Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / 2f099ebb73bac294783dd5ce28631d5f72d5438e / . / docs / CodingStandards.html
blob: 2d83f0bf62a660d607ce2e38653c4742dd121446 [file] [log] [blame]
Dan Gohmanf17a25c2007-07-18 16:29:46 +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 Brukman4dbc9d52008-12-10 23:07:02 +0000[diff] [blame]6 <title>LLVM Coding Standards</title>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]7</head>
8<body>
9
10<div class="doc_title">
Misha Brukman4dbc9d52008-12-10 23:07:02 +0000[diff] [blame]11 LLVM Coding Standards
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]12</div>
13
14<ol>
15 <li><a href="#introduction">Introduction</a></li>
16 <li><a href="#mechanicalissues">Mechanical Source Issues</a>
17 <ol>
18 <li><a href="#sourceformating">Source Code Formatting</a>
19 <ol>
20 <li><a href="#scf_commenting">Commenting</a></li>
21 <li><a href="#scf_commentformat">Comment Formatting</a></li>
22 <li><a href="#scf_includes"><tt>#include</tt> Style</a></li>
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>
27 <li><a href="#compilerissues">Compiler Issues</a>
28 <ol>
29 <li><a href="#ci_warningerrors">Treat Compiler Warnings Like
30 Errors</a></li>
31 <li><a href="#ci_portable_code">Write Portable Code</a></li>
32 <li><a href="#ci_class_struct">Use of class/struct Keywords</a></li>
33 </ol></li>
34 </ol></li>
35 <li><a href="#styleissues">Style Issues</a>
36 <ol>
37 <li><a href="#macro">The High Level Issues</a>
38 <ol>
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 Lattnereb1e8c32009-07-22 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>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]48 </ol></li>
49 <li><a href="#micro">The Low Level Issues</a>
50 <ol>
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>
53 <li><a href="#ll_virtual_anch">Provide a virtual method anchor for
54 classes in headers</a></li>
Chris Lattner01b8af32009-06-30 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 Lattnereb1e8c32009-07-22 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>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]59 <li><a href="#ll_avoidendl">Avoid <tt>std::endl</tt></a></li>
Daniel Dunbar2f099eb2009-07-24 23:04:51 +0000[diff] [blame^]60 <li><a href="#ll_raw_ostream">Use <tt>raw_ostream</tt></a</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]61 </ol></li>
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]62
63 <li><a href="#nano">Microscopic Details</a>
64 <ol>
65 <li><a href="#micro_spaceparen">Spaces Before Parentheses</a></li>
66 <li><a href="#micro_preincrement">Prefer Preincrement</a></li>
67 <li><a href="#micro_namespaceindent">Namespace Indentation</a></li>
68 <li><a href="#micro_anonns">Anonymous Namespaces</a></li>
69 </ol></li>
70
71
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]72 </ol></li>
73 <li><a href="#seealso">See Also</a></li>
74</ol>
75
76<div class="doc_author">
Chris Lattnere7346a92009-07-22 05:43:01 +0000[diff] [blame]77 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]78</div>
79
80
81<!-- *********************************************************************** -->
82<div class="doc_section">
83 <a name="introduction">Introduction</a>
84</div>
85<!-- *********************************************************************** -->
86
87<div class="doc_text">
88
89<p>This document attempts to describe a few coding standards that are being used
90in the LLVM source tree. Although no coding standards should be regarded as
91absolute requirements to be followed in all instances, coding standards can be
92useful.</p>
93
94<p>This document intentionally does not prescribe fixed standards for religious
95issues such as brace placement and space usage. For issues like this, follow
96the golden rule:</p>
97
98<blockquote>
99
100<p><b><a name="goldenrule">If you are adding a significant body of source to a
101project, feel free to use whatever style you are most comfortable with. If you
102are extending, enhancing, or bug fixing already implemented code, use the style
103that is already being used so that the source is uniform and easy to
104follow.</a></b></p>
105
106</blockquote>
107
108<p>The ultimate goal of these guidelines is the increase readability and
109maintainability of our common source base. If you have suggestions for topics to
110be included, please mail them to <a
111href="mailto:sabre@nondot.org">Chris</a>.</p>
112
113</div>
114
115<!-- *********************************************************************** -->
116<div class="doc_section">
117 <a name="mechanicalissues">Mechanical Source Issues</a>
118</div>
119<!-- *********************************************************************** -->
120
121<!-- ======================================================================= -->
122<div class="doc_subsection">
123 <a name="sourceformating">Source Code Formatting</a>
124</div>
125
126<!-- _______________________________________________________________________ -->
127<div class="doc_subsubsection">
128 <a name="scf_commenting">Commenting</a>
129</div>
130
131<div class="doc_text">
132
133<p>Comments are one critical part of readability and maintainability. Everyone
Chris Lattner48386692009-07-12 00:10:24 +0000[diff] [blame]134knows they should comment, so should you. When writing comments, write them as
135English prose, which means they should use proper capitalization, punctuation,
136etc. Although we all should probably
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]137comment our code more than we do, there are a few very critical places that
138documentation is very useful:</p>
139
140<b>File Headers</b>
141
142<p>Every source file should have a header on it that describes the basic
143purpose of the file. If a file does not have a header, it should not be
144checked into Subversion. Most source trees will probably have a standard
145file header format. The standard format for the LLVM source tree looks like
146this:</p>
147
148<div class="doc_code">
149<pre>
150//===-- llvm/Instruction.h - Instruction class definition -------*- C++ -*-===//
Misha Brukman0fd5f3e2009-01-02 16:58:42 +0000[diff] [blame]151//
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]152// The LLVM Compiler Infrastructure
153//
Chris Lattnerfb5959f2007-12-29 19:56:08 +0000[diff] [blame]154// This file is distributed under the University of Illinois Open Source
155// License. See LICENSE.TXT for details.
Misha Brukman0fd5f3e2009-01-02 16:58:42 +0000[diff] [blame]156//
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]157//===----------------------------------------------------------------------===//
158//
159// This file contains the declaration of the Instruction class, which is the
160// base class for all of the VM instructions.
