Gitiles
Code Review Sign In
gerrit-public.fairphone.software / toolchain / llvm-project / 611a1883e594e8b95f6ef8dd313c374085277d12 / . / llvm / docs / CodingStandards.html
blob: 8d460009457406434ff473f99cd4787b344b6494 [file] [log] [blame]
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
3<html>
4<head>
5 <link rel="stylesheet" href="llvm.css" type="text/css">
Misha Brukman6c2522a2008-12-10 23:07:02 +0000[diff] [blame]6 <title>LLVM Coding Standards</title>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]7</head>
8<body>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]9
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]10<div class="doc_title">
Misha Brukman6c2522a2008-12-10 23:07:02 +0000[diff] [blame]11 LLVM Coding Standards
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]12</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]13
14<ol>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]15 <li><a href="#introduction">Introduction</a></li>
Chris Lattner85014f42001-07-23 20:40:41 +0000[diff] [blame]16 <li><a href="#mechanicalissues">Mechanical Source Issues</a>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]17 <ol>
18 <li><a href="#sourceformating">Source Code Formatting</a>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]19 <ol>
20 <li><a href="#scf_commenting">Commenting</a></li>
21 <li><a href="#scf_commentformat">Comment Formatting</a></li>
Misha Brukman03f87d52004-10-26 16:18:43 +0000[diff] [blame]22 <li><a href="#scf_includes"><tt>#include</tt> Style</a></li>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]23 <li><a href="#scf_codewidth">Source Code Width</a></li>
24 <li><a href="#scf_spacestabs">Use Spaces Instead of Tabs</a></li>
25 <li><a href="#scf_indentation">Indent Code Consistently</a></li>
26 </ol></li>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]27 <li><a href="#compilerissues">Compiler Issues</a>
28 <ol>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]29 <li><a href="#ci_warningerrors">Treat Compiler Warnings Like
30 Errors</a></li>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]31 <li><a href="#ci_portable_code">Write Portable Code</a></li>
Reid Spencerbf6439f2004-09-23 16:03:48 +0000[diff] [blame]32 <li><a href="#ci_class_struct">Use of class/struct Keywords</a></li>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]33 </ol></li>
34 </ol></li>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]35 <li><a href="#styleissues">Style Issues</a>
36 <ol>
37 <li><a href="#macro">The High Level Issues</a>
38 <ol>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]39 <li><a href="#hl_module">A Public Header File <b>is</b> a
40 Module</a></li>
41 <li><a href="#hl_dontinclude">#include as Little as Possible</a></li>
42 <li><a href="#hl_privateheaders">Keep "internal" Headers
43 Private</a></li>
Chris Lattner6720d7e2009-07-22 05:40:54 +0000[diff] [blame]44 <li><a href="#hl_earlyexit">Use Early Exits and 'continue' to Simplify
45 Code</a></li>
46 <li><a href="#hl_predicateloops">Turn Predicate Loops into Predicate
47 Functions</a></li>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]48 </ol></li>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]49 <li><a href="#micro">The Low Level Issues</a>
50 <ol>
Chris Lattner85ea83e2006-07-27 04:24:14 +0000[diff] [blame]51 <li><a href="#ll_assert">Assert Liberally</a></li>
52 <li><a href="#ll_ns_std">Do not use 'using namespace std'</a></li>
Bill Wendling2b52dc12006-12-09 01:20:34 +0000[diff] [blame]53 <li><a href="#ll_virtual_anch">Provide a virtual method anchor for
Bill Wendlingb0629482006-12-09 01:35:43 +0000[diff] [blame]54 classes in headers</a></li>
Chris Lattner56c10942009-06-30 06:13:23 +0000[diff] [blame]55 <li><a href="#ll_end">Don't evaluate end() every time through a
56 loop</a></li>
Chris Lattner6720d7e2009-07-22 05:40:54 +0000[diff] [blame]57 <li><a href="#ll_iostream"><tt>#include &lt;iostream&gt;</tt> is
58 <em>forbidden</em></a></li>
Chris Lattner85ea83e2006-07-27 04:24:14 +0000[diff] [blame]59 <li><a href="#ll_avoidendl">Avoid <tt>std::endl</tt></a></li>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]60 </ol></li>
Chris Lattner6720d7e2009-07-22 05:40:54 +0000[diff] [blame]61
62 <li><a href="#nano">Microscopic Details</a>
63 <ol>
64 <li><a href="#micro_spaceparen">Spaces Before Parentheses</a></li>
65 <li><a href="#micro_preincrement">Prefer Preincrement</a></li>
66 <li><a href="#micro_namespaceindent">Namespace Indentation</a></li>
67 <li><a href="#micro_anonns">Anonymous Namespaces</a></li>
68 </ol></li>
69
70
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]71 </ol></li>
72 <li><a href="#seealso">See Also</a></li>
73</ol>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]74
Chris Lattner1dab1922004-05-23 21:05:07 +0000[diff] [blame]75<div class="doc_author">
Chris Lattner9567fff2009-07-22 05:43:01 +0000[diff] [blame]76 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a></p>
Chris Lattner1dab1922004-05-23 21:05:07 +0000[diff] [blame]77</div>
78
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]79
80<!-- *********************************************************************** -->
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]81<div class="doc_section">
82 <a name="introduction">Introduction</a>
83</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]84<!-- *********************************************************************** -->
85
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]86<div class="doc_text">
87
88<p>This document attempts to describe a few coding standards that are being used
89in the LLVM source tree. Although no coding standards should be regarded as
Misha Brukman0cedb1f2003-10-06 19:26:00 +0000[diff] [blame]90absolute requirements to be followed in all instances, coding standards can be
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]91useful.</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]92
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]93<p>This document intentionally does not prescribe fixed standards for religious
Misha Brukman0cedb1f2003-10-06 19:26:00 +0000[diff] [blame]94issues such as brace placement and space usage. For issues like this, follow
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]95the golden rule:</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]96
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]97<blockquote>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]98
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]99<p><b><a name="goldenrule">If you are adding a significant body of source to a
100project, feel free to use whatever style you are most comfortable with. If you
101are extending, enhancing, or bug fixing already implemented code, use the style
102that is already being used so that the source is uniform and easy to
103follow.</a></b></p>
104
105</blockquote>
106
107<p>The ultimate goal of these guidelines is the increase readability and
Misha Brukman0cedb1f2003-10-06 19:26:00 +0000[diff] [blame]108maintainability of our common source base. If you have suggestions for topics to
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]109be included, please mail them to <a
110href="mailto:sabre@nondot.org">Chris</a>.</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]111
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]112</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]113
114<!-- *********************************************************************** -->
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]115<div class="doc_section">
116 <a name="mechanicalissues">Mechanical Source Issues</a>
117</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]118<!-- *********************************************************************** -->
119
120<!-- ======================================================================= -->
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]121<div class="doc_subsection">
122 <a name="sourceformating">Source Code Formatting</a>
123</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]124
125<!-- _______________________________________________________________________ -->
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]126<div class="doc_subsubsection">
127 <a name="scf_commenting">Commenting</a>
128</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]129
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]130<div class="doc_text">
131
132<p>Comments are one critical part of readability and maintainability. Everyone
Chris Lattnerbbe228b2009-07-12 00:10:24 +0000[diff] [blame]133knows they should comment, so should you. When writing comments, write them as
134English prose, which means they should use proper capitalization, punctuation,
135etc. Although we all should probably
Misha Brukman0cedb1f2003-10-06 19:26:00 +0000[diff] [blame]136comment our code more than we do, there are a few very critical places that
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]137documentation is very useful:</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]138
Chris Lattner1dab1922004-05-23 21:05:07 +0000[diff] [blame]139<b>File Headers</b>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]140
Reid Spencerc7f87f22007-07-09 08:04:31 +0000[diff] [blame]141<p>Every source file should have a header on it that describes the basic
142purpose of the file. If a file does not have a header, it should not be
143checked into Subversion. Most source trees will probably have a standard
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]144file header format. The standard format for the LLVM source tree looks like
145this:</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]146
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]147<div class="doc_code">
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]148<pre>
Chris Lattnere6f4e072003-10-13 14:58:11 +0000[diff] [blame]149//===-- llvm/Instruction.h - Instruction class definition -------*- C++ -*-===//
Misha Brukman02805a62009-01-02 16:58:42 +0000[diff] [blame]150//
Chris Lattner1dab1922004-05-23 21:05:07 +0000[diff] [blame]151// The LLVM Compiler Infrastructure
152//
Chris Lattnerebf56662007-12-29 19:56:08 +0000[diff] [blame]153// This file is distributed under the University of Illinois Open Source
154// License. See LICENSE.TXT for details.
