Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / 391efe0de27f008e3e3e47af958ea7851ee40842 / . / docs / CodingStandards.html
blob: 2fe3bdc9e2bbb1073b9dcf254d1b861a7179a5da [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>
Chris Lattner598fcfd2009-07-28 22:54:04 +0000[diff] [blame]46 <li><a href="#hl_else_after_return">Don't use "else" after a
47 return</a></li>
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]48 <li><a href="#hl_predicateloops">Turn Predicate Loops into Predicate
49 Functions</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]50 </ol></li>
51 <li><a href="#micro">The Low Level Issues</a>
52 <ol>
53 <li><a href="#ll_assert">Assert Liberally</a></li>
54 <li><a href="#ll_ns_std">Do not use 'using namespace std'</a></li>
55 <li><a href="#ll_virtual_anch">Provide a virtual method anchor for
56 classes in headers</a></li>
Chris Lattner01b8af32009-06-30 06:13:23 +0000[diff] [blame]57 <li><a href="#ll_end">Don't evaluate end() every time through a
58 loop</a></li>
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]59 <li><a href="#ll_iostream"><tt>#include &lt;iostream&gt;</tt> is
60 <em>forbidden</em></a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]61 <li><a href="#ll_avoidendl">Avoid <tt>std::endl</tt></a></li>
Daniel Dunbar2f099eb2009-07-24 23:04:51 +0000[diff] [blame]62 <li><a href="#ll_raw_ostream">Use <tt>raw_ostream</tt></a</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]63 </ol></li>
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]64
65 <li><a href="#nano">Microscopic Details</a>
66 <ol>
67 <li><a href="#micro_spaceparen">Spaces Before Parentheses</a></li>
68 <li><a href="#micro_preincrement">Prefer Preincrement</a></li>
69 <li><a href="#micro_namespaceindent">Namespace Indentation</a></li>
70 <li><a href="#micro_anonns">Anonymous Namespaces</a></li>
71 </ol></li>
72
73
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]74 </ol></li>
75 <li><a href="#seealso">See Also</a></li>
76</ol>
77
78<div class="doc_author">
Chris Lattnere7346a92009-07-22 05:43:01 +0000[diff] [blame]79 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]80</div>
81
82
83<!-- *********************************************************************** -->
84<div class="doc_section">
85 <a name="introduction">Introduction</a>
86</div>
87<!-- *********************************************************************** -->
88
89<div class="doc_text">
90
91<p>This document attempts to describe a few coding standards that are being used
92in the LLVM source tree. Although no coding standards should be regarded as
93absolute requirements to be followed in all instances, coding standards can be
94useful.</p>
95
96<p>This document intentionally does not prescribe fixed standards for religious
97issues such as brace placement and space usage. For issues like this, follow
98the golden rule:</p>
99
100<blockquote>
101
102<p><b><a name="goldenrule">If you are adding a significant body of source to a
103project, feel free to use whatever style you are most comfortable with. If you
104are extending, enhancing, or bug fixing already implemented code, use the style
105that is already being used so that the source is uniform and easy to
106follow.</a></b></p>
107
108</blockquote>
109
110<p>The ultimate goal of these guidelines is the increase readability and
111maintainability of our common source base. If you have suggestions for topics to
112be included, please mail them to <a
113href="mailto:sabre@nondot.org">Chris</a>.</p>
114
115</div>
116
117<!-- *********************************************************************** -->
118<div class="doc_section">
119 <a name="mechanicalissues">Mechanical Source Issues</a>
120</div>
121<!-- *********************************************************************** -->
122
123<!-- ======================================================================= -->
124<div class="doc_subsection">
125 <a name="sourceformating">Source Code Formatting</a>
126</div>
127
128<!-- _______________________________________________________________________ -->
129<div class="doc_subsubsection">
130 <a name="scf_commenting">Commenting</a>
131</div>
132
133<div class="doc_text">
134
135<p>Comments are one critical part of readability and maintainability. Everyone
Chris Lattner48386692009-07-12 00:10:24 +0000[diff] [blame]136knows they should comment, so should you. When writing comments, write them as
137English prose, which means they should use proper capitalization, punctuation,
138etc. Although we all should probably
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]139comment our code more than we do, there are a few very critical places that
140documentation is very useful:</p>
141
142<b>File Headers</b>
143
144<p>Every source file should have a header on it that describes the basic
145purpose of the file. If a file does not have a header, it should not be
146checked into Subversion. Most source trees will probably have a standard
147file header format. The standard format for the LLVM source tree looks like
148this:</p>
149
150<div class="doc_code">
151<pre>
152//===-- llvm/Instruction.h - Instruction class definition -------*- C++ -*-===//
Misha Brukman0fd5f3e2009-01-02 16:58:42 +0000[diff] [blame]153//
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]154// The LLVM Compiler Infrastructure
155//
Chris Lattnerfb5959f2007-12-29 19:56:08 +0000[diff] [blame]156// This file is distributed under the University of Illinois Open Source
157// License. See LICENSE.TXT for details.
Misha Brukman0fd5f3e2009-01-02 16:58:42 +0000[diff] [blame]158//
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]159//===----------------------------------------------------------------------===//
160//
161// This file contains the declaration of the Instruction class, which is the
162// base class for all of the VM instructions.
163//
164//===----------------------------------------------------------------------===//
165</pre>
166</div>
167
Chris Lattnerfb5959f2007-12-29 19:56:08 +0000[diff] [blame]168<p>A few things to note about this particular format: The "<tt>-*- C++
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]169-*-</tt>" string on the first line is there to tell Emacs that the source file
170is a C++ file, not a C file (Emacs assumes .h files are C files by default).