161//
162//===----------------------------------------------------------------------===//
163</pre>
164</div>
165
Chris Lattnerfb5959f2007-12-29 19:56:08 +0000[diff] [blame]166<p>A few things to note about this particular format: The "<tt>-*- C++
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]167-*-</tt>" string on the first line is there to tell Emacs that the source file
168is a C++ file, not a C file (Emacs assumes .h files are C files by default).
169Note that this tag is not necessary in .cpp files. The name of the file is also
170on the first line, along with a very short description of the purpose of the
171file. This is important when printing out code and flipping though lots of
172pages.</p>
173
Chris Lattnerfb5959f2007-12-29 19:56:08 +0000[diff] [blame]174<p>The next section in the file is a concise note that defines the license
175that the file is released under. This makes it perfectly clear what terms the
176source code can be distributed under and should not be modified in any way.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]177
178<p>The main body of the description does not have to be very long in most cases.
179Here it's only two lines. If an algorithm is being implemented or something
180tricky is going on, a reference to the paper where it is published should be
181included, as well as any notes or "gotchas" in the code to watch out for.</p>
182
183<b>Class overviews</b>
184
185<p>Classes are one fundamental part of a good object oriented design. As such,
186a class definition should have a comment block that explains what the class is
187used for... if it's not obvious. If it's so completely obvious your grandma
188could figure it out, it's probably safe to leave it out. Naming classes
189something sane goes a long ways towards avoiding writing documentation.</p>
190
191
192<b>Method information</b>
193
194<p>Methods defined in a class (as well as any global functions) should also be
195documented properly. A quick note about what it does any a description of the
196borderline behaviour is all that is necessary here (unless something
197particularly tricky or insideous is going on). The hope is that people can
198figure out how to use your interfaces without reading the code itself... that is
199the goal metric.</p>
200
201<p>Good things to talk about here are what happens when something unexpected
202happens: does the method return null? Abort? Format your hard disk?</p>
203
204</div>
205
206<!-- _______________________________________________________________________ -->
207<div class="doc_subsubsection">
208 <a name="scf_commentformat">Comment Formatting</a>
209</div>
210
211<div class="doc_text">
212
213<p>In general, prefer C++ style (<tt>//</tt>) comments. They take less space,
214require less typing, don't have nesting problems, etc. There are a few cases
215when it is useful to use C style (<tt>/* */</tt>) comments however:</p>
216
217<ol>
218 <li>When writing a C code: Obviously if you are writing C code, use C style
219 comments.</li>
220 <li>When writing a header file that may be <tt>#include</tt>d by a C source
221 file.</li>
222 <li>When writing a source file that is used by a tool that only accepts C
223 style comments.</li>
224</ol>
225
226<p>To comment out a large block of code, use <tt>#if 0</tt> and <tt>#endif</tt>.
227These nest properly and are better behaved in general than C style comments.</p>
228
229</div>
230
231<!-- _______________________________________________________________________ -->
232<div class="doc_subsubsection">
233 <a name="scf_includes"><tt>#include</tt> Style</a>
234</div>
235
236<div class="doc_text">
237
238<p>Immediately after the <a href="#scf_commenting">header file comment</a> (and
239include guards if working on a header file), the <a
240href="#hl_dontinclude">minimal</a> list of <tt>#include</tt>s required by the
241file should be listed. We prefer these <tt>#include</tt>s to be listed in this
242order:</p>
243
244<ol>
245 <li><a href="#mmheader">Main Module header</a></li>
246 <li><a href="#hl_privateheaders">Local/Private Headers</a></li>
247 <li><tt>llvm/*</tt></li>
248 <li><tt>llvm/Analysis/*</tt></li>
249 <li><tt>llvm/Assembly/*</tt></li>
250 <li><tt>llvm/Bytecode/*</tt></li>
251 <li><tt>llvm/CodeGen/*</tt></li>
252 <li>...</li>
253 <li><tt>Support/*</tt></li>
254 <li><tt>Config/*</tt></li>
255 <li>System <tt>#includes</tt></li>
256</ol>
257
Nick Lewycky61819f72008-11-29 20:13:25 +0000[diff] [blame]258<p>... and each category should be sorted by name.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]259
260<p><a name="mmheader">The "Main Module Header"</a> file applies to .cpp file
261which implement an interface defined by a .h file. This <tt>#include</tt>
262should always be included <b>first</b> regardless of where it lives on the file
263system. By including a header file first in the .cpp files that implement the
264interfaces, we ensure that the header does not have any hidden dependencies
265which are not explicitly #included in the header, but should be. It is also a
266form of documentation in the .cpp file to indicate where the interfaces it
267implements are defined.</p>
268
269</div>
270
271<!-- _______________________________________________________________________ -->
272<div class="doc_subsubsection">
273 <a name="scf_codewidth">Source Code Width</a>
274</div>
275
276<div class="doc_text">
277
278<p>Write your code to fit within 80 columns of text. This helps those of us who
279like to print out code and look at your code in an xterm without resizing
280it.</p>
281
Chris Lattnerb5daf702008-07-08 05:12:37 +0000[diff] [blame]282<p>The longer answer is that there must be some limit to the width of the code
283in order to reasonably allow developers to have multiple files side-by-side in
284windows on a modest display. If you are going to pick a width limit, it is
285somewhat arbitrary but you might as well pick something standard. Going with
28690 columns (for example) instead of 80 columns wouldn't add any significant
287value and would be detrimental to printing out code. Also many other projects
288have standardized on 80 columns, so some people have already configured their
289editors for it (vs something else, like 90 columns).</p>
290
291<p>This is one of many contentious issues in coding standards, but is not up
292for debate.</p>
293
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]294</div>
295
296<!-- _______________________________________________________________________ -->
297<div class="doc_subsubsection">
298 <a name="scf_spacestabs">Use Spaces Instead of Tabs</a>
299</div>
300
301<div class="doc_text">
302
303<p>In all cases, prefer spaces to tabs in source files. People have different
304prefered indentation levels, and different styles of indentation that they
305like... this is fine. What isn't is that different editors/viewers expand tabs
306out to different tab stops. This can cause your code to look completely
307unreadable, and it is not worth dealing with.</p>
308
309<p>As always, follow the <a href="#goldenrule">Golden Rule</a> above: follow the
310style of existing code if your are modifying and extending it. If you like four
311spaces of indentation, <b>DO NOT</b> do that in the middle of a chunk of code
312with two spaces of indentation. Also, do not reindent a whole source file: it
313makes for incredible diffs that are absolutely worthless.</p>
314
315</div>
316
317<!-- _______________________________________________________________________ -->
318<div class="doc_subsubsection">
319 <a name="scf_indentation">Indent Code Consistently</a>
320</div>
321
322<div class="doc_text">
323
324<p>Okay, your first year of programming you were told that indentation is
325important. If you didn't believe and internalize this then, now is the time.