Misha Brukman02805a62009-01-02 16:58:42 +0000[diff] [blame]155//
Chris Lattner1dab1922004-05-23 21:05:07 +0000[diff] [blame]156//===----------------------------------------------------------------------===//
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]157//
158// This file contains the declaration of the Instruction class, which is the
159// base class for all of the VM instructions.
160//
161//===----------------------------------------------------------------------===//
162</pre>
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]163</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]164
Chris Lattnerebf56662007-12-29 19:56:08 +0000[diff] [blame]165<p>A few things to note about this particular format: The "<tt>-*- C++
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]166-*-</tt>" string on the first line is there to tell Emacs that the source file
Misha Brukman0d640e82004-07-28 22:37:57 +0000[diff] [blame]167is a C++ file, not a C file (Emacs assumes .h files are C files by default).
168Note that this tag is not necessary in .cpp files. The name of the file is also
169on the first line, along with a very short description of the purpose of the
170file. This is important when printing out code and flipping though lots of
171pages.</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]172
Chris Lattnerebf56662007-12-29 19:56:08 +0000[diff] [blame]173<p>The next section in the file is a concise note that defines the license
174that the file is released under. This makes it perfectly clear what terms the
175source code can be distributed under and should not be modified in any way.</p>
Chris Lattner1dab1922004-05-23 21:05:07 +0000[diff] [blame]176
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]177<p>The main body of the description does not have to be very long in most cases.
Misha Brukman0cedb1f2003-10-06 19:26:00 +0000[diff] [blame]178Here it's only two lines. If an algorithm is being implemented or something
179tricky is going on, a reference to the paper where it is published should be
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]180included, as well as any notes or "gotchas" in the code to watch out for.</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]181
Chris Lattner1dab1922004-05-23 21:05:07 +0000[diff] [blame]182<b>Class overviews</b>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]183
Jim Laskey88b5e792006-07-31 20:18:49 +0000[diff] [blame]184<p>Classes are one fundamental part of a good object oriented design. As such,
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]185a class definition should have a comment block that explains what the class is
Misha Brukman0cedb1f2003-10-06 19:26:00 +0000[diff] [blame]186used for... if it's not obvious. If it's so completely obvious your grandma
187could figure it out, it's probably safe to leave it out. Naming classes
Misha Brukman6d1686c2004-12-04 00:32:12 +0000[diff] [blame]188something sane goes a long ways towards avoiding writing documentation.</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]189
190
Chris Lattner1dab1922004-05-23 21:05:07 +0000[diff] [blame]191<b>Method information</b>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]192
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]193<p>Methods defined in a class (as well as any global functions) should also be
Misha Brukman0cedb1f2003-10-06 19:26:00 +0000[diff] [blame]194documented properly. A quick note about what it does any a description of the
195borderline behaviour is all that is necessary here (unless something
196particularly tricky or insideous is going on). The hope is that people can
197figure out how to use your interfaces without reading the code itself... that is
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]198the goal metric.</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]199
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]200<p>Good things to talk about here are what happens when something unexpected
201happens: does the method return null? Abort? Format your hard disk?</p>
202
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]203</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]204
205<!-- _______________________________________________________________________ -->
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]206<div class="doc_subsubsection">
207 <a name="scf_commentformat">Comment Formatting</a>
208</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]209
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]210<div class="doc_text">
211
212<p>In general, prefer C++ style (<tt>//</tt>) comments. They take less space,
Misha Brukman0cedb1f2003-10-06 19:26:00 +0000[diff] [blame]213require less typing, don't have nesting problems, etc. There are a few cases
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]214when it is useful to use C style (<tt>/* */</tt>) comments however:</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]215
216<ol>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]217 <li>When writing a C code: Obviously if you are writing C code, use C style
Misha Brukman6d1686c2004-12-04 00:32:12 +0000[diff] [blame]218 comments.</li>
Misha Brukman03f87d52004-10-26 16:18:43 +0000[diff] [blame]219 <li>When writing a header file that may be <tt>#include</tt>d by a C source
220 file.</li>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]221 <li>When writing a source file that is used by a tool that only accepts C
222 style comments.</li>
223</ol>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]224
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]225<p>To comment out a large block of code, use <tt>#if 0</tt> and <tt>#endif</tt>.