171Note that this tag is not necessary in .cpp files. The name of the file is also
172on the first line, along with a very short description of the purpose of the
173file. This is important when printing out code and flipping though lots of
174pages.</p>
175
Chris Lattnerfb5959f2007-12-29 19:56:08 +0000[diff] [blame]176<p>The next section in the file is a concise note that defines the license
177that the file is released under. This makes it perfectly clear what terms the
178source code can be distributed under and should not be modified in any way.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]179
180<p>The main body of the description does not have to be very long in most cases.
181Here it's only two lines. If an algorithm is being implemented or something
182tricky is going on, a reference to the paper where it is published should be
183included, as well as any notes or "gotchas" in the code to watch out for.</p>
184
185<b>Class overviews</b>
186
187<p>Classes are one fundamental part of a good object oriented design. As such,
188a class definition should have a comment block that explains what the class is
189used for... if it's not obvious. If it's so completely obvious your grandma
190could figure it out, it's probably safe to leave it out. Naming classes
191something sane goes a long ways towards avoiding writing documentation.</p>
192
193
194<b>Method information</b>
195
196<p>Methods defined in a class (as well as any global functions) should also be
197documented properly. A quick note about what it does any a description of the
198borderline behaviour is all that is necessary here (unless something
199particularly tricky or insideous is going on). The hope is that people can
200figure out how to use your interfaces without reading the code itself... that is
201the goal metric.</p>
202
203<p>Good things to talk about here are what happens when something unexpected
204happens: does the method return null? Abort? Format your hard disk?</p>
205
206</div>
207
208<!-- _______________________________________________________________________ -->
209<div class="doc_subsubsection">
210 <a name="scf_commentformat">Comment Formatting</a>
211</div>
212
213<div class="doc_text">
214
215<p>In general, prefer C++ style (<tt>//</tt>) comments. They take less space,
216require less typing, don't have nesting problems, etc. There are a few cases
217when it is useful to use C style (<tt>/* */</tt>) comments however:</p>
218
219<ol>
220 <li>When writing a C code: Obviously if you are writing C code, use C style
221 comments.</li>
222 <li>When writing a header file that may be <tt>#include</tt>d by a C source
223 file.</li>
224 <li>When writing a source file that is used by a tool that only accepts C
225 style comments.</li>
226</ol>
227
228<p>To comment out a large block of code, use <tt>#if 0</tt> and <tt>#endif</tt>.
229These nest properly and are better behaved in general than C style comments.</p>
230
231</div>
232
233<!-- _______________________________________________________________________ -->
234<div class="doc_subsubsection">
235 <a name="scf_includes"><tt>#include</tt> Style</a>
236</div>
237
238<div class="doc_text">
239
240<p>Immediately after the <a href="#scf_commenting">header file comment</a> (and
241include guards if working on a header file), the <a
242href="#hl_dontinclude">minimal</a> list of <tt>#include</tt>s required by the
243file should be listed. We prefer these <tt>#include</tt>s to be listed in this
244order:</p>
245
246<ol>
247 <li><a href="#mmheader">Main Module header</a></li>
248 <li><a href="#hl_privateheaders">Local/Private Headers</a></li>
249 <li><tt>llvm/*</tt></li>
250 <li><tt>llvm/Analysis/*</tt></li>
251 <li><tt>llvm/Assembly/*</tt></li>
252 <li><tt>llvm/Bytecode/*</tt></li>
253 <li><tt>llvm/CodeGen/*</tt></li>
254 <li>...</li>
255 <li><tt>Support/*</tt></li>
256 <li><tt>Config/*</tt></li>
257 <li>System <tt>#includes</tt></li>
258</ol>
259
Nick Lewycky61819f72008-11-29 20:13:25 +0000[diff] [blame]260<p>... and each category should be sorted by name.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]261
262<p><a name="mmheader">The "Main Module Header"</a> file applies to .cpp file
263which implement an interface defined by a .h file. This <tt>#include</tt>
264should always be included <b>first</b> regardless of where it lives on the file
265system. By including a header file first in the .cpp files that implement the
266interfaces, we ensure that the header does not have any hidden dependencies
267which are not explicitly #included in the header, but should be. It is also a
268form of documentation in the .cpp file to indicate where the interfaces it
269implements are defined.</p>
270
271</div>
272
273<!-- _______________________________________________________________________ -->
274<div class="doc_subsubsection">
275 <a name="scf_codewidth">Source Code Width</a>
276</div>
277
278<div class="doc_text">
279
280<p>Write your code to fit within 80 columns of text. This helps those of us who
281like to print out code and look at your code in an xterm without resizing
282it.</p>
283
Chris Lattnerb5daf702008-07-08 05:12:37 +0000[diff] [blame]284<p>The longer answer is that there must be some limit to the width of the code
285in order to reasonably allow developers to have multiple files side-by-side in
286windows on a modest display. If you are going to pick a width limit, it is
287somewhat arbitrary but you might as well pick something standard. Going with
28890 columns (for example) instead of 80 columns wouldn't add any significant
289value and would be detrimental to printing out code. Also many other projects
290have standardized on 80 columns, so some people have already configured their
291editors for it (vs something else, like 90 columns).</p>
292
293<p>This is one of many contentious issues in coding standards, but is not up
294for debate.</p>
295
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]296</div>
297
298<!-- _______________________________________________________________________ -->
299<div class="doc_subsubsection">
300 <a name="scf_spacestabs">Use Spaces Instead of Tabs</a>
301</div>
302
303<div class="doc_text">
304
305<p>In all cases, prefer spaces to tabs in source files. People have different
306prefered indentation levels, and different styles of indentation that they
307like... this is fine. What isn't is that different editors/viewers expand tabs
308out to different tab stops. This can cause your code to look completely
309unreadable, and it is not worth dealing with.</p>
310
311<p>As always, follow the <a href="#goldenrule">Golden Rule</a> above: follow the
312style of existing code if your are modifying and extending it. If you like four
313spaces of indentation, <b>DO NOT</b> do that in the middle of a chunk of code
314with two spaces of indentation. Also, do not reindent a whole source file: it
315makes for incredible diffs that are absolutely worthless.</p>
316
317</div>
318
319<!-- _______________________________________________________________________ -->
320<div class="doc_subsubsection">
321 <a name="scf_indentation">Indent Code Consistently</a>
322</div>
323
324<div class="doc_text">
325
326<p>Okay, your first year of programming you were told that indentation is
327important. If you didn't believe and internalize this then, now is the time.
