, we do not
				856	need to create duplicates of types with different sets of qualifiers (i.e. there
				857	is only a single heap allocated "int" type: "const int" and "volatile const int"
				858	both point to the same heap allocated "int" type). This reduces the heap size
				859	used to represent bits and also means we do not have to consider qualifiers when
				860	uniquing types (<a href="#Type">Type</a> does not even contain qualifiers).</p>
				861
				862	<p>In practice, on hosts where it is safe, the 3 type qualifiers are stored in
				863	the low bit of the pointer to the Type object. This means that QualType is
				864	exactly the same size as a pointer, and this works fine on any system where
				865	malloc'd objects are at least 8 byte aligned.</p>
Ted Kremenek	8bc0571	2007-10-10 23:01:43 +0000	[diff] [blame]	866
				867	<!-- ======================================================================= -->
Douglas Gregor	2e1cd42	2008-11-17 14:58:09 +0000	[diff] [blame]	868	<h3 id="DeclarationName">Declaration names</h3>
				869	<!-- ======================================================================= -->
				870
				871	<p>The <tt>DeclarationName</tt> class represents the name of a
				872	declaration in Clang. Declarations in the C family of languages can
Chris Lattner	3fcbb89	2008-11-23 08:32:53 +0000	[diff] [blame]	873	take several different forms. Most declarations are named by
Douglas Gregor	2e1cd42	2008-11-17 14:58:09 +0000	[diff] [blame]	874	simple identifiers, e.g., "<code>f</code>" and "<code>x</code>" in
				875	the function declaration <code>f(int x)</code>. In C++, declaration
				876	names can also name class constructors ("<code>Class</code>"
				877	in <code>struct Class { Class(); }</code>), class destructors
				878	("<code>~Class</code>"), overloaded operator names ("operator+"),
				879	and conversion functions ("<code>operator void const *</code>"). In
				880	Objective-C, declaration names can refer to the names of Objective-C
				881	methods, which involve the method name and the parameters,
Chris Lattner	3fcbb89	2008-11-23 08:32:53 +0000	[diff] [blame]	882	collectively called a <i>selector</i>, e.g.,
Douglas Gregor	2e1cd42	2008-11-17 14:58:09 +0000	[diff] [blame]	883	"<code>setWidth:height:</code>". Since all of these kinds of
Chris Lattner	3fcbb89	2008-11-23 08:32:53 +0000	[diff] [blame]	884	entities - variables, functions, Objective-C methods, C++
				885	constructors, destructors, and operators - are represented as
Douglas Gregor	2e1cd42	2008-11-17 14:58:09 +0000	[diff] [blame]	886	subclasses of Clang's common <code>NamedDecl</code>
				887	class, <code>DeclarationName</code> is designed to efficiently
				888	represent any kind of name.</p>
				889
				890	<p>Given
				891	a <code>DeclarationName</code> <code>N</code>, <code>N.getNameKind()</code>
Douglas Gregor	2def483	2008-11-17 20:34:05 +0000	[diff] [blame]	892	will produce a value that describes what kind of name <code>N</code>
Douglas Gregor	e94ca9e4	2008-11-18 14:39:36 +0000	[diff] [blame]	893	stores. There are 8 options (all of the names are inside
Douglas Gregor	2e1cd42	2008-11-17 14:58:09 +0000	[diff] [blame]	894	the <code>DeclarationName</code> class)</p>
				895	<dl>
				896	<dt>Identifier</dt>
				897	<dd>The name is a simple
				898	identifier. Use <code>N.getAsIdentifierInfo()</code> to retrieve the
				899	corresponding <code>IdentifierInfo*</code> pointing to the actual
				900	identifier. Note that C++ overloaded operators (e.g.,
				901	"<code>operator+</code>") are represented as special kinds of
				902	identifiers. Use <code>IdentifierInfo</code>'s <code>getOverloadedOperatorID</code>
				903	function to determine whether an identifier is an overloaded
				904	operator name.</dd>
				905
				906	<dt>ObjCZeroArgSelector, ObjCOneArgSelector,
				907	ObjCMultiArgSelector</dt>
				908	<dd>The name is an Objective-C selector, which can be retrieved as a
				909	<code>Selector</code> instance
				910	via <code>N.getObjCSelector()</code>. The three possible name
				911	kinds for Objective-C reflect an optimization within
				912	the <code>DeclarationName</code> class: both zero- and
				913	one-argument selectors are stored as a
				914	masked <code>IdentifierInfo</code> pointer, and therefore require
				915	very little space, since zero- and one-argument selectors are far
				916	more common than multi-argument selectors (which use a different
				917	structure).</dd>
				918
				919	<dt>CXXConstructorName</dt>
				920	<dd>The name is a C++ constructor
				921	name. Use <code>N.getCXXNameType()</code> to retrieve
				922	the <a href="#QualType">type</a> that this constructor is meant to
				923	construct. The type is always the canonical type, since all
				924	constructors for a given type have the same name.</dd>
				925
				926	<dt>CXXDestructorName</dt>
				927	<dd>The name is a C++ destructor
				928	name. Use <code>N.getCXXNameType()</code> to retrieve
				929	the <a href="#QualType">type</a> whose destructor is being
				930	named. This type is always a canonical type.</dd>
				931
				932	<dt>CXXConversionFunctionName</dt>