226These nest properly and are better behaved in general than C style comments.</p>
227
228</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]229
Chris Lattner245b5252003-08-07 21:45:47 +0000[diff] [blame]230<!-- _______________________________________________________________________ -->
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]231<div class="doc_subsubsection">
Misha Brukman03f87d52004-10-26 16:18:43 +0000[diff] [blame]232 <a name="scf_includes"><tt>#include</tt> Style</a>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]233</div>
Chris Lattner245b5252003-08-07 21:45:47 +0000[diff] [blame]234
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]235<div class="doc_text">
236
237<p>Immediately after the <a href="#scf_commenting">header file comment</a> (and
Chris Lattner245b5252003-08-07 21:45:47 +0000[diff] [blame]238include guards if working on a header file), the <a
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]239href="#hl_dontinclude">minimal</a> list of <tt>#include</tt>s required by the
240file should be listed. We prefer these <tt>#include</tt>s to be listed in this
241order:</p>
Chris Lattner245b5252003-08-07 21:45:47 +0000[diff] [blame]242
243<ol>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]244 <li><a href="#mmheader">Main Module header</a></li>
245 <li><a href="#hl_privateheaders">Local/Private Headers</a></li>
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]246 <li><tt>llvm/*</tt></li>
247 <li><tt>llvm/Analysis/*</tt></li>
248 <li><tt>llvm/Assembly/*</tt></li>
249 <li><tt>llvm/Bytecode/*</tt></li>
250 <li><tt>llvm/CodeGen/*</tt></li>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]251 <li>...</li>
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]252 <li><tt>Support/*</tt></li>
253 <li><tt>Config/*</tt></li>
254 <li>System <tt>#includes</tt></li>
Chris Lattner245b5252003-08-07 21:45:47 +0000[diff] [blame]255</ol>
256
Nick Lewycky35847802008-11-29 20:13:25 +0000[diff] [blame]257<p>... and each category should be sorted by name.</p>
Chris Lattner245b5252003-08-07 21:45:47 +0000[diff] [blame]258
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]259<p><a name="mmheader">The "Main Module Header"</a> file applies to .cpp file
Misha Brukman03f87d52004-10-26 16:18:43 +0000[diff] [blame]260which implement an interface defined by a .h file. This <tt>#include</tt>
261should always be included <b>first</b> regardless of where it lives on the file
262system. By including a header file first in the .cpp files that implement the
263interfaces, we ensure that the header does not have any hidden dependencies
264which are not explicitly #included in the header, but should be. It is also a
265form of documentation in the .cpp file to indicate where the interfaces it
266implements are defined.</p>
Chris Lattner245b5252003-08-07 21:45:47 +0000[diff] [blame]267
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]268</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]269
270<!-- _______________________________________________________________________ -->
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]271<div class="doc_subsubsection">
272 <a name="scf_codewidth">Source Code Width</a>
273</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]274
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]275<div class="doc_text">
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]276
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]277<p>Write your code to fit within 80 columns of text. This helps those of us who
278like to print out code and look at your code in an xterm without resizing
279it.</p>
280
Chris Lattnerfc12d2e2008-07-08 05:12:37 +0000[diff] [blame]281<p>The longer answer is that there must be some limit to the width of the code
282in order to reasonably allow developers to have multiple files side-by-side in
283windows on a modest display. If you are going to pick a width limit, it is
284somewhat arbitrary but you might as well pick something standard. Going with
28590 columns (for example) instead of 80 columns wouldn't add any significant
286value and would be detrimental to printing out code. Also many other projects
287have standardized on 80 columns, so some people have already configured their
288editors for it (vs something else, like 90 columns).</p>
289
290<p>This is one of many contentious issues in coding standards, but is not up
291for debate.</p>
292
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]293</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]294
295<!-- _______________________________________________________________________ -->
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]296<div class="doc_subsubsection">
297 <a name="scf_spacestabs">Use Spaces Instead of Tabs</a>
298</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]299
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]300<div class="doc_text">
301
302<p>In all cases, prefer spaces to tabs in source files. People have different
Misha Brukman0cedb1f2003-10-06 19:26:00 +0000[diff] [blame]303prefered indentation levels, and different styles of indentation that they
304like... this is fine. What isn't is that different editors/viewers expand tabs
305out to different tab stops. This can cause your code to look completely
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]306unreadable, and it is not worth dealing with.</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]307
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]308<p>As always, follow the <a href="#goldenrule">Golden Rule</a> above: follow the
Misha Brukman0cedb1f2003-10-06 19:26:00 +0000[diff] [blame]309style of existing code if your are modifying and extending it. If you like four
310spaces of indentation, <b>DO NOT</b> do that in the middle of a chunk of code
311with two spaces of indentation. Also, do not reindent a whole source file: it
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]312makes for incredible diffs that are absolutely worthless.</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]313
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]314</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]315
316<!-- _______________________________________________________________________ -->
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]317<div class="doc_subsubsection">
318 <a name="scf_indentation">Indent Code Consistently</a>
319</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]320
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]321<div class="doc_text">
322
323<p>Okay, your first year of programming you were told that indentation is
Misha Brukman0cedb1f2003-10-06 19:26:00 +0000[diff] [blame]324important. If you didn't believe and internalize this then, now is the time.
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]325Just do it.</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]326
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]327</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]328
329
330<!-- ======================================================================= -->
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]331<div class="doc_subsection">
332 <a name="compilerissues">Compiler Issues</a>
333</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]334
335
336<!-- _______________________________________________________________________ -->
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]337<div class="doc_subsubsection">
338 <a name="ci_warningerrors">Treat Compiler Warnings Like Errors</a>
339</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]340
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]341<div class="doc_text">
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]342
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]343<p>If your code has compiler warnings in it, something is wrong: you aren't
344casting values correctly, your have "questionable" constructs in your code, or
345you are doing something legitimately wrong. Compiler warnings can cover up
346legitimate errors in output and make dealing with a translation unit
347difficult.</p>
348
349<p>It is not possible to prevent all warnings from all compilers, nor is it
Misha Brukman0cedb1f2003-10-06 19:26:00 +0000[diff] [blame]350desirable. Instead, pick a standard compiler (like <tt>gcc</tt>) that provides
351a good thorough set of warnings, and stick to them. At least in the case of
352<tt>gcc</tt>, it is possible to work around any spurious errors by changing the
353syntax of the code slightly. For example, an warning that annoys me occurs when
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]354I write code like this:</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]355
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]356<div class="doc_code">
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]357<pre>
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]358if (V = getValue()) {
359 ...
360}
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]361</pre>
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]362</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]363
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]364<p><tt>gcc</tt> will warn me that I probably want to use the <tt>==</tt>
365operator, and that I probably mistyped it. In most cases, I haven't, and I
366really don't want the spurious errors. To fix this particular problem, I
367rewrite the code like this:</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]368
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]369<div class="doc_code">
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]370<pre>
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]371if ((V = getValue())) {
372 ...
373}
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]374</pre>
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]375</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]376
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]377<p>...which shuts <tt>gcc</tt> up. Any <tt>gcc</tt> warning that annoys you can
378be fixed by massaging the code appropriately.</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]379
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]380<p>These are the <tt>gcc</tt> warnings that I prefer to enable: <tt>-Wall
381-Winline -W -Wwrite-strings -Wno-unused</tt></p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]382
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]383</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]384
385<!-- _______________________________________________________________________ -->
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]386<div class="doc_subsubsection">
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]387 <a name="ci_portable_code">Write Portable Code</a>
388</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]389
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]390<div class="doc_text">
391
392<p>In almost all cases, it is possible and within reason to write completely
Misha Brukmanc3e78932003-07-28 21:57:18 +0000[diff] [blame]393portable code. If there are cases where it isn't possible to write portable
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]394code, isolate it behind a well defined (and well documented) interface.</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]395
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]396<p>In practice, this means that you shouldn't assume much about the host