326Just do it.</p>
327
328</div>
329
330
331<!-- ======================================================================= -->
332<div class="doc_subsection">
333 <a name="compilerissues">Compiler Issues</a>
334</div>
335
336
337<!-- _______________________________________________________________________ -->
338<div class="doc_subsubsection">
339 <a name="ci_warningerrors">Treat Compiler Warnings Like Errors</a>
340</div>
341
342<div class="doc_text">
343
344<p>If your code has compiler warnings in it, something is wrong: you aren't
345casting values correctly, your have "questionable" constructs in your code, or
346you are doing something legitimately wrong. Compiler warnings can cover up
347legitimate errors in output and make dealing with a translation unit
348difficult.</p>
349
350<p>It is not possible to prevent all warnings from all compilers, nor is it
351desirable. Instead, pick a standard compiler (like <tt>gcc</tt>) that provides
352a good thorough set of warnings, and stick to them. At least in the case of
353<tt>gcc</tt>, it is possible to work around any spurious errors by changing the
354syntax of the code slightly. For example, an warning that annoys me occurs when
355I write code like this:</p>
356
357<div class="doc_code">
358<pre>
359if (V = getValue()) {
360 ...
361}
362</pre>
363</div>
364
365<p><tt>gcc</tt> will warn me that I probably want to use the <tt>==</tt>
366operator, and that I probably mistyped it. In most cases, I haven't, and I
367really don't want the spurious errors. To fix this particular problem, I
368rewrite the code like this:</p>
369
370<div class="doc_code">
371<pre>
372if ((V = getValue())) {
373 ...
374}
375</pre>
376</div>
377
378<p>...which shuts <tt>gcc</tt> up. Any <tt>gcc</tt> warning that annoys you can
379be fixed by massaging the code appropriately.</p>
380
381<p>These are the <tt>gcc</tt> warnings that I prefer to enable: <tt>-Wall
382-Winline -W -Wwrite-strings -Wno-unused</tt></p>
383
384</div>
385
386<!-- _______________________________________________________________________ -->
387<div class="doc_subsubsection">
388 <a name="ci_portable_code">Write Portable Code</a>
389</div>
390
391<div class="doc_text">
392
393<p>In almost all cases, it is possible and within reason to write completely
394portable code. If there are cases where it isn't possible to write portable
395code, isolate it behind a well defined (and well documented) interface.</p>
396
397<p>In practice, this means that you shouldn't assume much about the host
398compiler, including its support for "high tech" features like partial
Chris Lattnerafec0252009-03-23 04:52:53 +0000[diff] [blame]399specialization of templates. If these features are used, they should only be
400an implementation detail of a library which has a simple exposed API.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]401
402</div>
403
404<!-- _______________________________________________________________________ -->
405<div class="doc_subsubsection">
406<a name="ci_class_struct">Use of <tt>class</tt> and <tt>struct</tt> Keywords</a>
407</div>
408<div class="doc_text">
409
410<p>In C++, the <tt>class</tt> and <tt>struct</tt> keywords can be used almost
411interchangeably. The only difference is when they are used to declare a class:
412<tt>class</tt> makes all members private by default while <tt>struct</tt> makes
413all members public by default.</p>
414
415<p>Unfortunately, not all compilers follow the rules and some will generate
416different symbols based on whether <tt>class</tt> or <tt>struct</tt> was used to
417declare the symbol. This can lead to problems at link time.</p>
418
419<p>So, the rule for LLVM is to always use the <tt>class</tt> keyword, unless
420<b>all</b> members are public, in which case <tt>struct</tt> is allowed.</p>
421
422</div>
423
424<!-- *********************************************************************** -->
425<div class="doc_section">
426 <a name="styleissues">Style Issues</a>
427</div>
428<!-- *********************************************************************** -->
429
430
431<!-- ======================================================================= -->
432<div class="doc_subsection">
433 <a name="macro">The High Level Issues</a>
434</div>
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]435<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]436
437
438<!-- _______________________________________________________________________ -->
439<div class="doc_subsubsection">
440 <a name="hl_module">A Public Header File <b>is</b> a Module</a>
441</div>
442
443<div class="doc_text">
444
445<p>C++ doesn't do too well in the modularity department. There is no real
446encapsulation or data hiding (unless you use expensive protocol classes), but it
447is what we have to work with. When you write a public header file (in the LLVM
448source tree, they live in the top level "include" directory), you are defining a
449module of functionality.</p>
450
451<p>Ideally, modules should be completely independent of each other, and their
452header files should only include the absolute minimum number of headers
453possible. A module is not just a class, a function, or a namespace: <a
454href="http://www.cuj.com/articles/2000/0002/0002c/0002c.htm">it's a collection
455of these</a> that defines an interface. This interface may be several
456functions, classes or data structures, but the important issue is how they work
457together.</p>
458
459<p>In general, a module should be implemented with one or more <tt>.cpp</tt>
460files. Each of these <tt>.cpp</tt> files should include the header that defines
461their interface first. This ensure that all of the dependences of the module
462header have been properly added to the module header itself, and are not
463implicit. System headers should be included after user headers for a
464translation unit.</p>
465
466</div>
467
468<!-- _______________________________________________________________________ -->
469<div class="doc_subsubsection">
470 <a name="hl_dontinclude"><tt>#include</tt> as Little as Possible</a>
471</div>
472
473<div class="doc_text">
474
475<p><tt>#include</tt> hurts compile time performance. Don't do it unless you
476have to, especially in header files.</p>
477
478<p>But wait, sometimes you need to have the definition of a class to use it, or
479to inherit from it. In these cases go ahead and <tt>#include</tt> that header
480file. Be aware however that there are many cases where you don't need to have
481the full definition of a class. If you are using a pointer or reference to a
482class, you don't need the header file. If you are simply returning a class
483instance from a prototyped function or method, you don't need it. In fact, for
484most cases, you simply don't need the definition of a class... and not
485<tt>#include</tt>'ing speeds up compilation.</p>
486
487<p>It is easy to try to go too overboard on this recommendation, however. You
488<b>must</b> include all of the header files that you are using -- you can
489include them either directly
490or indirectly (through another header file). To make sure that you don't
491accidently forget to include a header file in your module header, make sure to
492include your module header <b>first</b> in the implementation file (as mentioned
493above). This way there won't be any hidden dependencies that you'll find out
494about later...</p>
495
496</div>
497
498<!-- _______________________________________________________________________ -->
499<div class="doc_subsubsection">
500 <a name="hl_privateheaders">Keep "internal" Headers Private</a>
501</div>
502
503<div class="doc_text">
504
505<p>Many modules have a complex implementation that causes them to use more than
506one implementation (<tt>.cpp</tt>) file. It is often tempting to put the
507internal communication interface (helper classes, extra functions, etc) in the
508public module header file. Don't do this.</p>
509
510<p>If you really need to do something like this, put a private header file in
511the same directory as the source files, and include it locally. This ensures
512that your private interface remains private and undisturbed by outsiders.</p>
513
514<p>Note however, that it's okay to put extra implementation methods a public
515class itself... just make them private (or protected), and all is well.</p>
516
517</div>
518
519<!-- _______________________________________________________________________ -->
520<div class="doc_subsubsection">
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]521 <a name="hl_earlyexit">Use Early Exits and 'continue' to Simplify Code</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]522</div>
523
524<div class="doc_text">
525
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]526<p>When reading code, keep in mind how much state and how many previous
527decisions have to be remembered by the reader to understand a block of code.