328Just do it.</p>
329
330</div>
331
332
333<!-- ======================================================================= -->
334<div class="doc_subsection">
335 <a name="compilerissues">Compiler Issues</a>
336</div>
337
338
339<!-- _______________________________________________________________________ -->
340<div class="doc_subsubsection">
341 <a name="ci_warningerrors">Treat Compiler Warnings Like Errors</a>
342</div>
343
344<div class="doc_text">
345
346<p>If your code has compiler warnings in it, something is wrong: you aren't
347casting values correctly, your have "questionable" constructs in your code, or
348you are doing something legitimately wrong. Compiler warnings can cover up
349legitimate errors in output and make dealing with a translation unit
350difficult.</p>
351
352<p>It is not possible to prevent all warnings from all compilers, nor is it
353desirable. Instead, pick a standard compiler (like <tt>gcc</tt>) that provides
354a good thorough set of warnings, and stick to them. At least in the case of
355<tt>gcc</tt>, it is possible to work around any spurious errors by changing the
356syntax of the code slightly. For example, an warning that annoys me occurs when
357I write code like this:</p>
358
359<div class="doc_code">
360<pre>
361if (V = getValue()) {
362 ...
363}
364</pre>
365</div>
366
367<p><tt>gcc</tt> will warn me that I probably want to use the <tt>==</tt>
368operator, and that I probably mistyped it. In most cases, I haven't, and I
369really don't want the spurious errors. To fix this particular problem, I
370rewrite the code like this:</p>
371
372<div class="doc_code">
373<pre>
374if ((V = getValue())) {
375 ...
376}
377</pre>
378</div>
379
380<p>...which shuts <tt>gcc</tt> up. Any <tt>gcc</tt> warning that annoys you can
381be fixed by massaging the code appropriately.</p>
382
383<p>These are the <tt>gcc</tt> warnings that I prefer to enable: <tt>-Wall
384-Winline -W -Wwrite-strings -Wno-unused</tt></p>
385
386</div>
387
388<!-- _______________________________________________________________________ -->
389<div class="doc_subsubsection">
390 <a name="ci_portable_code">Write Portable Code</a>
391</div>
392
393<div class="doc_text">
394
395<p>In almost all cases, it is possible and within reason to write completely
396portable code. If there are cases where it isn't possible to write portable
397code, isolate it behind a well defined (and well documented) interface.</p>
398
399<p>In practice, this means that you shouldn't assume much about the host
400compiler, including its support for "high tech" features like partial
Chris Lattnerafec0252009-03-23 04:52:53 +0000[diff] [blame]401specialization of templates. If these features are used, they should only be
402an implementation detail of a library which has a simple exposed API.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]403
404</div>
405
406<!-- _______________________________________________________________________ -->
407<div class="doc_subsubsection">
408<a name="ci_class_struct">Use of <tt>class</tt> and <tt>struct</tt> Keywords</a>
409</div>
410<div class="doc_text">
411
412<p>In C++, the <tt>class</tt> and <tt>struct</tt> keywords can be used almost
413interchangeably. The only difference is when they are used to declare a class:
414<tt>class</tt> makes all members private by default while <tt>struct</tt> makes
415all members public by default.</p>
416
417<p>Unfortunately, not all compilers follow the rules and some will generate
418different symbols based on whether <tt>class</tt> or <tt>struct</tt> was used to
419declare the symbol. This can lead to problems at link time.</p>
420
421<p>So, the rule for LLVM is to always use the <tt>class</tt> keyword, unless
422<b>all</b> members are public, in which case <tt>struct</tt> is allowed.</p>
423
424</div>
425
426<!-- *********************************************************************** -->
427<div class="doc_section">
428 <a name="styleissues">Style Issues</a>
429</div>
430<!-- *********************************************************************** -->
431
432
433<!-- ======================================================================= -->
434<div class="doc_subsection">
435 <a name="macro">The High Level Issues</a>
436</div>
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]437<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]438
439
440<!-- _______________________________________________________________________ -->
441<div class="doc_subsubsection">
442 <a name="hl_module">A Public Header File <b>is</b> a Module</a>
443</div>
444
445<div class="doc_text">
446
447<p>C++ doesn't do too well in the modularity department. There is no real
448encapsulation or data hiding (unless you use expensive protocol classes), but it
449is what we have to work with. When you write a public header file (in the LLVM
450source tree, they live in the top level "include" directory), you are defining a
451module of functionality.</p>
452
453<p>Ideally, modules should be completely independent of each other, and their
454header files should only include the absolute minimum number of headers
455possible. A module is not just a class, a function, or a namespace: <a
456href="http://www.cuj.com/articles/2000/0002/0002c/0002c.htm">it's a collection
457of these</a> that defines an interface. This interface may be several
458functions, classes or data structures, but the important issue is how they work
459together.</p>
460
461<p>In general, a module should be implemented with one or more <tt>.cpp</tt>
462files. Each of these <tt>.cpp</tt> files should include the header that defines
463their interface first. This ensure that all of the dependences of the module
464header have been properly added to the module header itself, and are not
465implicit. System headers should be included after user headers for a
466translation unit.</p>
467
468</div>
469