397compiler, including its support for "high tech" features like partial
Chris Lattner30bc9722009-03-23 04:52:53 +0000[diff] [blame]398specialization of templates. If these features are used, they should only be
399an implementation detail of a library which has a simple exposed API.</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]400
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]401</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]402
Reid Spencerbf6439f2004-09-23 16:03:48 +0000[diff] [blame]403<!-- _______________________________________________________________________ -->
404<div class="doc_subsubsection">
405<a name="ci_class_struct">Use of <tt>class</tt> and <tt>struct</tt> Keywords</a>
406</div>
407<div class="doc_text">
Misha Brukmanf2499132004-10-26 15:45:13 +0000[diff] [blame]408
409<p>In C++, the <tt>class</tt> and <tt>struct</tt> keywords can be used almost
410interchangeably. The only difference is when they are used to declare a class:
411<tt>class</tt> makes all members private by default while <tt>struct</tt> makes
412all members public by default.</p>
413
414<p>Unfortunately, not all compilers follow the rules and some will generate
415different symbols based on whether <tt>class</tt> or <tt>struct</tt> was used to
416declare the symbol. This can lead to problems at link time.</p>
417
418<p>So, the rule for LLVM is to always use the <tt>class</tt> keyword, unless
419<b>all</b> members are public, in which case <tt>struct</tt> is allowed.</p>
420
Reid Spencerbf6439f2004-09-23 16:03:48 +0000[diff] [blame]421</div>
422
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]423<!-- *********************************************************************** -->
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]424<div class="doc_section">
425 <a name="styleissues">Style Issues</a>
426</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]427<!-- *********************************************************************** -->
428
429
430<!-- ======================================================================= -->
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]431<div class="doc_subsection">
432 <a name="macro">The High Level Issues</a>
433</div>
Chris Lattner6720d7e2009-07-22 05:40:54 +0000[diff] [blame]434<!-- ======================================================================= -->
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]435
436
437<!-- _______________________________________________________________________ -->
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]438<div class="doc_subsubsection">
439 <a name="hl_module">A Public Header File <b>is</b> a Module</a>
440</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]441
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]442<div class="doc_text">
443
444<p>C++ doesn't do too well in the modularity department. There is no real
Misha Brukmanc3e78932003-07-28 21:57:18 +0000[diff] [blame]445encapsulation or data hiding (unless you use expensive protocol classes), but it
446is what we have to work with. When you write a public header file (in the LLVM
447source tree, they live in the top level "include" directory), you are defining a
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]448module of functionality.</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]449
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]450<p>Ideally, modules should be completely independent of each other, and their
Misha Brukmanc3e78932003-07-28 21:57:18 +0000[diff] [blame]451header files should only include the absolute minimum number of headers
452possible. A module is not just a class, a function, or a namespace: <a
453href="http://www.cuj.com/articles/2000/0002/0002c/0002c.htm">it's a collection
454of these</a> that defines an interface. This interface may be several
455functions, classes or data structures, but the important issue is how they work
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]456together.</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]457
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]458<p>In general, a module should be implemented with one or more <tt>.cpp</tt>
459files. Each of these <tt>.cpp</tt> files should include the header that defines
460their interface first. This ensure that all of the dependences of the module
461header have been properly added to the module header itself, and are not
462implicit. System headers should be included after user headers for a
463translation unit.</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]464
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]465</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]466
467<!-- _______________________________________________________________________ -->
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]468<div class="doc_subsubsection">
Misha Brukman03f87d52004-10-26 16:18:43 +0000[diff] [blame]469 <a name="hl_dontinclude"><tt>#include</tt> as Little as Possible</a>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]470</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]471
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]472<div class="doc_text">
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]473
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]474<p><tt>#include</tt> hurts compile time performance. Don't do it unless you
475have to, especially in header files.</p>
476
477<p>But wait, sometimes you need to have the definition of a class to use it, or
Misha Brukman03f87d52004-10-26 16:18:43 +0000[diff] [blame]478to inherit from it. In these cases go ahead and <tt>#include</tt> that header
479file. Be aware however that there are many cases where you don't need to have
480the full definition of a class. If you are using a pointer or reference to a
481class, you don't need the header file. If you are simply returning a class
482instance from a prototyped function or method, you don't need it. In fact, for
483most cases, you simply don't need the definition of a class... and not
484<tt>#include</tt>'ing speeds up compilation.</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]485
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]486<p>It is easy to try to go too overboard on this recommendation, however. You
Chris Lattner756f83f2007-02-10 18:35:31 +0000[diff] [blame]487<b>must</b> include all of the header files that you are using -- you can
488include them either directly
Misha Brukmanc3e78932003-07-28 21:57:18 +0000[diff] [blame]489or indirectly (through another header file). To make sure that you don't
490accidently forget to include a header file in your module header, make sure to
491include your module header <b>first</b> in the implementation file (as mentioned
492above). This way there won't be any hidden dependencies that you'll find out
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]493about later...</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]494
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]495</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]496
497<!-- _______________________________________________________________________ -->
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]498<div class="doc_subsubsection">
499 <a name="hl_privateheaders">Keep "internal" Headers Private</a>
500</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]501
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]502<div class="doc_text">
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]503
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]504<p>Many modules have a complex implementation that causes them to use more than
505one implementation (<tt>.cpp</tt>) file. It is often tempting to put the
506internal communication interface (helper classes, extra functions, etc) in the
Misha Brukman6d1686c2004-12-04 00:32:12 +0000[diff] [blame]507public module header file. Don't do this.</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]508
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]509<p>If you really need to do something like this, put a private header file in
510the same directory as the source files, and include it locally. This ensures
511that your private interface remains private and undisturbed by outsiders.</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]512
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]513<p>Note however, that it's okay to put extra implementation methods a public
514class itself... just make them private (or protected), and all is well.</p>
515
516</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]517
Bill Wendling2b52dc12006-12-09 01:20:34 +0000[diff] [blame]518<!-- _______________________________________________________________________ -->
519<div class="doc_subsubsection">
Chris Lattner6720d7e2009-07-22 05:40:54 +0000[diff] [blame]520 <a name="hl_earlyexit">Use Early Exits and 'continue' to Simplify Code</a>
Bill Wendling2b52dc12006-12-09 01:20:34 +0000[diff] [blame]521</div>
522
523<div class="doc_text">
524
Chris Lattner6720d7e2009-07-22 05:40:54 +0000[diff] [blame]525<p>When reading code, keep in mind how much state and how many previous
526decisions have to be remembered by the reader to understand a block of code.
527Aim to reduce indentation where possible when it doesn't make it more difficult
528to understand the code. One great way to do this is by making use of early
529exits and the 'continue' keyword in long loops. As an example of using an early
530exit from a function, consider this "bad" code:</p>
Bill Wendling2b52dc12006-12-09 01:20:34 +0000[diff] [blame]531
Chris Lattner6720d7e2009-07-22 05:40:54 +0000[diff] [blame]532<div class="doc_code">
533<pre>
534Value *DoSomething(Instruction *I) {
535 if (!isa&lt;TerminatorInst&gt;(I) &amp;&amp;
536 I-&gt;hasOneUse() &amp;&amp; SomeOtherThing(I)) {
537 ... some long code ....
538 }
539
540 return 0;
541}
542</pre>
543</div>
Bill Wendling2b52dc12006-12-09 01:20:34 +0000[diff] [blame]544
Chris Lattner6720d7e2009-07-22 05:40:54 +0000[diff] [blame]545<p>This code has several problems if the body of the 'if' is large. When you're
546looking at the top of the function, it isn't immediately clear that this
547<em>only</em> does interesting things with non-terminator instructions, and only
548applies to things with the other predicates. Second, it is relatively difficult
549to describe (in comments) why these predicates are important because the if
550statement makes it difficult to lay out the comments. Third, when you're deep
551within the body of the code, it is indented an extra level. Finally, when
552reading the top of the function, it isn't clear what the result is if the
553predicate isn't true, you have to read to the end of the function to know that
554it returns null.</p>
Matthijs Kooijmanca4ffaa2008-07-30 12:14:10 +0000[diff] [blame]555
Chris Lattner6720d7e2009-07-22 05:40:54 +0000[diff] [blame]556<p>It is much preferred to format the code like this:</p>
557
558<div class="doc_code">
559<pre>
560Value *DoSomething(Instruction *I) {
561 // Terminators never need 'something' done to them because, ...
562 if (isa&lt;TerminatorInst&gt;(I))
563 return 0;
564
565 // We conservatively avoid transforming instructions with multiple uses
566 // because goats like cheese.
567 if (!I-&gt;hasOneUse())
568 return 0;
569
570 // This is really just here for example.
571 if (!SomeOtherThing(I))
572 return 0;
573
574 ... some long code ....