528Aim to reduce indentation where possible when it doesn't make it more difficult
529to understand the code. One great way to do this is by making use of early
530exits and the 'continue' keyword in long loops. As an example of using an early
531exit from a function, consider this "bad" code:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]532
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]533<div class="doc_code">
534<pre>
535Value *DoSomething(Instruction *I) {
536 if (!isa&lt;TerminatorInst&gt;(I) &amp;&amp;
537 I-&gt;hasOneUse() &amp;&amp; SomeOtherThing(I)) {
538 ... some long code ....
539 }
540
541 return 0;
542}
543</pre>
544</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]545
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]546<p>This code has several problems if the body of the 'if' is large. When you're
547looking at the top of the function, it isn't immediately clear that this
548<em>only</em> does interesting things with non-terminator instructions, and only
549applies to things with the other predicates. Second, it is relatively difficult
550to describe (in comments) why these predicates are important because the if
551statement makes it difficult to lay out the comments. Third, when you're deep
552within the body of the code, it is indented an extra level. Finally, when
553reading the top of the function, it isn't clear what the result is if the
554predicate isn't true, you have to read to the end of the function to know that
555it returns null.</p>
Matthijs Kooijmanc4c0af22008-07-30 12:14:10 +0000[diff] [blame]556
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]557<p>It is much preferred to format the code like this:</p>
558
559<div class="doc_code">
560<pre>
561Value *DoSomething(Instruction *I) {
562 // Terminators never need 'something' done to them because, ...
563 if (isa&lt;TerminatorInst&gt;(I))
564 return 0;
565
566 // We conservatively avoid transforming instructions with multiple uses
567 // because goats like cheese.
568 if (!I-&gt;hasOneUse())
569 return 0;
570
571 // This is really just here for example.
572 if (!SomeOtherThing(I))
573 return 0;
574
575 ... some long code ....
576}
577</pre>
578</div>
579
580<p>This fixes these problems. A similar problem frequently happens in for
581loops. A silly example is something like this:</p>
582
583<div class="doc_code">
584<pre>
585 for (BasicBlock::iterator II = BB-&gt;begin(), E = BB-&gt;end(); II != E; ++II) {
586 if (BinaryOperator *BO = dyn_cast&lt;BinaryOperator&gt;(II)) {
587 Value *LHS = BO-&gt;getOperand(0);
588 Value *RHS = BO-&gt;getOperand(1);
589 if (LHS != RHS) {
590 ...
591 }
592 }
593 }
594</pre>
595</div>
596
597<p>When you have very very small loops, this sort of structure is fine, but if
598it exceeds more than 10-15 lines, it becomes difficult for people to read and
599understand at a glance.
600The problem with this sort of code is that it gets very nested very quickly,
601meaning that the reader of the code has to keep a lot of context in their brain
602to remember what is going immediately on in the loop, because they don't know
603if/when the if conditions will have elses etc. It is strongly preferred to
604structure the loop like this:</p>
605
606<div class="doc_code">
607<pre>
608 for (BasicBlock::iterator II = BB-&gt;begin(), E = BB-&gt;end(); II != E; ++II) {
609 BinaryOperator *BO = dyn_cast&lt;BinaryOperator&gt;(II);
610 if (!BO) continue;
611
612 Value *LHS = BO-&gt;getOperand(0);
613 Value *RHS = BO-&gt;getOperand(1);
614 if (LHS == RHS) continue;
615 }
616</pre>
617</div>
618
619<p>This has all the benefits of using early exits from functions: it reduces
620nesting of the loop, it makes it easier to describe why the conditions are true,
621and it makes it obvious to the reader that there is no "else" coming up that
622they have to push context into their brain for. If a loop is large, this can
623be a big understandability win.</p>
624
625</div>
626
627
628<!-- _______________________________________________________________________ -->
629<div class="doc_subsubsection">
630 <a name="hl_predicateloops">Turn Predicate Loops into Predicate Functions</a>
631</div>
632
633<div class="doc_text">
634
Chris Lattner5af04d22009-07-22 16:54:14 +0000[diff] [blame]635<p>It is very common to write small loops that just compute a boolean
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]636 value. There are a number of ways that people commonly write these, but an
637 example of this sort of thing is:</p>
638
639<div class="doc_code">
640<pre>
641 <b>bool FoundFoo = false;</b>
642 for (unsigned i = 0, e = BarList.size(); i != e; ++i)
643 if (BarList[i]-&gt;isFoo()) {
644 <b>FoundFoo = true;</b>
645 break;
646 }
647
648 <b>if (FoundFoo) {</b>
649 ...