470<!-- _______________________________________________________________________ -->
471<div class="doc_subsubsection">
472 <a name="hl_dontinclude"><tt>#include</tt> as Little as Possible</a>
473</div>
474
475<div class="doc_text">
476
477<p><tt>#include</tt> hurts compile time performance. Don't do it unless you
478have to, especially in header files.</p>
479
480<p>But wait, sometimes you need to have the definition of a class to use it, or
481to inherit from it. In these cases go ahead and <tt>#include</tt> that header
482file. Be aware however that there are many cases where you don't need to have
483the full definition of a class. If you are using a pointer or reference to a
484class, you don't need the header file. If you are simply returning a class
485instance from a prototyped function or method, you don't need it. In fact, for
486most cases, you simply don't need the definition of a class... and not
487<tt>#include</tt>'ing speeds up compilation.</p>
488
489<p>It is easy to try to go too overboard on this recommendation, however. You
490<b>must</b> include all of the header files that you are using -- you can
491include them either directly
492or indirectly (through another header file). To make sure that you don't
493accidently forget to include a header file in your module header, make sure to
494include your module header <b>first</b> in the implementation file (as mentioned
495above). This way there won't be any hidden dependencies that you'll find out
496about later...</p>
497
498</div>
499
500<!-- _______________________________________________________________________ -->
501<div class="doc_subsubsection">
502 <a name="hl_privateheaders">Keep "internal" Headers Private</a>
503</div>
504
505<div class="doc_text">
506
507<p>Many modules have a complex implementation that causes them to use more than
508one implementation (<tt>.cpp</tt>) file. It is often tempting to put the
509internal communication interface (helper classes, extra functions, etc) in the
510public module header file. Don't do this.</p>
511
512<p>If you really need to do something like this, put a private header file in
513the same directory as the source files, and include it locally. This ensures
514that your private interface remains private and undisturbed by outsiders.</p>
515
516<p>Note however, that it's okay to put extra implementation methods a public
517class itself... just make them private (or protected), and all is well.</p>
518
519</div>
520
521<!-- _______________________________________________________________________ -->
522<div class="doc_subsubsection">
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]523 <a name="hl_earlyexit">Use Early Exits and 'continue' to Simplify Code</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]524</div>
525
526<div class="doc_text">
527
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]528<p>When reading code, keep in mind how much state and how many previous
529decisions have to be remembered by the reader to understand a block of code.
530Aim to reduce indentation where possible when it doesn't make it more difficult
531to understand the code. One great way to do this is by making use of early
532exits and the 'continue' keyword in long loops. As an example of using an early
533exit from a function, consider this "bad" code:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]534
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]535<div class="doc_code">
536<pre>
537Value *DoSomething(Instruction *I) {
538 if (!isa&lt;TerminatorInst&gt;(I) &amp;&amp;
539 I-&gt;hasOneUse() &amp;&amp; SomeOtherThing(I)) {
540 ... some long code ....
541 }
542
543 return 0;
544}
545</pre>
546</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]547
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]548<p>This code has several problems if the body of the 'if' is large. When you're
549looking at the top of the function, it isn't immediately clear that this
550<em>only</em> does interesting things with non-terminator instructions, and only
551applies to things with the other predicates. Second, it is relatively difficult
552to describe (in comments) why these predicates are important because the if
553statement makes it difficult to lay out the comments. Third, when you're deep
554within the body of the code, it is indented an extra level. Finally, when
555reading the top of the function, it isn't clear what the result is if the
556predicate isn't true, you have to read to the end of the function to know that
557it returns null.</p>
Matthijs Kooijmanc4c0af22008-07-30 12:14:10 +0000[diff] [blame]558
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]559<p>It is much preferred to format the code like this:</p>
560
561<div class="doc_code">
562<pre>
563Value *DoSomething(Instruction *I) {
564 // Terminators never need 'something' done to them because, ...
565 if (isa&lt;TerminatorInst&gt;(I))
566 return 0;
567
568 // We conservatively avoid transforming instructions with multiple uses
569 // because goats like cheese.
570 if (!I-&gt;hasOneUse())
571 return 0;
572
573 // This is really just here for example.
574 if (!SomeOtherThing(I))
575 return 0;
576
577 ... some long code ....
578}
579</pre>
580</div>
581
582<p>This fixes these problems. A similar problem frequently happens in for
583loops. A silly example is something like this:</p>
584
585<div class="doc_code">
586<pre>
587 for (BasicBlock::iterator II = BB-&gt;begin(), E = BB-&gt;end(); II != E; ++II) {
588 if (BinaryOperator *BO = dyn_cast&lt;BinaryOperator&gt;(II)) {
589 Value *LHS = BO-&gt;getOperand(0);
590 Value *RHS = BO-&gt;getOperand(1);
591 if (LHS != RHS) {
592 ...
593 }
594 }
595 }
596</pre>
597</div>
598
599<p>When you have very very small loops, this sort of structure is fine, but if
600it exceeds more than 10-15 lines, it becomes difficult for people to read and
601understand at a glance.