575}
576</pre>
577</div>
578
579<p>This fixes these problems. A similar problem frequently happens in for
580loops. A silly example is something like this:</p>
581
582<div class="doc_code">
583<pre>
584 for (BasicBlock::iterator II = BB-&gt;begin(), E = BB-&gt;end(); II != E; ++II) {
585 if (BinaryOperator *BO = dyn_cast&lt;BinaryOperator&gt;(II)) {
586 Value *LHS = BO-&gt;getOperand(0);
587 Value *RHS = BO-&gt;getOperand(1);
588 if (LHS != RHS) {
589 ...
590 }
591 }
592 }
593</pre>
594</div>
595
596<p>When you have very very small loops, this sort of structure is fine, but if
597it exceeds more than 10-15 lines, it becomes difficult for people to read and
598understand at a glance.
599The problem with this sort of code is that it gets very nested very quickly,
600meaning that the reader of the code has to keep a lot of context in their brain
601to remember what is going immediately on in the loop, because they don't know
602if/when the if conditions will have elses etc. It is strongly preferred to
603structure the loop like this:</p>
604
605<div class="doc_code">
606<pre>
607 for (BasicBlock::iterator II = BB-&gt;begin(), E = BB-&gt;end(); II != E; ++II) {
608 BinaryOperator *BO = dyn_cast&lt;BinaryOperator&gt;(II);
609 if (!BO) continue;
610
611 Value *LHS = BO-&gt;getOperand(0);
612 Value *RHS = BO-&gt;getOperand(1);
613 if (LHS == RHS) continue;
614 }
615</pre>
616</div>
617
618<p>This has all the benefits of using early exits from functions: it reduces
619nesting of the loop, it makes it easier to describe why the conditions are true,
620and it makes it obvious to the reader that there is no "else" coming up that
621they have to push context into their brain for. If a loop is large, this can
622be a big understandability win.</p>
623
624</div>
625
626
627<!-- _______________________________________________________________________ -->
628<div class="doc_subsubsection">
629 <a name="hl_predicateloops">Turn Predicate Loops into Predicate Functions</a>
630</div>
631
632<div class="doc_text">
633
634<p>It is very common to write inline functions that just compute a boolean
635 value. There are a number of ways that people commonly write these, but an
636 example of this sort of thing is:</p>
637
638<div class="doc_code">
639<pre>
640 <b>bool FoundFoo = false;</b>
641 for (unsigned i = 0, e = BarList.size(); i != e; ++i)
642 if (BarList[i]-&gt;isFoo()) {
643 <b>FoundFoo = true;</b>
644 break;
645 }
646
647 <b>if (FoundFoo) {</b>
648 ...
649 }
650</pre>
651</div>
652
653<p>This sort of code is awkward to write, and is almost always a bad sign.
654Instead of this sort of loop, we strongly prefer to use a predicate function
655(which may be <a href="#micro_anonns">static</a>) that uses
656<a href="#hl_earlyexit">early exits</a> to compute the predicate. Code like
657this would be preferred:
658</p>
659
660
661<div class="doc_code">
662<pre>
663/// ListContainsFoo - Return true if the specified list has an element that is
664/// a foo.
665static bool ListContainsFoo(const std::vector&lt;Bar*&gt; &amp;List) {
666 for (unsigned i = 0, e = List.size(); i != e; ++i)
667 if (List[i]-&gt;isFoo())
668 return true;
669 return false;
670}
671...
672
673 <b>if (ListContainsFoo(BarList)) {</b>
674 ...
675 }
676</pre>
677</div>
678
679<p>There are many reasons for doing this: it reduces indentation and factors out
680code which can often be shared by other code that checks for the same predicate.
681More importantly, it <em>forces you to pick a name</em> for the function, and
682forces you to write a comment for it. In this silly example, this doesn't add
683much value. However, if the condition is complex, this can make it a lot easier
684for the reader to understand the code that queries for this predicate. Instead
Chris Lattner611a1882009-07-22 16:30:39 +0000[diff] [blame^]685of being faced with the in-line details of how we check to see if the BarList
Chris Lattner6720d7e2009-07-22 05:40:54 +0000[diff] [blame]686contains a foo, we can trust the function name and continue reading with better
687locality.</p>
Bill Wendling2b52dc12006-12-09 01:20:34 +0000[diff] [blame]688
689</div>
690
691
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]692<!-- ======================================================================= -->
Chris Lattner1dab1922004-05-23 21:05:07 +0000[diff] [blame]693<div class="doc_subsection">
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]694 <a name="micro">The Low Level Issues</a>
695</div>
Chris Lattner6720d7e2009-07-22 05:40:54 +0000[diff] [blame]696<!-- ======================================================================= -->
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]697
698
699<!-- _______________________________________________________________________ -->
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]700<div class="doc_subsubsection">
Chris Lattner85ea83e2006-07-27 04:24:14 +0000[diff] [blame]701 <a name="ll_assert">Assert Liberally</a>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]702</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]703
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]704<div class="doc_text">
705
706<p>Use the "<tt>assert</tt>" function to its fullest. Check all of your
Misha Brukmanc3e78932003-07-28 21:57:18 +0000[diff] [blame]707preconditions and assumptions, you never know when a bug (not neccesarily even
708yours) might be caught early by an assertion, which reduces debugging time
709dramatically. The "<tt>&lt;cassert&gt;</tt>" header file is probably already
710included by the header files you are using, so it doesn't cost anything to use
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]711it.</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]712
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]713<p>To further assist with debugging, make sure to put some kind of error message
714in the assertion statement (which is printed if the assertion is tripped). This
Misha Brukmanc3e78932003-07-28 21:57:18 +0000[diff] [blame]715helps the poor debugging make sense of why an assertion is being made and
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]716enforced, and hopefully what to do about it. Here is one complete example:</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]717
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]718<div class="doc_code">
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]719<pre>
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]720inline Value *getOperand(unsigned i) {
721 assert(i &lt; Operands.size() &amp;&amp; "getOperand() out of range!");
722 return Operands[i];
723}
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]724</pre>
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]725</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]726
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]727<p>Here are some examples:</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]728
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]729<div class="doc_code">
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]730<pre>
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]731assert(Ty-&gt;isPointerType() &amp;&amp; "Can't allocate a non pointer type!");