650 }
651</pre>
652</div>
653
654<p>This sort of code is awkward to write, and is almost always a bad sign.
655Instead of this sort of loop, we strongly prefer to use a predicate function
656(which may be <a href="#micro_anonns">static</a>) that uses
Chris Lattner5af04d22009-07-22 16:54:14 +0000[diff] [blame]657<a href="#hl_earlyexit">early exits</a> to compute the predicate. We prefer
658the code to be structured like this:
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]659</p>
660
661
662<div class="doc_code">
663<pre>
664/// ListContainsFoo - Return true if the specified list has an element that is
665/// a foo.
666static bool ListContainsFoo(const std::vector&lt;Bar*&gt; &amp;List) {
667 for (unsigned i = 0, e = List.size(); i != e; ++i)
668 if (List[i]-&gt;isFoo())
669 return true;
670 return false;
671}
672...
673
674 <b>if (ListContainsFoo(BarList)) {</b>
675 ...
676 }
677</pre>
678</div>
679
680<p>There are many reasons for doing this: it reduces indentation and factors out
681code which can often be shared by other code that checks for the same predicate.
682More importantly, it <em>forces you to pick a name</em> for the function, and
683forces you to write a comment for it. In this silly example, this doesn't add
684much value. However, if the condition is complex, this can make it a lot easier
685for the reader to understand the code that queries for this predicate. Instead
Chris Lattner6ead3272009-07-22 16:30:39 +0000[diff] [blame]686of being faced with the in-line details of how we check to see if the BarList
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]687contains a foo, we can trust the function name and continue reading with better
688locality.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]689
690</div>
691
692
693<!-- ======================================================================= -->
694<div class="doc_subsection">
695 <a name="micro">The Low Level Issues</a>
696</div>
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]697<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]698
699
700<!-- _______________________________________________________________________ -->
701<div class="doc_subsubsection">
702 <a name="ll_assert">Assert Liberally</a>
703</div>
704
705<div class="doc_text">
706
707<p>Use the "<tt>assert</tt>" function to its fullest. Check all of your
708preconditions and assumptions, you never know when a bug (not neccesarily even
709yours) might be caught early by an assertion, which reduces debugging time
710dramatically. The "<tt>&lt;cassert&gt;</tt>" header file is probably already
711included by the header files you are using, so it doesn't cost anything to use
712it.</p>
713
714<p>To further assist with debugging, make sure to put some kind of error message
715in the assertion statement (which is printed if the assertion is tripped). This
716helps the poor debugging make sense of why an assertion is being made and
717enforced, and hopefully what to do about it. Here is one complete example:</p>
718
719<div class="doc_code">
720<pre>
721inline Value *getOperand(unsigned i) {
722 assert(i &lt; Operands.size() &amp;&amp; "getOperand() out of range!");
723 return Operands[i];
724}
725</pre>
726</div>
727
728<p>Here are some examples:</p>
729
730<div class="doc_code">
731<pre>
732assert(Ty-&gt;isPointerType() &amp;&amp; "Can't allocate a non pointer type!");
733
734assert((Opcode == Shl || Opcode == Shr) &amp;&amp; "ShiftInst Opcode invalid!");
735
736assert(idx &lt; getNumSuccessors() &amp;&amp; "Successor # out of range!");
737
738assert(V1.getType() == V2.getType() &amp;&amp; "Constant types must be identical!");
739
740assert(isa&lt;PHINode&gt;(Succ-&gt;front()) &amp;&amp; "Only works on PHId BBs!");
741</pre>
742</div>
743
744<p>You get the idea...</p>
745
Nick Lewycky009216a2008-05-31 23:54:55 +0000[diff] [blame]746<p>Please be aware when adding assert statements that not all compilers are aware of
747the semantics of the assert. In some places, asserts are used to indicate a piece of
748code that should not be reached. These are typically of the form:</p>
749
750<div class="doc_code">
751<pre>
Dan Gohman4dfac702008-11-24 17:18:39 +0000[diff] [blame]752assert(0 &amp;&amp; "Some helpful error message");
Nick Lewycky009216a2008-05-31 23:54:55 +0000[diff] [blame]753</pre>
754</div>
755
756<p>When used in a function that returns a value, they should be followed with a return
757statement and a comment indicating that this line is never reached. This will prevent
758a compiler which is unable to deduce that the assert statement never returns from
759generating a warning.</p>
760
761<div class="doc_code">
762<pre>
Dan Gohman4dfac702008-11-24 17:18:39 +0000[diff] [blame]763assert(0 &amp;&amp; "Some helpful error message");
Nick Lewycky009216a2008-05-31 23:54:55 +0000[diff] [blame]764// Not reached
765return 0;
766</pre>
767</div>
768
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]769</div>
770
771<!-- _______________________________________________________________________ -->
772<div class="doc_subsubsection">
773 <a name="ll_ns_std">Do not use '<tt>using namespace std</tt>'</a>
774</div>
775
776<div class="doc_text">
777<p>In LLVM, we prefer to explicitly prefix all identifiers from the standard
778namespace with an "<tt>std::</tt>" prefix, rather than rely on
779"<tt>using namespace std;</tt>".</p>
780
781<p> In header files, adding a '<tt>using namespace XXX</tt>' directive pollutes
Misha Brukmanbcc53732008-12-11 19:37:04 +0000[diff] [blame]782the namespace of any source file that <tt>#include</tt>s the header. This is
783clearly a bad thing.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]784
785<p>In implementation files (e.g. .cpp files), the rule is more of a stylistic
786rule, but is still important. Basically, using explicit namespace prefixes
787makes the code <b>clearer</b>, because it is immediately obvious what facilities
788are being used and where they are coming from, and <b>more portable</b>, because
789namespace clashes cannot occur between LLVM code and other namespaces. The
790portability rule is important because different standard library implementations
791expose different symbols (potentially ones they shouldn't), and future revisions
792to the C++ standard will add more symbols to the <tt>std</tt> namespace. As
793such, we never use '<tt>using namespace std;</tt>' in LLVM.</p>
794
795<p>The exception to the general rule (i.e. it's not an exception for
796the <tt>std</tt> namespace) is for implementation files. For example, all of
797the code in the LLVM project implements code that lives in the 'llvm' namespace.