				933	<dd>The name is a C++ conversion function. Conversion functions are
				934	named according to the type they convert to, e.g., "<code>operator void
				935	const *</code>". Use <code>N.getCXXNameType()</code> to retrieve
				936	the type that this conversion function converts to. This type is
				937	always a canonical type.</dd>
Douglas Gregor	e94ca9e4	2008-11-18 14:39:36 +0000	[diff] [blame]	938
				939	<dt>CXXOperatorName</dt>
				940	<dd>The name is a C++ overloaded operator name. Overloaded operators
				941	are named according to their spelling, e.g.,
				942	"<code>operator+</code>" or "<code>operator new
				943	[]</code>". Use <code>N.getCXXOverloadedOperator()</code> to
				944	retrieve the overloaded operator (a value of
				945	type <code>OverloadedOperatorKind</code>).</dd>
Douglas Gregor	2e1cd42	2008-11-17 14:58:09 +0000	[diff] [blame]	946	</dl>
				947
				948	<p><code>DeclarationName</code>s are cheap to create, copy, and
				949	compare. They require only a single pointer's worth of storage in
Douglas Gregor	e94ca9e4	2008-11-18 14:39:36 +0000	[diff] [blame]	950	the common cases (identifiers, zero-
Douglas Gregor	2e1cd42	2008-11-17 14:58:09 +0000	[diff] [blame]	951	and one-argument Objective-C selectors) and use dense, uniqued
				952	storage for the other kinds of
				953	names. Two <code>DeclarationName</code>s can be compared for
				954	equality (<code>==</code>, <code>!=</code>) using a simple bitwise
				955	comparison, can be ordered
				956	with <code><</code>, <code>></code>, <code><=</code>,
				957	and <code>>=</code> (which provide a lexicographical ordering for
				958	normal identifiers but an unspecified ordering for other kinds of
				959	names), and can be placed into LLVM <code>DenseMap</code>s
				960	and <code>DenseSet</code>s.</p>
				961
				962	<p><code>DeclarationName</code> instances can be created in different
				963	ways depending on what kind of name the instance will store. Normal
Douglas Gregor	e94ca9e4	2008-11-18 14:39:36 +0000	[diff] [blame]	964	identifiers (<code>IdentifierInfo</code> pointers) and Objective-C selectors
Douglas Gregor	2e1cd42	2008-11-17 14:58:09 +0000	[diff] [blame]	965	(<code>Selector</code>) can be implicitly converted
				966	to <code>DeclarationName</code>s. Names for C++ constructors,
Douglas Gregor	e94ca9e4	2008-11-18 14:39:36 +0000	[diff] [blame]	967	destructors, conversion functions, and overloaded operators can be retrieved from
Douglas Gregor	2e1cd42	2008-11-17 14:58:09 +0000	[diff] [blame]	968	the <code>DeclarationNameTable</code>, an instance of which is
				969	available as <code>ASTContext::DeclarationNames</code>. The member
				970	functions <code>getCXXConstructorName</code>, <code>getCXXDestructorName</code>,
Douglas Gregor	e94ca9e4	2008-11-18 14:39:36 +0000	[diff] [blame]	971	<code>getCXXConversionFunctionName</code>, and <code>getCXXOperatorName</code>, respectively,
				972	return <code>DeclarationName</code> instances for the four kinds of
Douglas Gregor	2e1cd42	2008-11-17 14:58:09 +0000	[diff] [blame]	973	C++ special function names.</p>
				974
				975	<!-- ======================================================================= -->
Douglas Gregor	074149e	2009-01-05 19:45:36 +0000	[diff] [blame]	976	<h3 id="DeclContext">Declaration contexts</h3>
				977	<!-- ======================================================================= -->
				978	<p>Every declaration in a program exists within some <i>declaration
				979	context</i>, such as a translation unit, namespace, class, or
				980	function. Declaration contexts in Clang are represented by
				981	the <code>DeclContext</code> class, from which the various
				982	declaration-context AST nodes
				983	(<code>TranslationUnitDecl</code>, <code>NamespaceDecl</code>, <code>RecordDecl</code>, <code>FunctionDecl</code>,
				984	etc.) will derive. The <code>DeclContext</code> class provides
				985	several facilities common to each declaration context:</p>
				986	<dl>
				987	<dt>Source-centric vs. Semantics-centric View of Declarations</dt>
				988	<dd><code>DeclContext</code> provides two views of the declarations
				989	stored within a declaration context. The source-centric view
				990	accurately represents the program source code as written, including
				991	multiple declarations of entities where present (see the
				992	section <a href="#Redeclarations">Redeclarations and
				993	Overloads</a>), while the semantics-centric view represents the
				994	program semantics. The two views are kept synchronized by semantic
				995	analysis while the ASTs are being constructed.</dd>
				996
				997	<dt>Storage of declarations within that context</dt>
				998	<dd>Every declaration context can contain some number of
				999	declarations. For example, a C++ class (represented
				1000	by <code>RecordDecl</code>) contains various member functions,
				1001	fields, nested types, and so on. All of these declarations will be
				1002	stored within the <code>DeclContext</code>, and one can iterate
				1003	over the declarations via
				1004	[<code>DeclContext::decls_begin()</code>,