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]732
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]733assert((Opcode == Shl || Opcode == Shr) &amp;&amp; "ShiftInst Opcode invalid!");
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]734
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]735assert(idx &lt; getNumSuccessors() &amp;&amp; "Successor # out of range!");
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]736
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]737assert(V1.getType() == V2.getType() &amp;&amp; "Constant types must be identical!");
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]738
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]739assert(isa&lt;PHINode&gt;(Succ-&gt;front()) &amp;&amp; "Only works on PHId BBs!");
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]740</pre>
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]741</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]742
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]743<p>You get the idea...</p>
744
Nick Lewyckyfb75d422008-05-31 23:54:55 +0000[diff] [blame]745<p>Please be aware when adding assert statements that not all compilers are aware of
746the semantics of the assert. In some places, asserts are used to indicate a piece of
747code that should not be reached. These are typically of the form:</p>
748
749<div class="doc_code">
750<pre>
Dan Gohman8ef44982008-11-24 17:18:39 +0000[diff] [blame]751assert(0 &amp;&amp; "Some helpful error message");
Nick Lewyckyfb75d422008-05-31 23:54:55 +0000[diff] [blame]752</pre>
753</div>
754
755<p>When used in a function that returns a value, they should be followed with a return
756statement and a comment indicating that this line is never reached. This will prevent
757a compiler which is unable to deduce that the assert statement never returns from
758generating a warning.</p>
759
760<div class="doc_code">
761<pre>
Dan Gohman8ef44982008-11-24 17:18:39 +0000[diff] [blame]762assert(0 &amp;&amp; "Some helpful error message");
Nick Lewyckyfb75d422008-05-31 23:54:55 +0000[diff] [blame]763// Not reached
764return 0;
765</pre>
766</div>
767
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]768</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]769
Chris Lattner01e81e62006-01-01 21:59:22 +0000[diff] [blame]770<!-- _______________________________________________________________________ -->
771<div class="doc_subsubsection">
Bill Wendlingb0629482006-12-09 01:35:43 +0000[diff] [blame]772 <a name="ll_ns_std">Do not use '<tt>using namespace std</tt>'</a>
Chris Lattner01e81e62006-01-01 21:59:22 +0000[diff] [blame]773</div>
774
775<div class="doc_text">
776<p>In LLVM, we prefer to explicitly prefix all identifiers from the standard
Bill Wendlingb0629482006-12-09 01:35:43 +0000[diff] [blame]777namespace with an "<tt>std::</tt>" prefix, rather than rely on
778"<tt>using namespace std;</tt>".</p>
Chris Lattner01e81e62006-01-01 21:59:22 +0000[diff] [blame]779
Bill Wendlingb0629482006-12-09 01:35:43 +0000[diff] [blame]780<p> In header files, adding a '<tt>using namespace XXX</tt>' directive pollutes
Misha Brukman586a15c2008-12-11 19:37:04 +0000[diff] [blame]781the namespace of any source file that <tt>#include</tt>s the header. This is
782clearly a bad thing.</p>
Chris Lattner01e81e62006-01-01 21:59:22 +0000[diff] [blame]783
Bill Wendlingb0629482006-12-09 01:35:43 +0000[diff] [blame]784<p>In implementation files (e.g. .cpp files), the rule is more of a stylistic
785rule, but is still important. Basically, using explicit namespace prefixes
786makes the code <b>clearer</b>, because it is immediately obvious what facilities
787are being used and where they are coming from, and <b>more portable</b>, because
788namespace clashes cannot occur between LLVM code and other namespaces. The
789portability rule is important because different standard library implementations
790expose different symbols (potentially ones they shouldn't), and future revisions
791to the C++ standard will add more symbols to the <tt>std</tt> namespace. As
792such, we never use '<tt>using namespace std;</tt>' in LLVM.</p>
Chris Lattner01e81e62006-01-01 21:59:22 +0000[diff] [blame]793
Bill Wendlingb0629482006-12-09 01:35:43 +0000[diff] [blame]794<p>The exception to the general rule (i.e. it's not an exception for
795the <tt>std</tt> namespace) is for implementation files. For example, all of
796the code in the LLVM project implements code that lives in the 'llvm' namespace.
797As such, it is ok, and actually clearer, for the .cpp files to have a '<tt>using
798namespace llvm</tt>' directive at their top, after the <tt>#include</tt>s. The
799general form of this rule is that any .cpp file that implements code in any
800namespace may use that namespace (and its parents'), but should not use any
801others.</p>
Chris Lattner01e81e62006-01-01 21:59:22 +0000[diff] [blame]802
803</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]804
805<!-- _______________________________________________________________________ -->
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]806<div class="doc_subsubsection">
Bill Wendlingb0629482006-12-09 01:35:43 +0000[diff] [blame]807 <a name="ll_virtual_anch">Provide a virtual method anchor for classes
808 in headers</a>
Chris Lattner85ea83e2006-07-27 04:24:14 +0000[diff] [blame]809</div>
810
811<div class="doc_text">
812
813<p>If a class is defined in a header file and has a v-table (either it has
814virtual methods or it derives from classes with virtual methods), it must
815always have at least one out-of-line virtual method in the class. Without
Misha Brukman586a15c2008-12-11 19:37:04 +0000[diff] [blame]816this, the compiler will copy the vtable and RTTI into every <tt>.o</tt> file
817that <tt>#include</tt>s the header, bloating <tt>.o</tt> file sizes and
818increasing link times.</p>
Chris Lattner85ea83e2006-07-27 04:24:14 +0000[diff] [blame]819
820</div>
821
Chris Lattner56c10942009-06-30 06:13:23 +0000[diff] [blame]822<!-- _______________________________________________________________________ -->
823<div class="doc_subsubsection">
824 <a name="ll_end">Don't evaluate end() every time through a loop</a>
825</div>
826
827<div class="doc_text">
828
829<p>Because C++ doesn't have a standard "foreach" loop (though it can be emulated
830with macros and may be coming in C++'0x) we end up writing a lot of loops that
831manually iterate from begin to end on a variety of containers or through other
832data structures. One common mistake is to write a loop in this style:</p>
833
834<div class="doc_code">
835<pre>
836 BasicBlock *BB = ...
837 for (BasicBlock::iterator I = BB->begin(); I != <b>BB->end()</b>; ++I)
838 ... use I ...
839</pre>
840</div>
841
842<p>The problem with this construct is that it evaluates "<tt>BB->end()</tt>"
843every time through the loop. Instead of writing the loop like this, we strongly
844prefer loops to be written so that they evaluate it once before the loop starts.
845A convenient way to do this is like so:</p>
846
847<div class="doc_code">
848<pre>
849 BasicBlock *BB = ...
850 for (BasicBlock::iterator I = BB->begin(), E = <b>BB->end()</b>; I != E; ++I)
851 ... use I ...