602The problem with this sort of code is that it gets very nested very quickly,
603meaning that the reader of the code has to keep a lot of context in their brain
604to remember what is going immediately on in the loop, because they don't know
605if/when the if conditions will have elses etc. It is strongly preferred to
606structure the loop like this:</p>
607
608<div class="doc_code">
609<pre>
610 for (BasicBlock::iterator II = BB-&gt;begin(), E = BB-&gt;end(); II != E; ++II) {
611 BinaryOperator *BO = dyn_cast&lt;BinaryOperator&gt;(II);
612 if (!BO) continue;
613
614 Value *LHS = BO-&gt;getOperand(0);
615 Value *RHS = BO-&gt;getOperand(1);
616 if (LHS == RHS) continue;
617 }
618</pre>
619</div>
620
621<p>This has all the benefits of using early exits from functions: it reduces
622nesting of the loop, it makes it easier to describe why the conditions are true,
623and it makes it obvious to the reader that there is no "else" coming up that
624they have to push context into their brain for. If a loop is large, this can
625be a big understandability win.</p>
626
627</div>
628
Chris Lattner598fcfd2009-07-28 22:54:04 +0000[diff] [blame]629<!-- _______________________________________________________________________ -->
630<div class="doc_subsubsection">
631 <a name="hl_else_after_return">Don't use "else" after a return</a>
632</div>
633
634<div class="doc_text">
635
636<p>For similar reasons above (reduction of indentation and easier reading),
637 please do not use "else" or "else if" after something that interrupts
638 control flow like return, break, continue, goto, etc. For example, this is
639 "bad":</p>
640
641<div class="doc_code">
642<pre>
643 case 'J': {
644 if (Signed) {
645 Type = Context.getsigjmp_bufType();
646 if (Type.isNull()) {
647 Error = ASTContext::GE_Missing_sigjmp_buf;
648 return QualType();
649 } else {
650 break;
651 }
652 } else {
653 Type = Context.getjmp_bufType();
654 if (Type.isNull()) {
655 Error = ASTContext::GE_Missing_jmp_buf;
656 return QualType();
657 } else {
658 break;
659 }
660 }
661 }
662 }
663</pre>
664</div>
665
666<p>It is better to write this something like:</p>
667
668<div class="doc_code">
669<pre>
670 case 'J':
671 if (Signed) {
672 Type = Context.getsigjmp_bufType();
673 if (Type.isNull()) {
674 Error = ASTContext::GE_Missing_sigjmp_buf;
675 return QualType();
676 }
677 } else {
678 Type = Context.getjmp_bufType();
679 if (Type.isNull()) {
680 Error = ASTContext::GE_Missing_jmp_buf;
681 return QualType();
682 }
683 }
684 break;
685</pre>
686</div>
687
688<p>Or better yet (in this case), as:</p>
689
690<div class="doc_code">
691<pre>
692 case 'J':
693 if (Signed)
694 Type = Context.getsigjmp_bufType();
695 else
696 Type = Context.getjmp_bufType();
697
698 if (Type.isNull()) {
699 Error = Signed ? ASTContext::GE_Missing_sigjmp_buf :
700 ASTContext::GE_Missing_jmp_buf;
701 return QualType();
702 }
703 break;
704</pre>
705</div>
706
707<p>The idea is to reduce indentation and the amount of code you have to keep
708 track of when reading the code.</p>
709
710</div>
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]711
712<!-- _______________________________________________________________________ -->
713<div class="doc_subsubsection">
714 <a name="hl_predicateloops">Turn Predicate Loops into Predicate Functions</a>
715</div>
716
717<div class="doc_text">
718
Chris Lattner5af04d22009-07-22 16:54:14 +0000[diff] [blame]719<p>It is very common to write small loops that just compute a boolean
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]720 value. There are a number of ways that people commonly write these, but an
721 example of this sort of thing is:</p>
722
723<div class="doc_code">
724<pre>
725 <b>bool FoundFoo = false;</b>
726 for (unsigned i = 0, e = BarList.size(); i != e; ++i)
727 if (BarList[i]-&gt;isFoo()) {
728 <b>FoundFoo = true;</b>
729 break;
730 }
731
732 <b>if (FoundFoo) {</b>
733 ...
734 }
735</pre>
736</div>
737
738<p>This sort of code is awkward to write, and is almost always a bad sign.
739Instead of this sort of loop, we strongly prefer to use a predicate function
740(which may be <a href="#micro_anonns">static</a>) that uses
Chris Lattner5af04d22009-07-22 16:54:14 +0000[diff] [blame]741<a href="#hl_earlyexit">early exits</a> to compute the predicate. We prefer
742the code to be structured like this:
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]743</p>
744
745
746<div class="doc_code">
747<pre>
748/// ListContainsFoo - Return true if the specified list has an element that is
749/// a foo.
750static bool ListContainsFoo(const std::vector&lt;Bar*&gt; &amp;List) {
751 for (unsigned i = 0, e = List.size(); i != e; ++i)
752 if (List[i]-&gt;isFoo())
753 return true;
754 return false;
755}
756...
757
758 <b>if (ListContainsFoo(BarList)) {</b>
759 ...
760 }
761</pre>
762</div>
763
764<p>There are many reasons for doing this: it reduces indentation and factors out
765code which can often be shared by other code that checks for the same predicate.
766More importantly, it <em>forces you to pick a name</em> for the function, and
767forces you to write a comment for it. In this silly example, this doesn't add
768much value. However, if the condition is complex, this can make it a lot easier
769for the reader to understand the code that queries for this predicate. Instead
Chris Lattner6ead3272009-07-22 16:30:39 +0000[diff] [blame]770of 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]771contains a foo, we can trust the function name and continue reading with better
772locality.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]773
774</div>
775
776
777<!-- ======================================================================= -->
778<div class="doc_subsection">
779 <a name="micro">The Low Level Issues</a>
780</div>
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]781<!-- ======================================================================= -->
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]782
783
784<!-- _______________________________________________________________________ -->
785<div class="doc_subsubsection">
786 <a name="ll_assert">Assert Liberally</a>
787</div>
788
789<div class="doc_text">
790
791<p>Use the "<tt>assert</tt>" function to its fullest. Check all of your
792preconditions and assumptions, you never know when a bug (not neccesarily even
793yours) might be caught early by an assertion, which reduces debugging time
794dramatically. The "<tt>&lt;cassert&gt;</tt>" header file is probably already