798As such, it is ok, and actually clearer, for the .cpp files to have a '<tt>using
799namespace llvm</tt>' directive at their top, after the <tt>#include</tt>s. The
800general form of this rule is that any .cpp file that implements code in any
801namespace may use that namespace (and its parents'), but should not use any
802others.</p>
803
804</div>
805
806<!-- _______________________________________________________________________ -->
807<div class="doc_subsubsection">
808 <a name="ll_virtual_anch">Provide a virtual method anchor for classes
809 in headers</a>
810</div>
811
812<div class="doc_text">
813
814<p>If a class is defined in a header file and has a v-table (either it has
815virtual methods or it derives from classes with virtual methods), it must
816always have at least one out-of-line virtual method in the class. Without
Misha Brukmanbcc53732008-12-11 19:37:04 +0000[diff] [blame]817this, the compiler will copy the vtable and RTTI into every <tt>.o</tt> file
818that <tt>#include</tt>s the header, bloating <tt>.o</tt> file sizes and
819increasing link times.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]820
821</div>
822
Chris Lattner01b8af32009-06-30 06:13:23 +0000[diff] [blame]823<!-- _______________________________________________________________________ -->
824<div class="doc_subsubsection">
825 <a name="ll_end">Don't evaluate end() every time through a loop</a>
826</div>
827
828<div class="doc_text">
829
830<p>Because C++ doesn't have a standard "foreach" loop (though it can be emulated
831with macros and may be coming in C++'0x) we end up writing a lot of loops that
832manually iterate from begin to end on a variety of containers or through other
833data structures. One common mistake is to write a loop in this style:</p>
834
835<div class="doc_code">
836<pre>
837 BasicBlock *BB = ...
838 for (BasicBlock::iterator I = BB->begin(); I != <b>BB->end()</b>; ++I)
839 ... use I ...
840</pre>
841</div>
842
843<p>The problem with this construct is that it evaluates "<tt>BB->end()</tt>"
844every time through the loop. Instead of writing the loop like this, we strongly
845prefer loops to be written so that they evaluate it once before the loop starts.
846A convenient way to do this is like so:</p>
847
848<div class="doc_code">
849<pre>
850 BasicBlock *BB = ...
851 for (BasicBlock::iterator I = BB->begin(), E = <b>BB->end()</b>; I != E; ++I)
852 ... use I ...
853</pre>
854</div>
855
856<p>The observant may quickly point out that these two loops may have different
857semantics: if the container (a basic block in this case) is being mutated, then
858"<tt>BB->end()</tt>" may change its value every time through the loop and the
859second loop may not in fact be correct. If you actually do depend on this
Chris Lattnerdc43b342009-06-30 06:20:03 +0000[diff] [blame]860behavior, please write the loop in the first form and add a comment indicating
Chris Lattner01b8af32009-06-30 06:13:23 +0000[diff] [blame]861that you did it intentionally.</p>
862
863<p>Why do we prefer the second form (when correct)? Writing the loop in the
864first form has two problems: First it may be less efficient than evaluating it
865at the start of the loop. In this case, the cost is probably minor: a few extra
866loads every time through the loop. However, if the base expression is more
867complex, then the cost can rise quickly. I've seen loops where the end
868expression was actually something like: "<tt>SomeMap[x]->end()</tt>" and map
Chris Lattner088affa2009-06-30 06:27:54 +0000[diff] [blame]869lookups really aren't cheap. By writing it in the second form consistently, you
Chris Lattner01b8af32009-06-30 06:13:23 +0000[diff] [blame]870eliminate the issue entirely and don't even have to think about it.</p>
871
Chris Lattnerdc43b342009-06-30 06:20:03 +0000[diff] [blame]872<p>The second (even bigger) issue is that writing the loop in the first form
Chris Lattner01b8af32009-06-30 06:13:23 +0000[diff] [blame]873hints to the reader that the loop is mutating the container (a fact that a
874comment would handily confirm!). If you write the loop in the second form, it
875is immediately obvious without even looking at the body of the loop that the
876container isn't being modified, which makes it easier to read the code and
877understand what it does.</p>
878
879<p>While the second form of the loop is a few extra keystrokes, we do strongly
880prefer it.</p>
881
882</div>
883
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]884<!-- _______________________________________________________________________ -->
885<div class="doc_subsubsection">
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]886 <a name="ll_iostream"><tt>#include &lt;iostream&gt;</tt> is forbidden</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]887</div>
888
889<div class="doc_text">
890
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]891<p>The use of <tt>#include &lt;iostream&gt;</tt> in library files is
892hereby <b><em>forbidden</em></b>. The primary reason for doing this is to
893support clients using LLVM libraries as part of larger systems. In particular,
894we statically link LLVM into some dynamic libraries. Even if LLVM isn't used,
895the static c'tors are run whenever an application start up that uses the dynamic
896library. There are two problems with this:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]897
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]898<ol>
899 <li>The time to run the static c'tors impacts startup time of
900 applications&mdash;a critical time for GUI apps.</li>
901 <li>The static c'tors cause the app to pull many extra pages of memory off the
902 disk: both the code for the static c'tors in each <tt>.o</tt> file and the
903 small amount of data that gets touched. In addition, touched/dirty pages
904 put more pressure on the VM system on low-memory machines.</li>
905</ol>
906
907<p>Note that using the other stream headers (<tt>&lt;sstream&gt;</tt> for
908example) is allowed normally, it is just <tt>&lt;iostream&gt;</tt> that is
909causing problems.</p>
910
Daniel Dunbar2f099eb2009-07-24 23:04:51 +0000[diff] [blame^]911<p>In addition, new code should always
912use <a href="#ll_raw_ostream"><tt>raw_ostream</tt></a> or
913the <tt>llvm::MemoryBuffer</tt> API (for reading in files).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]914
915</div>
916
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]917