852</pre>
853</div>
854
855<p>The observant may quickly point out that these two loops may have different
856semantics: if the container (a basic block in this case) is being mutated, then
857"<tt>BB->end()</tt>" may change its value every time through the loop and the
858second loop may not in fact be correct. If you actually do depend on this
Chris Lattner6ad089c2009-06-30 06:20:03 +0000[diff] [blame]859behavior, please write the loop in the first form and add a comment indicating
Chris Lattner56c10942009-06-30 06:13:23 +0000[diff] [blame]860that you did it intentionally.</p>
861
862<p>Why do we prefer the second form (when correct)? Writing the loop in the
863first form has two problems: First it may be less efficient than evaluating it
864at the start of the loop. In this case, the cost is probably minor: a few extra
865loads every time through the loop. However, if the base expression is more
866complex, then the cost can rise quickly. I've seen loops where the end
867expression was actually something like: "<tt>SomeMap[x]->end()</tt>" and map
Chris Lattner67f2a4ac2009-06-30 06:27:54 +0000[diff] [blame]868lookups really aren't cheap. By writing it in the second form consistently, you
Chris Lattner56c10942009-06-30 06:13:23 +0000[diff] [blame]869eliminate the issue entirely and don't even have to think about it.</p>
870
Chris Lattner6ad089c2009-06-30 06:20:03 +0000[diff] [blame]871<p>The second (even bigger) issue is that writing the loop in the first form
Chris Lattner56c10942009-06-30 06:13:23 +0000[diff] [blame]872hints to the reader that the loop is mutating the container (a fact that a
873comment would handily confirm!). If you write the loop in the second form, it
874is immediately obvious without even looking at the body of the loop that the
875container isn't being modified, which makes it easier to read the code and
876understand what it does.</p>
877
878<p>While the second form of the loop is a few extra keystrokes, we do strongly
879prefer it.</p>
880
881</div>
882
Chris Lattner85ea83e2006-07-27 04:24:14 +0000[diff] [blame]883<!-- _______________________________________________________________________ -->
884<div class="doc_subsubsection">
Chris Lattner6720d7e2009-07-22 05:40:54 +0000[diff] [blame]885 <a name="ll_iostream"><tt>#include &lt;iostream&gt;</tt> is forbidden</a>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]886</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]887
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]888<div class="doc_text">
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]889
Chris Lattner6720d7e2009-07-22 05:40:54 +0000[diff] [blame]890<p>The use of <tt>#include &lt;iostream&gt;</tt> in library files is
891hereby <b><em>forbidden</em></b>. The primary reason for doing this is to
892support clients using LLVM libraries as part of larger systems. In particular,
893we statically link LLVM into some dynamic libraries. Even if LLVM isn't used,
894the static c'tors are run whenever an application start up that uses the dynamic
895library. There are two problems with this:</p>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]896
Chris Lattner6720d7e2009-07-22 05:40:54 +0000[diff] [blame]897<ol>
898 <li>The time to run the static c'tors impacts startup time of
899 applications&mdash;a critical time for GUI apps.</li>
900 <li>The static c'tors cause the app to pull many extra pages of memory off the
901 disk: both the code for the static c'tors in each <tt>.o</tt> file and the
902 small amount of data that gets touched. In addition, touched/dirty pages
903 put more pressure on the VM system on low-memory machines.</li>
904</ol>
905
906<p>Note that using the other stream headers (<tt>&lt;sstream&gt;</tt> for
907example) is allowed normally, it is just <tt>&lt;iostream&gt;</tt> that is
908causing problems.</p>
909
910<p>The preferred replacement for stream functionality is the
911<tt>llvm::raw_ostream</tt> class (for writing to output streams of various
912sorts) and the <tt>llvm::MemoryBuffer</tt> API (for reading in files).</p>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]913
914</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]915
Chris Lattner6720d7e2009-07-22 05:40:54 +0000[diff] [blame]916
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]917<!-- _______________________________________________________________________ -->
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]918<div class="doc_subsubsection">
Chris Lattner85ea83e2006-07-27 04:24:14 +0000[diff] [blame]919 <a name="ll_avoidendl">Avoid <tt>std::endl</tt></a>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]920</div>
Chris Lattner850d4f62002-01-20 19:01:26 +0000[diff] [blame]921
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]922<div class="doc_text">
923
Chris Lattner1dab1922004-05-23 21:05:07 +0000[diff] [blame]924<p>The <tt>std::endl</tt> modifier, when used with iostreams outputs a newline
925to the output stream specified. In addition to doing this, however, it also
926flushes the output stream. In other words, these are equivalent:</p>
Chris Lattner850d4f62002-01-20 19:01:26 +0000[diff] [blame]927
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]928<div class="doc_code">
Chris Lattner850d4f62002-01-20 19:01:26 +0000[diff] [blame]929<pre>
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]930std::cout &lt;&lt; std::endl;
931std::cout &lt;&lt; '\n' &lt;&lt; std::flush;
Chris Lattner850d4f62002-01-20 19:01:26 +0000[diff] [blame]932</pre>
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]933</div>
Chris Lattner850d4f62002-01-20 19:01:26 +0000[diff] [blame]934
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]935<p>Most of the time, you probably have no reason to flush the output stream, so
Misha Brukmanb2246152004-07-28 22:31:54 +0000[diff] [blame]936it's better to use a literal <tt>'\n'</tt>.</p>
Chris Lattner850d4f62002-01-20 19:01:26 +0000[diff] [blame]937
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]938</div>
Chris Lattner850d4f62002-01-20 19:01:26 +0000[diff] [blame]939
Bill Wendling2b52dc12006-12-09 01:20:34 +0000[diff] [blame]940
Chris Lattner6720d7e2009-07-22 05:40:54 +0000[diff] [blame]941<!-- ======================================================================= -->
942<div class="doc_subsection">
943 <a name="nano">Microscopic Details</a>
944</div>
945<!-- ======================================================================= -->
946
947<p>This section describes preferred low-level formatting guidelines along with
948reasoning on why we prefer them.</p>
949
950<!-- _______________________________________________________________________ -->
951<div class="doc_subsubsection">
952 <a name="micro_spaceparen">Spaces Before Parentheses</a>
953</div>
954
955<div class="doc_text">
956
957<p>We prefer to put a space before a parentheses only in control flow
958statements, but not in normal function call expressions and function-like
959macros. For example, this is good:</p>
960
961<div class="doc_code">
962<pre>
963 <b>if (</b>x) ...
964 <b>for (</b>i = 0; i != 100; ++i) ...
965 <b>while (</b>llvm_rocks) ...
966
967 <b>somefunc(</b>42);
968 <b><a href="#ll_assert">assert</a>(</b>3 != 4 &amp;&amp; "laws of math are failing me");
969
970 a = <b>foo(</b>42, 92) + <b>bar(</b>x);
971 </pre>
972</div>
973
974<p>... and this is bad:</p>
975
976<div class="doc_code">
977<pre>
978 <b>if(</b>x) ...
979 <b>for(</b>i = 0; i != 100; ++i) ...
980 <b>while(</b>llvm_rocks) ...
981
982 <b>somefunc (</b>42);
983 <b><a href="#ll_assert">assert</a> (</b>3 != 4 &amp;&amp; "laws of math are failing me");
984
985 a = <b>foo (</b>42, 92) + <b>bar (</b>x);
986</pre>
987</div>
988
989<p>The reason for doing this is not completely arbitrary. This style makes
990 control flow operators stand out more, and makes expressions flow better. The
991 function call operator binds very tightly as a postfix operator. Putting
992 a space after a function name (as in the last example) makes it appear that
993 the code might bind the arguments of the left-hand-side of a binary operator
994 with the argument list of a function and the name of the right side. More
995 specifically, it is easy to misread the "a" example as:</p>
996
997<div class="doc_code">
998<pre>
999 a = foo <b>(</b>(42, 92) + bar<b>)</b> (x);
1000</pre>
1001</div>
1002
1003<p>... when skimming through the code. By avoiding a space in a function, we
1004avoid this misinterpretation.</p>
1005
1006</div>
1007
1008<!-- _______________________________________________________________________ -->
1009<div class="doc_subsubsection">
1010 <a name="micro_preincrement">Prefer Preincrement</a>
1011</div>
1012
1013<div class="doc_text">
1014
1015<p>Hard fast rule: Preincrement (<tt>++X</tt>) may be no slower than
1016postincrement (<tt>X++</tt>) and could very well be a lot faster than it. Use
1017preincrementation whenever possible.</p>
1018
1019<p>The semantics of postincrement include making a copy of the value being
1020incremented, returning it, and then preincrementing the "work value". For
1021primitive types, this isn't a big deal... but for iterators, it can be a huge
1022issue (for example, some iterators contains stack and set objects in them...
1023copying an iterator could invoke the copy ctor's of these as well). In general,
1024get in the habit of always using preincrement, and you won't have a problem.</p>
1025
1026</div>
1027
1028<!-- _______________________________________________________________________ -->
1029<div class="doc_subsubsection">
1030 <a name="micro_namespaceindent">Namespace Indentation</a>
1031</div>
1032
1033<div class="doc_text">
1034
1035<p>
1036In general, we strive to reduce indentation where ever possible. This is useful
1037because we want code to <a href="#scf_codewidth">fit into 80 columns</a> without
1038wrapping horribly, but also because it makes it easier to understand the code.