795included by the header files you are using, so it doesn't cost anything to use
796it.</p>
797
798<p>To further assist with debugging, make sure to put some kind of error message
799in the assertion statement (which is printed if the assertion is tripped). This
800helps the poor debugging make sense of why an assertion is being made and
801enforced, and hopefully what to do about it. Here is one complete example:</p>
802
803<div class="doc_code">
804<pre>
805inline Value *getOperand(unsigned i) {
806 assert(i &lt; Operands.size() &amp;&amp; "getOperand() out of range!");
807 return Operands[i];
808}
809</pre>
810</div>
811
812<p>Here are some examples:</p>
813
814<div class="doc_code">
815<pre>
816assert(Ty-&gt;isPointerType() &amp;&amp; "Can't allocate a non pointer type!");
817
818assert((Opcode == Shl || Opcode == Shr) &amp;&amp; "ShiftInst Opcode invalid!");
819
820assert(idx &lt; getNumSuccessors() &amp;&amp; "Successor # out of range!");
821
822assert(V1.getType() == V2.getType() &amp;&amp; "Constant types must be identical!");
823
824assert(isa&lt;PHINode&gt;(Succ-&gt;front()) &amp;&amp; "Only works on PHId BBs!");
825</pre>
826</div>
827
828<p>You get the idea...</p>
829
Nick Lewycky009216a2008-05-31 23:54:55 +0000[diff] [blame]830<p>Please be aware when adding assert statements that not all compilers are aware of
831the semantics of the assert. In some places, asserts are used to indicate a piece of
832code that should not be reached. These are typically of the form:</p>
833
834<div class="doc_code">
835<pre>
Dan Gohman4dfac702008-11-24 17:18:39 +0000[diff] [blame]836assert(0 &amp;&amp; "Some helpful error message");
Nick Lewycky009216a2008-05-31 23:54:55 +0000[diff] [blame]837</pre>
838</div>
839
840<p>When used in a function that returns a value, they should be followed with a return
841statement and a comment indicating that this line is never reached. This will prevent
842a compiler which is unable to deduce that the assert statement never returns from
843generating a warning.</p>
844
845<div class="doc_code">
846<pre>
Dan Gohman4dfac702008-11-24 17:18:39 +0000[diff] [blame]847assert(0 &amp;&amp; "Some helpful error message");
Nick Lewycky009216a2008-05-31 23:54:55 +0000[diff] [blame]848// Not reached
849return 0;
850</pre>
851</div>
852
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]853</div>
854
855<!-- _______________________________________________________________________ -->
856<div class="doc_subsubsection">
857 <a name="ll_ns_std">Do not use '<tt>using namespace std</tt>'</a>
858</div>
859
860<div class="doc_text">
861<p>In LLVM, we prefer to explicitly prefix all identifiers from the standard
862namespace with an "<tt>std::</tt>" prefix, rather than rely on
863"<tt>using namespace std;</tt>".</p>
864
865<p> In header files, adding a '<tt>using namespace XXX</tt>' directive pollutes
Misha Brukmanbcc53732008-12-11 19:37:04 +0000[diff] [blame]866the namespace of any source file that <tt>#include</tt>s the header. This is
867clearly a bad thing.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]868
869<p>In implementation files (e.g. .cpp files), the rule is more of a stylistic
870rule, but is still important. Basically, using explicit namespace prefixes
871makes the code <b>clearer</b>, because it is immediately obvious what facilities
872are being used and where they are coming from, and <b>more portable</b>, because
873namespace clashes cannot occur between LLVM code and other namespaces. The
874portability rule is important because different standard library implementations
875expose different symbols (potentially ones they shouldn't), and future revisions
876to the C++ standard will add more symbols to the <tt>std</tt> namespace. As
877such, we never use '<tt>using namespace std;</tt>' in LLVM.</p>
878
879<p>The exception to the general rule (i.e. it's not an exception for
880the <tt>std</tt> namespace) is for implementation files. For example, all of
881the code in the LLVM project implements code that lives in the 'llvm' namespace.
882As such, it is ok, and actually clearer, for the .cpp files to have a '<tt>using
883namespace llvm</tt>' directive at their top, after the <tt>#include</tt>s. The
884general form of this rule is that any .cpp file that implements code in any
885namespace may use that namespace (and its parents'), but should not use any
886others.</p>
887
888</div>
889
890<!-- _______________________________________________________________________ -->
891<div class="doc_subsubsection">
892 <a name="ll_virtual_anch">Provide a virtual method anchor for classes
893 in headers</a>
894</div>
895
896<div class="doc_text">
897
898<p>If a class is defined in a header file and has a v-table (either it has
899virtual methods or it derives from classes with virtual methods), it must
900always have at least one out-of-line virtual method in the class. Without
Misha Brukmanbcc53732008-12-11 19:37:04 +0000[diff] [blame]901this, the compiler will copy the vtable and RTTI into every <tt>.o</tt> file
902that <tt>#include</tt>s the header, bloating <tt>.o</tt> file sizes and
903increasing link times.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]904
905</div>
906
Chris Lattner01b8af32009-06-30 06:13:23 +0000[diff] [blame]907<!-- _______________________________________________________________________ -->
908<div class="doc_subsubsection">
909 <a name="ll_end">Don't evaluate end() every time through a loop</a>
910</div>
911
912<div class="doc_text">
913
914<p>Because C++ doesn't have a standard "foreach" loop (though it can be emulated
915with macros and may be coming in C++'0x) we end up writing a lot of loops that
916manually iterate from begin to end on a variety of containers or through other
917data structures. One common mistake is to write a loop in this style:</p>
918
919<div class="doc_code">
920<pre>
921 BasicBlock *BB = ...
922 for (BasicBlock::iterator I = BB->begin(); I != <b>BB->end()</b>; ++I)
923 ... use I ...
924</pre>
925</div>
926
927<p>The problem with this construct is that it evaluates "<tt>BB->end()</tt>"
928every time through the loop. Instead of writing the loop like this, we strongly
929prefer loops to be written so that they evaluate it once before the loop starts.
930A convenient way to do this is like so:</p>
931
932<div class="doc_code">
933<pre>
934 BasicBlock *BB = ...
935 for (BasicBlock::iterator I = BB->begin(), E = <b>BB->end()</b>; I != E; ++I)
936 ... use I ...