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]918<!-- _______________________________________________________________________ -->
919<div class="doc_subsubsection">
920 <a name="ll_avoidendl">Avoid <tt>std::endl</tt></a>
921</div>
922
923<div class="doc_text">
924
925<p>The <tt>std::endl</tt> modifier, when used with iostreams outputs a newline
926to the output stream specified. In addition to doing this, however, it also
927flushes the output stream. In other words, these are equivalent:</p>
928
929<div class="doc_code">
930<pre>
931std::cout &lt;&lt; std::endl;
932std::cout &lt;&lt; '\n' &lt;&lt; std::flush;
933</pre>
934</div>
935
936<p>Most of the time, you probably have no reason to flush the output stream, so
937it's better to use a literal <tt>'\n'</tt>.</p>
938
939</div>
940
941
Daniel Dunbar2f099eb2009-07-24 23:04:51 +0000[diff] [blame^]942<!-- _______________________________________________________________________ -->
943<div class="doc_subsubsection">
944 <a name="ll_raw_ostream">Use <tt>raw_ostream</tt></a>
945</div>
946
947<div class="doc_text">
948
949<p>LLVM includes a lightweight, simple, and efficient stream implementation
950in <tt>llvm/Support/raw_ostream.h</tt> which provides all of the common features
951of <tt>std::iostream</tt>. All new code should use <tt>raw_ostream</tt> instead
952of <tt>iostream</tt>.</p>
953
954<p>Unlike <tt>std::iostream</tt>, <tt>raw_ostream</tt> is not a template and can
955be forward declared as <tt>class raw_ostream</tt>. Public headers should
956generally not include the <tt>raw_ostream</tt> header, but use forward
957declarations and constant references to <tt>raw_ostream</tt> instances.</p>
958
959</div>
960
961
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]962<!-- ======================================================================= -->
963<div class="doc_subsection">
964 <a name="nano">Microscopic Details</a>
965</div>
966<!-- ======================================================================= -->
967
968<p>This section describes preferred low-level formatting guidelines along with
969reasoning on why we prefer them.</p>
970
971<!-- _______________________________________________________________________ -->
972<div class="doc_subsubsection">
973 <a name="micro_spaceparen">Spaces Before Parentheses</a>
974</div>
975
976<div class="doc_text">
977
978<p>We prefer to put a space before a parentheses only in control flow
979statements, but not in normal function call expressions and function-like
980macros. For example, this is good:</p>
981
982<div class="doc_code">
983<pre>
984 <b>if (</b>x) ...
985 <b>for (</b>i = 0; i != 100; ++i) ...
986 <b>while (</b>llvm_rocks) ...
987
988 <b>somefunc(</b>42);
989 <b><a href="#ll_assert">assert</a>(</b>3 != 4 &amp;&amp; "laws of math are failing me");
990
991 a = <b>foo(</b>42, 92) + <b>bar(</b>x);
992 </pre>
993</div>
994
995<p>... and this is bad:</p>
996
997<div class="doc_code">
998<pre>
999 <b>if(</b>x) ...
1000 <b>for(</b>i = 0; i != 100; ++i) ...
1001 <b>while(</b>llvm_rocks) ...
1002
1003 <b>somefunc (</b>42);
1004 <b><a href="#ll_assert">assert</a> (</b>3 != 4 &amp;&amp; "laws of math are failing me");
1005
1006 a = <b>foo (</b>42, 92) + <b>bar (</b>x);
1007</pre>
1008</div>
1009
1010<p>The reason for doing this is not completely arbitrary. This style makes
1011 control flow operators stand out more, and makes expressions flow better. The
1012 function call operator binds very tightly as a postfix operator. Putting
1013 a space after a function name (as in the last example) makes it appear that
1014 the code might bind the arguments of the left-hand-side of a binary operator
1015 with the argument list of a function and the name of the right side. More
1016 specifically, it is easy to misread the "a" example as:</p>
1017
1018<div class="doc_code">
1019<pre>
1020 a = foo <b>(</b>(42, 92) + bar<b>)</b> (x);
1021</pre>
1022</div>
1023
1024<p>... when skimming through the code. By avoiding a space in a function, we
1025avoid this misinterpretation.</p>
1026
1027</div>
1028
1029<!-- _______________________________________________________________________ -->
1030<div class="doc_subsubsection">
1031 <a name="micro_preincrement">Prefer Preincrement</a>
1032</div>
1033
1034<div class="doc_text">
1035
1036<p>Hard fast rule: Preincrement (<tt>++X</tt>) may be no slower than
1037postincrement (<tt>X++</tt>) and could very well be a lot faster than it. Use
1038preincrementation whenever possible.</p>
1039
1040<p>The semantics of postincrement include making a copy of the value being
1041incremented, returning it, and then preincrementing the "work value". For
1042primitive types, this isn't a big deal... but for iterators, it can be a huge
1043issue (for example, some iterators contains stack and set objects in them...
1044copying an iterator could invoke the copy ctor's of these as well). In general,
1045get in the habit of always using preincrement, and you won't have a problem.</p>
1046
1047</div>
1048
1049<!-- _______________________________________________________________________ -->
1050<div class="doc_subsubsection">
1051 <a name="micro_namespaceindent">Namespace Indentation</a>
1052</div>
1053
1054<div class="doc_text">
1055
1056<p>
1057In general, we strive to reduce indentation where ever possible. This is useful
1058because we want code to <a href="#scf_codewidth">fit into 80 columns</a> without
1059wrapping horribly, but also because it makes it easier to understand the code.
1060Namespaces are a funny thing: they are often large, and we often desire to put
1061lots of stuff into them (so they can be large). Other times they are tiny,
1062because they just hold an enum or something similar. In order to balance this,
1063we use different approaches for small versus large namespaces.
1064</p>
1065
1066<p>
1067If a namespace definition is small and <em>easily</em> fits on a screen (say,
1068less than 35 lines of code), then you should indent its body. Here's an
1069example:
1070</p>
1071
1072<div class="doc_code">
1073<pre>
Chris Lattner5af04d22009-07-22 16:54:14 +0000[diff] [blame]1074namespace llvm {
1075 namespace X86 {
1076 /// RelocationType - An enum for the x86 relocation codes. Note that
1077 /// the terminology here doesn't follow x86 convention - word means
1078 /// 32-bit and dword means 64-bit.
1079 enum RelocationType {
1080 /// reloc_pcrel_word - PC relative relocation, add the relocated value to
1081 /// the value already in memory, after we adjust it for where the PC is.