				1005	<code>DeclContext::decls_end()</code>). This mechanism provides
				1006	the source-centric view of declarations in the context.</dd>
				1007
				1008	<dt>Lookup of declarations within that context</dt>
				1009	<dd>The <code>DeclContext</code> structure provides efficient name
				1010	lookup for names within that declaration context. For example,
				1011	if <code>N</code> is a namespace we can look for the
				1012	name <code>N::f</code>
				1013	using <code>DeclContext::lookup</code>. The lookup itself is
				1014	based on a lazily-constructed array (for declaration contexts
				1015	with a small number of declarations) or hash table (for
				1016	declaration contexts with more declarations). The lookup
				1017	operation provides the semantics-centric view of the declarations
				1018	in the context.</dd>
				1019
				1020	<dt>Ownership of declarations</dt>
				1021	<dd>The <code>DeclContext</code> owns all of the declarations that
				1022	were declared within its declaration context, and is responsible
				1023	for the management of their memory as well as their
				1024	(de-)serialization.</dd>
				1025	</dl>
				1026
Douglas Gregor	4afa39d	2009-01-20 01:17:11 +0000	[diff] [blame]	1027	<p>All declarations are stored within a declaration context, and one
				1028	can query
				1029	information about the context in which each declaration lives. One
Douglas Gregor	074149e	2009-01-05 19:45:36 +0000	[diff] [blame]	1030	can retrieve the <code>DeclContext</code> that contains a
Douglas Gregor	4afa39d	2009-01-20 01:17:11 +0000	[diff] [blame]	1031	particular <code>Decl</code>
				1032	using <code>Decl::getDeclContext</code>. However, see the
Douglas Gregor	074149e	2009-01-05 19:45:36 +0000	[diff] [blame]	1033	section <a href="#LexicalAndSemanticContexts">Lexical and Semantic
				1034	Contexts</a> for more information about how to interpret this
				1035	context information.</p>
				1036
				1037	<h4 id="Redeclarations">Redeclarations and Overloads</h4>
				1038	<p>Within a translation unit, it is common for an entity to be
				1039	declared several times. For example, we might declare a function "f"
				1040	and then later re-declare it as part of an inlined definition:</p>
				1041
				1042	<pre>
				1043	void f(int x, int y, int z = 1);
				1044
				1045	inline void f(int x, int y, int z) { /* ... */ }
				1046	</pre>
				1047
				1048	<p>The representation of "f" differs in the source-centric and
				1049	semantics-centric views of a declaration context. In the
				1050	source-centric view, all redeclarations will be present, in the
				1051	order they occurred in the source code, making
				1052	this view suitable for clients that wish to see the structure of
				1053	the source code. In the semantics-centric view, only the most recent "f"
				1054	will be found by the lookup, since it effectively replaces the first
				1055	declaration of "f".</p>
				1056
				1057	<p>In the semantics-centric view, overloading of functions is
				1058	represented explicitly. For example, given two declarations of a
				1059	function "g" that are overloaded, e.g.,</p>
				1060	<pre>
				1061	void g();
				1062	void g(int);
				1063	</pre>
				1064	<p>the <code>DeclContext::lookup</code> operation will return
				1065	an <code>OverloadedFunctionDecl</code> that contains both
				1066	declarations of "g". Clients that perform semantic analysis on a
				1067	program that is not concerned with the actual source code will
				1068	primarily use this semantics-centric view.</p>
				1069
				1070	<h4 id="LexicalAndSemanticContexts">Lexical and Semantic Contexts</h4>
Douglas Gregor	4afa39d	2009-01-20 01:17:11 +0000	[diff] [blame]	1071	<p>Each declaration has two potentially different
Douglas Gregor	074149e	2009-01-05 19:45:36 +0000	[diff] [blame]	1072	declaration contexts: a <i>lexical</i> context, which corresponds to
				1073	the source-centric view of the declaration context, and
				1074	a <i>semantic</i> context, which corresponds to the
				1075	semantics-centric view. The lexical context is accessible
Douglas Gregor	4afa39d	2009-01-20 01:17:11 +0000	[diff] [blame]	1076	via <code>Decl::getLexicalDeclContext</code> while the
Douglas Gregor	074149e	2009-01-05 19:45:36 +0000	[diff] [blame]	1077	semantic context is accessible
Douglas Gregor	4afa39d	2009-01-20 01:17:11 +0000	[diff] [blame]	1078	via <code>Decl::getDeclContext</code>, both of which return
Douglas Gregor	074149e	2009-01-05 19:45:36 +0000	[diff] [blame]	1079	<code>DeclContext</code> pointers. For most declarations, the two
				1080	contexts are identical. For example:</p>
				1081
				1082	<pre>
				1083	class X {
				1084	public:
				1085	void f(int x);
				1086	};
				1087	</pre>
				1088
				1089	<p>Here, the semantic and lexical contexts of <code>X::f</code> are
				1090	the <code>DeclContext</code> associated with the
				1091	class <code>X</code> (itself stored as a <code>RecordDecl</code> AST
				1092	node). However, we can now define <code>X::f</code> out-of-line:</p>
				1093
				1094	<pre>
				1095	void X::f(int x = 17) { /* ... */ }
				1096	</pre>
				1097
				1098	<p>This definition of has different lexical and semantic