937</pre>
938</div>
939
940<p>The observant may quickly point out that these two loops may have different
941semantics: if the container (a basic block in this case) is being mutated, then
942"<tt>BB->end()</tt>" may change its value every time through the loop and the
943second loop may not in fact be correct. If you actually do depend on this
Chris Lattnerdc43b342009-06-30 06:20:03 +0000[diff] [blame]944behavior, please write the loop in the first form and add a comment indicating
Chris Lattner01b8af32009-06-30 06:13:23 +0000[diff] [blame]945that you did it intentionally.</p>
946
947<p>Why do we prefer the second form (when correct)? Writing the loop in the
948first form has two problems: First it may be less efficient than evaluating it
949at the start of the loop. In this case, the cost is probably minor: a few extra
950loads every time through the loop. However, if the base expression is more
951complex, then the cost can rise quickly. I've seen loops where the end
952expression was actually something like: "<tt>SomeMap[x]->end()</tt>" and map
Chris Lattner088affa2009-06-30 06:27:54 +0000[diff] [blame]953lookups really aren't cheap. By writing it in the second form consistently, you
Chris Lattner01b8af32009-06-30 06:13:23 +0000[diff] [blame]954eliminate the issue entirely and don't even have to think about it.</p>
955
Chris Lattnerdc43b342009-06-30 06:20:03 +0000[diff] [blame]956<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]957hints to the reader that the loop is mutating the container (a fact that a
958comment would handily confirm!). If you write the loop in the second form, it
959is immediately obvious without even looking at the body of the loop that the
960container isn't being modified, which makes it easier to read the code and
961understand what it does.</p>
962
963<p>While the second form of the loop is a few extra keystrokes, we do strongly
964prefer it.</p>
965
966</div>
967
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]968<!-- _______________________________________________________________________ -->
969<div class="doc_subsubsection">
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]970 <a name="ll_iostream"><tt>#include &lt;iostream&gt;</tt> is forbidden</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]971</div>
972
973<div class="doc_text">
974
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]975<p>The use of <tt>#include &lt;iostream&gt;</tt> in library files is
976hereby <b><em>forbidden</em></b>. The primary reason for doing this is to
977support clients using LLVM libraries as part of larger systems. In particular,
978we statically link LLVM into some dynamic libraries. Even if LLVM isn't used,
979the static c'tors are run whenever an application start up that uses the dynamic
980library. There are two problems with this:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]981
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]982<ol>
983 <li>The time to run the static c'tors impacts startup time of
984 applications&mdash;a critical time for GUI apps.</li>
985 <li>The static c'tors cause the app to pull many extra pages of memory off the
986 disk: both the code for the static c'tors in each <tt>.o</tt> file and the
987 small amount of data that gets touched. In addition, touched/dirty pages
988 put more pressure on the VM system on low-memory machines.</li>
989</ol>
990
991<p>Note that using the other stream headers (<tt>&lt;sstream&gt;</tt> for
992example) is allowed normally, it is just <tt>&lt;iostream&gt;</tt> that is
993causing problems.</p>
994
Daniel Dunbar2f099eb2009-07-24 23:04:51 +0000[diff] [blame]995<p>In addition, new code should always
996use <a href="#ll_raw_ostream"><tt>raw_ostream</tt></a> or
997the <tt>llvm::MemoryBuffer</tt> API (for reading in files).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]998
999</div>
1000
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]1001
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1002<!-- _______________________________________________________________________ -->
1003<div class="doc_subsubsection">
1004 <a name="ll_avoidendl">Avoid <tt>std::endl</tt></a>
1005</div>
1006
1007<div class="doc_text">
1008
1009<p>The <tt>std::endl</tt> modifier, when used with iostreams outputs a newline
1010to the output stream specified. In addition to doing this, however, it also
1011flushes the output stream. In other words, these are equivalent:</p>
1012
1013<div class="doc_code">
1014<pre>
1015std::cout &lt;&lt; std::endl;
1016std::cout &lt;&lt; '\n' &lt;&lt; std::flush;
1017</pre>
1018</div>
1019
1020<p>Most of the time, you probably have no reason to flush the output stream, so
1021it's better to use a literal <tt>'\n'</tt>.</p>
1022
1023</div>
1024
1025
Daniel Dunbar2f099eb2009-07-24 23:04:51 +0000[diff] [blame]1026<!-- _______________________________________________________________________ -->
1027<div class="doc_subsubsection">
1028 <a name="ll_raw_ostream">Use <tt>raw_ostream</tt></a>
1029</div>
1030
1031<div class="doc_text">
1032
1033<p>LLVM includes a lightweight, simple, and efficient stream implementation
1034in <tt>llvm/Support/raw_ostream.h</tt> which provides all of the common features
Daniel Dunbared0da882009-07-24 23:54:34 +0000[diff] [blame]1035of <tt>std::ostream</tt>. All new code should use <tt>raw_ostream</tt> instead
1036of <tt>ostream</tt>.</p>
Daniel Dunbar2f099eb2009-07-24 23:04:51 +0000[diff] [blame]1037
Daniel Dunbared0da882009-07-24 23:54:34 +0000[diff] [blame]1038<p>Unlike <tt>std::ostream</tt>, <tt>raw_ostream</tt> is not a template and can
Daniel Dunbar2f099eb2009-07-24 23:04:51 +0000[diff] [blame]1039be forward declared as <tt>class raw_ostream</tt>. Public headers should
1040generally not include the <tt>raw_ostream</tt> header, but use forward
1041declarations and constant references to <tt>raw_ostream</tt> instances.</p>
1042
1043</div>
1044
1045
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]1046<!-- ======================================================================= -->
1047<div class="doc_subsection">
1048 <a name="nano">Microscopic Details</a>
1049</div>
1050<!-- ======================================================================= -->
1051
1052<p>This section describes preferred low-level formatting guidelines along with
1053reasoning on why we prefer them.</p>
1054
1055<!-- _______________________________________________________________________ -->
1056<div class="doc_subsubsection">
1057 <a name="micro_spaceparen">Spaces Before Parentheses</a>
1058</div>
1059
1060<div class="doc_text">
1061
1062<p>We prefer to put a space before a parentheses only in control flow
1063statements, but not in normal function call expressions and function-like
1064macros. For example, this is good:</p>
1065
1066<div class="doc_code">
1067<pre>
1068 <b>if (</b>x) ...