1082 reloc_pcrel_word = 0,
1083
1084 /// reloc_picrel_word - PIC base relative relocation, add the relocated
1085 /// value to the value already in memory, after we adjust it for where the
1086 /// PIC base is.
1087 reloc_picrel_word = 1,
1088
1089 /// reloc_absolute_word, reloc_absolute_dword - Absolute relocation, just
1090 /// add the relocated value to the value already in memory.
1091 reloc_absolute_word = 2,
1092 reloc_absolute_dword = 3
1093 };
1094 }
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]1095}
1096</pre>
1097</div>
1098
1099<p>Since the body is small, indenting adds value because it makes it very clear
1100where the namespace starts and ends, and it is easy to take the whole thing in
1101in one "gulp" when reading the code. If the blob of code in the namespace is
1102larger (as it typically is in a header in the llvm or clang namespaces), do not
1103indent the code, and add a comment indicating what namespace is being closed.
1104For example:</p>
1105
1106<div class="doc_code">
1107<pre>
1108namespace llvm {
1109namespace knowledge {
1110
1111/// Grokable - This class represents things that Smith can have an intimate
1112/// understanding of and contains the data associated with it.
1113class Grokable {
1114...
1115public:
1116 explicit Grokable() { ... }
1117 virtual ~Grokable() = 0;
1118
1119 ...
1120
1121};
1122
1123} // end namespace knowledge
1124} // end namespace llvm
1125</pre>
1126</div>
1127
1128<p>Because the class is large, we don't expect that the reader can easily
1129understand the entire concept in a glance, and the end of the file (where the
1130namespaces end) may be a long ways away from the place they open. As such,
1131indenting the contents of the namespace doesn't add any value, and detracts from
1132the readability of the class. In these cases it is best to <em>not</em> indent
1133the contents of the namespace.</p>
1134
1135</div>
1136
1137<!-- _______________________________________________________________________ -->
1138<div class="doc_subsubsection">
1139 <a name="micro_anonns">Anonymous Namespaces</a>
1140</div>
1141
1142<div class="doc_text">
1143
Chris Lattner5af04d22009-07-22 16:54:14 +0000[diff] [blame]1144<p>After talking about namespaces in general, you may be wondering about
1145anonymous namespaces in particular.
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]1146Anonymous namespaces are a great language feature that tells the C++ compiler
1147that the contents of the namespace are only visible within the current
1148translation unit, allowing more aggressive optimization and eliminating the
1149possibility of symbol name collisions. Anonymous namespaces are to C++ as
1150"static" is to C functions and global variables. While "static" is available
1151in C++, anonymous namespaces are more general: they can make entire classes
1152private to a file.</p>
1153
1154<p>The problem with anonymous namespaces is that they naturally want to
1155encourage indentation of their body, and they reduce locality of reference: if
1156you see a random function definition in a C++ file, it is easy to see if it is
1157marked static, but seeing if it is in an anonymous namespace requires scanning
1158a big chunk of the file.</p>
1159
1160<p>Because of this, we have a simple guideline: make anonymous namespaces as
1161small as possible, and only use them for class declarations. For example, this
1162is good:</p>
1163
1164<div class="doc_code">
1165<pre>
1166<b>namespace {</b>
1167 class StringSort {
1168 ...
1169 public:
1170 StringSort(...)
1171 bool operator&lt;(const char *RHS) const;
1172 };
1173<b>} // end anonymous namespace</b>
1174
1175static void Helper() {
1176 ...
1177}
1178
1179bool StringSort::operator&lt;(const char *RHS) const {
1180 ...
1181}
1182
1183</pre>
1184</div>
1185
1186<p>This is bad:</p>
1187
1188
1189<div class="doc_code">
1190<pre>
1191<b>namespace {</b>
1192class StringSort {
1193...
1194public:
1195 StringSort(...)
1196 bool operator&lt;(const char *RHS) const;
1197};
1198
1199void Helper() {
1200 ...
1201}
1202
1203bool StringSort::operator&lt;(const char *RHS) const {
1204 ...
1205}
1206
1207<b>} // end anonymous namespace</b>
1208
1209</pre>
1210</div>
1211
1212
1213<p>This is bad specifically because if you're looking at "Helper" in the middle
1214of a large C++ file, that you have no immediate way to tell if it is local to
1215the file. When it is marked static explicitly, this is immediately obvious.
1216Also, there is no reason to enclose the definition of "operator&lt;" in the
Chris Lattner5af04d22009-07-22 16:54:14 +0000[diff] [blame]1217namespace just because it was declared there.
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]1218</p>
1219
1220</div>
1221
1222
1223
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1224<!-- *********************************************************************** -->
1225<div class="doc_section">
1226 <a name="seealso">See Also</a>
1227</div>
1228<!-- *********************************************************************** -->
1229
1230<div class="doc_text">
1231
1232<p>A lot of these comments and recommendations have been culled for other
1233sources. Two particularly important books for our work are:</p>
1234
1235<ol>
1236
Chris Lattnerb5955b72007-11-09 21:49:08 +0000[diff] [blame]1237<li><a href="http://www.amazon.com/Effective-Specific-Addison-Wesley-Professional-Computing/dp/0321334876">Effective
1238C++</a> by Scott Meyers. Also
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1239interesting and useful are "More Effective C++" and "Effective STL" by the same
1240author.</li>
1241
Chris Lattnerb5955b72007-11-09 21:49:08 +0000[diff] [blame]1242<li>Large-Scale C++ Software Design by John Lakos</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1243
1244</ol>
1245
1246<p>If you get some free time, and you haven't read them: do so, you might learn
1247something.</p>
1248
1249</div>
1250
1251<!-- *********************************************************************** -->
1252
1253<hr>
1254<address>
1255 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman947321d2008-12-11 17:34:48 +0000[diff] [blame]1256 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1257 <a href="http://validator.w3.org/check/referer"><img
Misha Brukman947321d2008-12-11 17:34:48 +0000[diff] [blame]1258 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1259
1260 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
1261 <a href="http://llvm.org">LLVM Compiler Infrastructure</a><br>
1262 Last modified: $Date$
1263</address>
1264
1265</body>
1266</html>