				1099	contexts. The lexical context corresponds to the declaration
				1100	context in which the actual declaration occurred in the source
				1101	code, e.g., the translation unit containing <code>X</code>. Thus,
				1102	this declaration of <code>X::f</code> can be found by traversing
				1103	the declarations provided by
				1104	[<code>decls_begin()</code>, <code>decls_end()</code>) in the
				1105	translation unit.</p>
				1106
				1107	<p>The semantic context of <code>X::f</code> corresponds to the
				1108	class <code>X</code>, since this member function is (semantically) a
				1109	member of <code>X</code>. Lookup of the name <code>f</code> into
				1110	the <code>DeclContext</code> associated with <code>X</code> will
				1111	then return the definition of <code>X::f</code> (including
				1112	information about the default argument).</p>
				1113
				1114	<h4 id="TransparentContexts">Transparent Declaration Contexts</h4>
				1115	<p>In C and C++, there are several contexts in which names that are
				1116	logically declared inside another declaration will actually "leak"
				1117	out into the enclosing scope from the perspective of name
				1118	lookup. The most obvious instance of this behavior is in
				1119	enumeration types, e.g.,</p>
				1120	<pre>
				1121	enum Color {
				1122	Red,
				1123	Green,
				1124	Blue
				1125	};
				1126	</pre>
				1127
				1128	<p>Here, <code>Color</code> is an enumeration, which is a declaration
				1129	context that contains the
				1130	enumerators <code>Red</code>, <code>Green</code>,
				1131	and <code>Blue</code>. Thus, traversing the list of declarations
				1132	contained in the enumeration <code>Color</code> will
				1133	yield <code>Red</code>, <code>Green</code>,
				1134	and <code>Blue</code>. However, outside of the scope
				1135	of <code>Color</code> one can name the enumerator <code>Red</code>
				1136	without qualifying the name, e.g.,</p>
				1137
				1138	<pre>
				1139	Color c = Red;
				1140	</pre>
				1141
				1142	<p>There are other entities in C++ that provide similar behavior. For
				1143	example, linkage specifications that use curly braces:</p>
				1144
				1145	<pre>
				1146	extern "C" {
				1147	void f(int);
				1148	void g(int);
				1149	}
				1150	// f and g are visible here
				1151	</pre>
				1152
				1153	<p>For source-level accuracy, we treat the linkage specification and
				1154	enumeration type as a
				1155	declaration context in which its enclosed declarations ("Red",
				1156	"Green", and "Blue"; "f" and "g")
				1157	are declared. However, these declarations are visible outside of the
				1158	scope of the declaration context.</p>
				1159
				1160	<p>These language features (and several others, described below) have
				1161	roughly the same set of
				1162	requirements: declarations are declared within a particular lexical
				1163	context, but the declarations are also found via name lookup in
				1164	scopes enclosing the declaration itself. This feature is implemented
				1165	via <i>transparent</i> declaration contexts
				1166	(see <code>DeclContext::isTransparentContext()</code>), whose
				1167	declarations are visible in the nearest enclosing non-transparent
				1168	declaration context. This means that the lexical context of the
				1169	declaration (e.g., an enumerator) will be the
				1170	transparent <code>DeclContext</code> itself, as will the semantic
				1171	context, but the declaration will be visible in every outer context
				1172	up to and including the first non-transparent declaration context (since
				1173	transparent declaration contexts can be nested).</p>
				1174
				1175	<p>The transparent <code>DeclContexts</code> are:</p>
				1176	<ul>
				1177	<li>Enumerations (but not C++0x "scoped enumerations"):
				1178	<pre>
				1179	enum Color {
				1180	Red,
				1181	Green,
				1182	Blue
				1183	};
				1184	// Red, Green, and Blue are in scope
				1185	</pre></li>
				1186	<li>C++ linkage specifications:
				1187	<pre>
				1188	extern "C" {
				1189	void f(int);
				1190	void g(int);
				1191	}
				1192	// f and g are in scope
				1193	</pre></li>
				1194	<li>Anonymous unions and structs:
				1195	<pre>
				1196	struct LookupTable {
				1197	bool IsVector;
				1198	union {
				1199	std::vector<Item> *Vector;
				1200	std::set<Item> *Set;
				1201	};
				1202	};
				1203
				1204	LookupTable LT;
				1205	LT.Vector = 0; // Okay: finds Vector inside the unnamed union
				1206	</pre>
				1207	</li>
				1208	<li>C++0x inline namespaces:
				1209	<pre>
				1210	namespace mylib {
				1211	inline namespace debug {
				1212	class X;
				1213	}
				1214	}
				1215	mylib::X *xp; // okay: mylib::X refers to mylib::debug::X
				1216	</pre>
				1217	</li>
				1218	</ul>
				1219
				1220
				1221	<h4 id="MultiDeclContext">Multiply-Defined Declaration Contexts</h4>
				1222	<p>C++ namespaces have the interesting--and, so far, unique--property that
				1223	the namespace can be defined multiple times, and the declarations
				1224	provided by each namespace definition are effectively merged (from