1039Namespaces are a funny thing: they are often large, and we often desire to put
1040lots of stuff into them (so they can be large). Other times they are tiny,
1041because they just hold an enum or something similar. In order to balance this,
1042we use different approaches for small versus large namespaces.
1043</p>
1044
1045<p>
1046If a namespace definition is small and <em>easily</em> fits on a screen (say,
1047less than 35 lines of code), then you should indent its body. Here's an
1048example:
1049</p>
1050
1051<div class="doc_code">
1052<pre>
1053/// SomeCrazyThing - This namespace contains flags for ...
1054namespace SomeCrazyThing {
1055 enum foo {
1056 /// X - This is the X flag, which is ...
1057 X = 1,
1058
1059 /// Y - This is the Y flag, which is ...
1060 Y = 2,
1061
1062 /// Z - This is the Z flag, which is ...
1063 Z = 4,
1064
1065 /// ALL_FLAGS - This is the union of all flags.
1066 ALL_FLAGS = 7
1067 };
1068}
1069</pre>
1070</div>
1071
1072<p>Since the body is small, indenting adds value because it makes it very clear
1073where the namespace starts and ends, and it is easy to take the whole thing in
1074in one "gulp" when reading the code. If the blob of code in the namespace is
1075larger (as it typically is in a header in the llvm or clang namespaces), do not
1076indent the code, and add a comment indicating what namespace is being closed.
1077For example:</p>
1078
1079<div class="doc_code">
1080<pre>
1081namespace llvm {
1082namespace knowledge {
1083
1084/// Grokable - This class represents things that Smith can have an intimate
1085/// understanding of and contains the data associated with it.
1086class Grokable {
1087...
1088public:
1089 explicit Grokable() { ... }
1090 virtual ~Grokable() = 0;
1091
1092 ...
1093
1094};
1095
1096} // end namespace knowledge
1097} // end namespace llvm
1098</pre>
1099</div>
1100
1101<p>Because the class is large, we don't expect that the reader can easily
1102understand the entire concept in a glance, and the end of the file (where the
1103namespaces end) may be a long ways away from the place they open. As such,
1104indenting the contents of the namespace doesn't add any value, and detracts from
1105the readability of the class. In these cases it is best to <em>not</em> indent
1106the contents of the namespace.</p>
1107
1108</div>
1109
1110<!-- _______________________________________________________________________ -->
1111<div class="doc_subsubsection">
1112 <a name="micro_anonns">Anonymous Namespaces</a>
1113</div>
1114
1115<div class="doc_text">
1116
1117<p>A common topic after talking about namespaces is anonymous namespaces.
1118Anonymous namespaces are a great language feature that tells the C++ compiler
1119that the contents of the namespace are only visible within the current
1120translation unit, allowing more aggressive optimization and eliminating the
1121possibility of symbol name collisions. Anonymous namespaces are to C++ as
1122"static" is to C functions and global variables. While "static" is available
1123in C++, anonymous namespaces are more general: they can make entire classes
1124private to a file.</p>
1125
1126<p>The problem with anonymous namespaces is that they naturally want to
1127encourage indentation of their body, and they reduce locality of reference: if
1128you see a random function definition in a C++ file, it is easy to see if it is
1129marked static, but seeing if it is in an anonymous namespace requires scanning
1130a big chunk of the file.</p>
1131
1132<p>Because of this, we have a simple guideline: make anonymous namespaces as
1133small as possible, and only use them for class declarations. For example, this
1134is good:</p>
1135
1136<div class="doc_code">
1137<pre>
1138<b>namespace {</b>
1139 class StringSort {
1140 ...
1141 public:
1142 StringSort(...)
1143 bool operator&lt;(const char *RHS) const;
1144 };
1145<b>} // end anonymous namespace</b>
1146
1147static void Helper() {
1148 ...
1149}
1150
1151bool StringSort::operator&lt;(const char *RHS) const {
1152 ...
1153}
1154
1155</pre>
1156</div>
1157
1158<p>This is bad:</p>
1159
1160
1161<div class="doc_code">
1162<pre>
1163<b>namespace {</b>
1164class StringSort {
1165...
1166public:
1167 StringSort(...)
1168 bool operator&lt;(const char *RHS) const;
1169};
1170
1171void Helper() {
1172 ...
1173}
1174
1175bool StringSort::operator&lt;(const char *RHS) const {
1176 ...
1177}
1178
1179<b>} // end anonymous namespace</b>
1180
1181</pre>
1182</div>
1183
1184
1185<p>This is bad specifically because if you're looking at "Helper" in the middle
1186of a large C++ file, that you have no immediate way to tell if it is local to
1187the file. When it is marked static explicitly, this is immediately obvious.
1188Also, there is no reason to enclose the definition of "operator&lt;" in the
1189namespace since it was declared there.
1190</p>
1191
1192</div>
1193
1194
1195
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]1196<!-- *********************************************************************** -->
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]1197<div class="doc_section">
1198 <a name="seealso">See Also</a>
1199</div>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]1200<!-- *********************************************************************** -->
1201
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]1202<div class="doc_text">
1203
1204<p>A lot of these comments and recommendations have been culled for other
1205sources. Two particularly important books for our work are:</p>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]1206
1207<ol>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]1208
Chris Lattnerf9f08bf2007-11-09 21:49:08 +0000[diff] [blame]1209<li><a href="http://www.amazon.com/Effective-Specific-Addison-Wesley-Professional-Computing/dp/0321334876">Effective
1210C++</a> by Scott Meyers. Also
Chris Lattner1dab1922004-05-23 21:05:07 +0000[diff] [blame]1211interesting and useful are "More Effective C++" and "Effective STL" by the same
1212author.</li>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]1213
Chris Lattnerf9f08bf2007-11-09 21:49:08 +0000[diff] [blame]1214<li>Large-Scale C++ Software Design by John Lakos</li>
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]1215
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]1216</ol>
1217
1218<p>If you get some free time, and you haven't read them: do so, you might learn
Misha Brukman6d1686c2004-12-04 00:32:12 +0000[diff] [blame]1219something.</p>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]1220
1221</div>
1222
Chris Lattnerac457c42001-07-09 03:27:08 +0000[diff] [blame]1223<!-- *********************************************************************** -->
1224
1225<hr>
Misha Brukmand4f290a2004-05-12 18:37:22 +0000[diff] [blame]1226<address>
1227 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman86242e12008-12-11 17:34:48 +0000[diff] [blame]1228 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Misha Brukmand4f290a2004-05-12 18:37:22 +0000[diff] [blame]1229 <a href="http://validator.w3.org/check/referer"><img
Misha Brukman86242e12008-12-11 17:34:48 +0000[diff] [blame]1230 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]1231
Misha Brukmand4f290a2004-05-12 18:37:22 +0000[diff] [blame]1232 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
Reid Spencerca058542006-03-14 05:39:39 +0000[diff] [blame]1233 <a href="http://llvm.org">LLVM Compiler Infrastructure</a><br>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]1234 Last modified: $Date$
Misha Brukmand4f290a2004-05-12 18:37:22 +0000[diff] [blame]1235</address>
Misha Brukmanf196dbb2003-10-24 17:57:33 +0000[diff] [blame]1236
1237</body>
1238</html>