				1225	the semantic point of view). For example, the following two code
				1226	snippets are semantically indistinguishable:</p>
				1227	<pre>
				1228	// Snippet #1:
				1229	namespace N {
				1230	void f();
				1231	}
				1232	namespace N {
				1233	void f(int);
				1234	}
				1235
				1236	// Snippet #2:
				1237	namespace N {
				1238	void f();
				1239	void f(int);
				1240	}
				1241	</pre>
				1242
				1243	<p>In Clang's representation, the source-centric view of declaration
				1244	contexts will actually have two separate <code>NamespaceDecl</code>
				1245	nodes in Snippet #1, each of which is a declaration context that
				1246	contains a single declaration of "f". However, the semantics-centric
				1247	view provided by name lookup into the namespace <code>N</code> for
				1248	"f" will return an <code>OverloadedFunctionDecl</code> that contains
				1249	both declarations of "f".</p>
				1250
				1251	<p><code>DeclContext</code> manages multiply-defined declaration
				1252	contexts internally. The
				1253	function <code>DeclContext::getPrimaryContext</code> retrieves the
				1254	"primary" context for a given <code>DeclContext</code> instance,
				1255	which is the <code>DeclContext</code> responsible for maintaining
				1256	the lookup table used for the semantics-centric view. Given the
				1257	primary context, one can follow the chain
				1258	of <code>DeclContext</code> nodes that define additional
				1259	declarations via <code>DeclContext::getNextContext</code>. Note that
				1260	these functions are used internally within the lookup and insertion
				1261	methods of the <code>DeclContext</code>, so the vast majority of
				1262	clients can ignore them.</p>
				1263
				1264	<!-- ======================================================================= -->
Ted Kremenek	8bc0571	2007-10-10 23:01:43 +0000	[diff] [blame]	1265	<h3 id="CFG">The <tt>CFG</tt> class</h3>
				1266	<!-- ======================================================================= -->
				1267
				1268	<p>The <tt>CFG</tt> class is designed to represent a source-level
				1269	control-flow graph for a single statement (<tt>Stmt*</tt>). Typically
				1270	instances of <tt>CFG</tt> are constructed for function bodies (usually
				1271	an instance of <tt>CompoundStmt</tt>), but can also be instantiated to
				1272	represent the control-flow of any class that subclasses <tt>Stmt</tt>,
				1273	which includes simple expressions. Control-flow graphs are especially
				1274	useful for performing
				1275	<a href="http://en.wikipedia.org/wiki/Data_flow_analysis#Sensitivities">flow-
				1276	or path-sensitive</a> program analyses on a given function.</p>
				1277
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	1278	<!-- ============ -->
Ted Kremenek	8bc0571	2007-10-10 23:01:43 +0000	[diff] [blame]	1279	<h4>Basic Blocks</h4>
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	1280	<!-- ============ -->
Ted Kremenek	8bc0571	2007-10-10 23:01:43 +0000	[diff] [blame]	1281
				1282	<p>Concretely, an instance of <tt>CFG</tt> is a collection of basic
				1283	blocks. Each basic block is an instance of <tt>CFGBlock</tt>, which
				1284	simply contains an ordered sequence of <tt>Stmt*</tt> (each referring
				1285	to statements in the AST). The ordering of statements within a block
				1286	indicates unconditional flow of control from one statement to the
				1287	next. <a href="#ConditionalControlFlow">Conditional control-flow</a>
				1288	is represented using edges between basic blocks. The statements
				1289	within a given <tt>CFGBlock</tt> can be traversed using
				1290	the <tt>CFGBlock::*iterator</tt> interface.</p>
				1291
				1292	<p>
Ted Kremenek	18e17e7	2007-10-18 22:50:52 +0000	[diff] [blame]	1293	A <tt>CFG</tt> object owns the instances of <tt>CFGBlock</tt> within
Ted Kremenek	8bc0571	2007-10-10 23:01:43 +0000	[diff] [blame]	1294	the control-flow graph it represents. Each <tt>CFGBlock</tt> within a
				1295	CFG is also uniquely numbered (accessible
				1296	via <tt>CFGBlock::getBlockID()</tt>). Currently the number is
				1297	based on the ordering the blocks were created, but no assumptions
				1298	should be made on how <tt>CFGBlock</tt>s are numbered other than their
				1299	numbers are unique and that they are numbered from 0..N-1 (where N is
				1300	the number of basic blocks in the CFG).</p>
				1301
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	1302	<!-- ===================== -->
Ted Kremenek	8bc0571	2007-10-10 23:01:43 +0000	[diff] [blame]	1303	<h4>Entry and Exit Blocks</h4>
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	1304	<!-- ===================== -->
Ted Kremenek	8bc0571	2007-10-10 23:01:43 +0000	[diff] [blame]	1305
				1306	Each instance of <tt>CFG</tt> contains two special blocks:
				1307	an <i>entry</i> block (accessible via <tt>CFG::getEntry()</tt>), which
				1308	has no incoming edges, and an <i>exit</i> block (accessible
				1309	via <tt>CFG::getExit()</tt>), which has no outgoing edges. Neither
				1310	block contains any statements, and they serve the role of providing a
				1311	clear entrance and exit for a body of code such as a function body.
				1312	The presence of these empty blocks greatly simplifies the
				1313	implementation of many analyses built on top of CFGs.
				1314
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	1315	<!-- ===================================================== -->
Ted Kremenek	8bc0571	2007-10-10 23:01:43 +0000	[diff] [blame]	1316	<h4 id ="ConditionalControlFlow">Conditional Control-Flow</h4>
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	1317	<!-- ===================================================== -->
Ted Kremenek	8bc0571	2007-10-10 23:01:43 +0000	[diff] [blame]	1318
				1319	<p>Conditional control-flow (such as those induced by if-statements
				1320	and loops) is represented as edges between <tt>CFGBlock</tt>s.
				1321	Because different C language constructs can induce control-flow,
				1322	each <tt>CFGBlock</tt> also records an extra <tt>Stmt*</tt> that
				1323	represents the <i>terminator</i> of the block. A terminator is simply
				1324	the statement that caused the control-flow, and is used to identify
				1325	the nature of the conditional control-flow between blocks. For
				1326	example, in the case of an if-statement, the terminator refers to
				1327	the <tt>IfStmt</tt> object in the AST that represented the given
				1328	branch.</p>
				1329
				1330	<p>To illustrate, consider the following code example:</p>
				1331
				1332	<code>
				1333	int foo(int x) {<br>
				1334	x = x + 1;<br>
				1335	<br>
				1336	if (x > 2) x++;<br>
				1337	else {<br>
				1338	x += 2;<br>
				1339	x *= 2;<br>
				1340	}<br>
				1341	<br>
				1342	return x;<br>
				1343	}
				1344	</code>
				1345
				1346	<p>After invoking the parser+semantic analyzer on this code fragment,
				1347	the AST of the body of <tt>foo</tt> is referenced by a
				1348	single <tt>Stmt*</tt>. We can then construct an instance
				1349	of <tt>CFG</tt> representing the control-flow graph of this function
				1350	body by single call to a static class method:</p>
				1351
				1352	<code>
				1353	Stmt* FooBody = ...<br>
				1354	CFG* FooCFG = <b>CFG::buildCFG</b>(FooBody);
				1355	</code>
				1356
				1357	<p>It is the responsibility of the caller of <tt>CFG::buildCFG</tt>
				1358	to <tt>delete</tt> the returned <tt>CFG*</tt> when the CFG is no
				1359	longer needed.</p>
				1360
				1361	<p>Along with providing an interface to iterate over
				1362	its <tt>CFGBlock</tt>s, the <tt>CFG</tt> class also provides methods
				1363	that are useful for debugging and visualizing CFGs. For example, the
				1364	method
				1365	<tt>CFG::dump()</tt> dumps a pretty-printed version of the CFG to
				1366	standard error. This is especially useful when one is using a
				1367	debugger such as gdb. For example, here is the output
				1368	of <tt>FooCFG->dump()</tt>:</p>
				1369
				1370	<code>
				1371	[ B5 (ENTRY) ]<br>
				1372	Predecessors (0):<br>
				1373	Successors (1): B4<br>
				1374	<br>
				1375	[ B4 ]<br>
				1376	1: x = x + 1<br>
				1377	2: (x > 2)<br>
				1378	<b>T: if [B4.2]</b><br>
				1379	Predecessors (1): B5<br>
				1380	Successors (2): B3 B2<br>
				1381	<br>
				1382	[ B3 ]<br>
				1383	1: x++<br>
				1384	Predecessors (1): B4<br>
				1385	Successors (1): B1<br>
				1386	<br>
				1387	[ B2 ]<br>
				1388	1: x += 2<br>
				1389	2: x *= 2<br>
				1390	Predecessors (1): B4<br>
				1391	Successors (1): B1<br>
				1392	<br>
				1393	[ B1 ]<br>
				1394	1: return x;<br>
				1395	Predecessors (2): B2 B3<br>
				1396	Successors (1): B0<br>
				1397	<br>
				1398	[ B0 (EXIT) ]<br>
				1399	Predecessors (1): B1<br>
				1400	Successors (0):
				1401	</code>
				1402
				1403	<p>For each block, the pretty-printed output displays for each block
				1404	the number of <i>predecessor</i> blocks (blocks that have outgoing
				1405	control-flow to the given block) and <i>successor</i> blocks (blocks
				1406	that have control-flow that have incoming control-flow from the given
				1407	block). We can also clearly see the special entry and exit blocks at
				1408	the beginning and end of the pretty-printed output. For the entry
				1409	block (block B5), the number of predecessor blocks is 0, while for the
				1410	exit block (block B0) the number of successor blocks is 0.</p>
				1411
				1412	<p>The most interesting block here is B4, whose outgoing control-flow
				1413	represents the branching caused by the sole if-statement
				1414	in <tt>foo</tt>. Of particular interest is the second statement in
				1415	the block, <b><tt>(x > 2)</tt></b>, and the terminator, printed
				1416	as <b><tt>if [B4.2]</tt></b>. The second statement represents the
				1417	evaluation of the condition of the if-statement, which occurs before
				1418	the actual branching of control-flow. Within the <tt>CFGBlock</tt>
				1419	for B4, the <tt>Stmt*</tt> for the second statement refers to the
				1420	actual expression in the AST for <b><tt>(x > 2)</tt></b>. Thus
				1421	pointers to subclasses of <tt>Expr</tt> can appear in the list of
				1422	statements in a block, and not just subclasses of <tt>Stmt</tt> that
				1423	refer to proper C statements.</p>
				1424
				1425	<p>The terminator of block B4 is a pointer to the <tt>IfStmt</tt>
				1426	object in the AST. The pretty-printer outputs <b><tt>if
				1427	[B4.2]</tt></b> because the condition expression of the if-statement
				1428	has an actual place in the basic block, and thus the terminator is
				1429	essentially
				1430	<i>referring</i> to the expression that is the second statement of
				1431	block B4 (i.e., B4.2). In this manner, conditions for control-flow
				1432	(which also includes conditions for loops and switch statements) are
				1433	hoisted into the actual basic block.</p>
				1434
Chris Lattner	62fd278	2008-11-22 21:41:31 +0000	[diff] [blame]	1435	<!-- ===================== -->
				1436	<!-- <h4>Implicit Control-Flow</h4> -->
				1437	<!-- ===================== -->
Ted Kremenek	8bc0571	2007-10-10 23:01:43 +0000	[diff] [blame]	1438
				1439	<!--
				1440	<p>A key design principle of the <tt>CFG</tt> class was to not require
				1441	any transformations to the AST in order to represent control-flow.
				1442	Thus the <tt>CFG</tt> does not perform any "lowering" of the
				1443	statements in an AST: loops are not transformed into guarded gotos,
				1444	short-circuit operations are not converted to a set of if-statements,
				1445	and so on.</p>
				1446	-->
Ted Kremenek	17a295d	2008-06-11 06:19:49 +0000	[diff] [blame]	1447
Chris Lattner	7bad199	2008-11-16 21:48:07 +0000	[diff] [blame]	1448
				1449	<!-- ======================================================================= -->
				1450	<h3 id="Constants">Constant Folding in the Clang AST</h3>
				1451	<!-- ======================================================================= -->
				1452
				1453	<p>There are several places where constants and constant folding matter a lot to
				1454	the Clang front-end. First, in general, we prefer the AST to retain the source
				1455	code as close to how the user wrote it as possible. This means that if they
				1456	wrote "5+4", we want to keep the addition and two constants in the AST, we don't
				1457	want to fold to "9". This means that constant folding in various ways turns
				1458	into a tree walk that needs to handle the various cases.</p>
				1459
				1460	<p>However, there are places in both C and C++ that require constants to be
				1461	folded. For example, the C standard defines what an "integer constant
				1462	expression" (i-c-e) is with very precise and specific requirements. The
				1463	language then requires i-c-e's in a lot of places (for example, the size of a
				1464	bitfield, the value for a case statement, etc). For these, we have to be able
				1465	to constant fold the constants, to do semantic checks (e.g. verify bitfield size
				1466	is non-negative and that case statements aren't duplicated). We aim for Clang
				1467	to be very pedantic about this, diagnosing cases when the code does not use an
				1468	i-c-e where one is required, but accepting the code unless running with
				1469	<tt>-pedantic-errors</tt>.</p>
				1470
				1471	<p>Things get a little bit more tricky when it comes to compatibility with
				1472	real-world source code. Specifically, GCC has historically accepted a huge
				1473	superset of expressions as i-c-e's, and a lot of real world code depends on this
				1474	unfortuate accident of history (including, e.g., the glibc system headers). GCC
				1475	accepts anything its "fold" optimizer is capable of reducing to an integer
				1476	constant, which means that the definition of what it accepts changes as its
				1477	optimizer does. One example is that GCC accepts things like "case X-X:" even
				1478	when X is a variable, because it can fold this to 0.</p>
				1479
				1480	<p>Another issue are how constants interact with the extensions we support, such
				1481	as __builtin_constant_p, __builtin_inf, __extension__ and many others. C99
				1482	obviously does not specify the semantics of any of these extensions, and the
				1483	definition of i-c-e does not include them. However, these extensions are often
				1484	used in real code, and we have to have a way to reason about them.</p>
				1485
				1486	<p>Finally, this is not just a problem for semantic analysis. The code
				1487	generator and other clients have to be able to fold constants (e.g. to
				1488	initialize global variables) and has to handle a superset of what C99 allows.
				1489	Further, these clients can benefit from extended information. For example, we
				1490	know that "foo()\|\|1" always evaluates to true, but we can't replace the
				1491	expression with true because it has side effects.</p>
				1492
				1493	<!-- ======================= -->
				1494	<h4>Implementation Approach</h4>
				1495	<!-- ======================= -->
				1496
				1497	<p>After trying several different approaches, we've finally converged on a
				1498	design (Note, at the time of this writing, not all of this has been implemented,
				1499	consider this a design goal!). Our basic approach is to define a single
				1500	recursive method evaluation method (<tt>Expr::Evaluate</tt>), which is
				1501	implemented in <tt>AST/ExprConstant.cpp</tt>. Given an expression with 'scalar'
				1502	type (integer, fp, complex, or pointer) this method returns the following
				1503	information:</p>
				1504
				1505	<ul>
				1506	<li>Whether the expression is an integer constant expression, a general
				1507	constant that was folded but has no side effects, a general constant that
				1508	was folded but that does have side effects, or an uncomputable/unfoldable
				1509	value.
				1510	</li>
				1511	<li>If the expression was computable in any way, this method returns the APValue
				1512	for the result of the expression.</li>
				1513	<li>If the expression is not evaluatable at all, this method returns
				1514	information on one of the problems with the expression. This includes a
				1515	SourceLocation for where the problem is, and a diagnostic ID that explains
				1516	the problem. The diagnostic should be have ERROR type.</li>
				1517	<li>If the expression is not an integer constant expression, this method returns
				1518	information on one of the problems with the expression. This includes a
				1519	SourceLocation for where the problem is, and a diagnostic ID that explains
				1520	the problem. The diagnostic should be have EXTENSION type.</li>
				1521	</ul>
				1522
				1523	<p>This information gives various clients the flexibility that they want, and we
				1524	will eventually have some helper methods for various extensions. For example,
				1525	Sema should have a <tt>Sema::VerifyIntegerConstantExpression</tt> method, which
				1526	calls Evaluate on the expression. If the expression is not foldable, the error
				1527	is emitted, and it would return true. If the expression is not an i-c-e, the
				1528	EXTENSION diagnostic is emitted. Finally it would return false to indicate that
				1529	the AST is ok.</p>
				1530
				1531	<p>Other clients can use the information in other ways, for example, codegen can
				1532	just use expressions that are foldable in any way.</p>
				1533
				1534	<!-- ========== -->
				1535	<h4>Extensions</h4>
				1536	<!-- ========== -->
				1537
Chris Lattner	552de0a	2008-11-23 08:16:56 +0000	[diff] [blame]	1538	<p>This section describes how some of the various extensions Clang supports
Chris Lattner	7bad199	2008-11-16 21:48:07 +0000	[diff] [blame]	1539	interacts with constant evaluation:</p>
				1540
				1541	<ul>
				1542	<li><b><tt>__extension__</tt></b>: The expression form of this extension causes
				1543	any evaluatable subexpression to be accepted as an integer constant
				1544	expression.</li>
				1545	<li><b><tt>__builtin_constant_p</tt></b>: This returns true (as a integer
Chris Lattner	28daa53	2008-12-12 06:55:44 +0000	[diff] [blame]	1546	constant expression) if the operand is any evaluatable constant. As a
				1547	special case, if <tt>__builtin_constant_p</tt> is the (potentially
				1548	parenthesized) condition of a conditional operator expression ("?:"), only
Chris Lattner	42b83dd	2008-12-12 18:00:51 +0000	[diff] [blame]	1549	the true side of the conditional operator is considered, and it is evaluated
				1550	with full constant folding.</li>
Chris Lattner	7bad199	2008-11-16 21:48:07 +0000	[diff] [blame]	1551	<li><b><tt>__builtin_choose_expr</tt></b>: The condition is required to be an
				1552	integer constant expression, but we accept any constant as an "extension of
				1553	an extension". This only evaluates one operand depending on which way the
				1554	condition evaluates.</li>
				1555	<li><b><tt>__builtin_classify_type</tt></b>: This always returns an integer
				1556	constant expression.</li>
				1557	<li><b><tt>__builtin_inf,nan,..</tt></b>: These are treated just like a
				1558	floating-point literal.</li>
				1559	<li><b><tt>__builtin_abs,copysign,..</tt></b>: These are constant folded as
				1560	general constant expressions.</li>
				1561	</ul>
				1562
				1563
				1564
				1565
Ted Kremenek	17a295d	2008-06-11 06:19:49 +0000	[diff] [blame]	1566	</div>
				1567	</body>
Douglas Gregor	2e1cd42	2008-11-17 14:58:09 +0000	[diff] [blame]	1568	</html>