1069 <b>for (</b>i = 0; i != 100; ++i) ...
1070 <b>while (</b>llvm_rocks) ...
1071
1072 <b>somefunc(</b>42);
1073 <b><a href="#ll_assert">assert</a>(</b>3 != 4 &amp;&amp; "laws of math are failing me");
1074
1075 a = <b>foo(</b>42, 92) + <b>bar(</b>x);
1076 </pre>
1077</div>
1078
1079<p>... and this is bad:</p>
1080
1081<div class="doc_code">
1082<pre>
1083 <b>if(</b>x) ...
1084 <b>for(</b>i = 0; i != 100; ++i) ...
1085 <b>while(</b>llvm_rocks) ...
1086
1087 <b>somefunc (</b>42);
1088 <b><a href="#ll_assert">assert</a> (</b>3 != 4 &amp;&amp; "laws of math are failing me");
1089
1090 a = <b>foo (</b>42, 92) + <b>bar (</b>x);
1091</pre>
1092</div>
1093
1094<p>The reason for doing this is not completely arbitrary. This style makes
1095 control flow operators stand out more, and makes expressions flow better. The
1096 function call operator binds very tightly as a postfix operator. Putting
1097 a space after a function name (as in the last example) makes it appear that
1098 the code might bind the arguments of the left-hand-side of a binary operator
1099 with the argument list of a function and the name of the right side. More
1100 specifically, it is easy to misread the "a" example as:</p>
1101
1102<div class="doc_code">
1103<pre>
1104 a = foo <b>(</b>(42, 92) + bar<b>)</b> (x);
1105</pre>
1106</div>
1107
1108<p>... when skimming through the code. By avoiding a space in a function, we
1109avoid this misinterpretation.</p>
1110
1111</div>
1112
1113<!-- _______________________________________________________________________ -->
1114<div class="doc_subsubsection">
1115 <a name="micro_preincrement">Prefer Preincrement</a>
1116</div>
1117
1118<div class="doc_text">
1119
1120<p>Hard fast rule: Preincrement (<tt>++X</tt>) may be no slower than
1121postincrement (<tt>X++</tt>) and could very well be a lot faster than it. Use
1122preincrementation whenever possible.</p>
1123
1124<p>The semantics of postincrement include making a copy of the value being
1125incremented, returning it, and then preincrementing the "work value". For
1126primitive types, this isn't a big deal... but for iterators, it can be a huge
1127issue (for example, some iterators contains stack and set objects in them...
1128copying an iterator could invoke the copy ctor's of these as well). In general,
1129get in the habit of always using preincrement, and you won't have a problem.</p>
1130
1131</div>
1132
1133<!-- _______________________________________________________________________ -->
1134<div class="doc_subsubsection">
1135 <a name="micro_namespaceindent">Namespace Indentation</a>
1136</div>
1137
1138<div class="doc_text">
1139
1140<p>
1141In general, we strive to reduce indentation where ever possible. This is useful
1142because we want code to <a href="#scf_codewidth">fit into 80 columns</a> without
1143wrapping horribly, but also because it makes it easier to understand the code.
1144Namespaces are a funny thing: they are often large, and we often desire to put
1145lots of stuff into them (so they can be large). Other times they are tiny,
1146because they just hold an enum or something similar. In order to balance this,
1147we use different approaches for small versus large namespaces.
1148</p>
1149
1150<p>
1151If a namespace definition is small and <em>easily</em> fits on a screen (say,
1152less than 35 lines of code), then you should indent its body. Here's an
1153example:
1154</p>
1155
1156<div class="doc_code">
1157<pre>
Chris Lattner5af04d22009-07-22 16:54:14 +0000[diff] [blame]1158namespace llvm {
1159 namespace X86 {
1160 /// RelocationType - An enum for the x86 relocation codes. Note that
1161 /// the terminology here doesn't follow x86 convention - word means
1162 /// 32-bit and dword means 64-bit.
1163 enum RelocationType {
1164 /// reloc_pcrel_word - PC relative relocation, add the relocated value to
1165 /// the value already in memory, after we adjust it for where the PC is.
1166 reloc_pcrel_word = 0,
1167
1168 /// reloc_picrel_word - PIC base relative relocation, add the relocated
1169 /// value to the value already in memory, after we adjust it for where the
1170 /// PIC base is.
1171 reloc_picrel_word = 1,
1172
1173 /// reloc_absolute_word, reloc_absolute_dword - Absolute relocation, just
1174 /// add the relocated value to the value already in memory.
1175 reloc_absolute_word = 2,
1176 reloc_absolute_dword = 3
1177 };
1178 }
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]1179}
1180</pre>
1181</div>
1182
1183<p>Since the body is small, indenting adds value because it makes it very clear
1184where the namespace starts and ends, and it is easy to take the whole thing in
1185in one "gulp" when reading the code. If the blob of code in the namespace is
1186larger (as it typically is in a header in the llvm or clang namespaces), do not
1187indent the code, and add a comment indicating what namespace is being closed.
1188For example:</p>
1189
1190<div class="doc_code">
1191<pre>
1192namespace llvm {
1193namespace knowledge {
1194
1195/// Grokable - This class represents things that Smith can have an intimate
1196/// understanding of and contains the data associated with it.
1197class Grokable {
1198...
1199public:
1200 explicit Grokable() { ... }
1201 virtual ~Grokable() = 0;
1202
1203 ...
1204
1205};
1206
1207} // end namespace knowledge
1208} // end namespace llvm
1209</pre>
1210</div>
1211
1212<p>Because the class is large, we don't expect that the reader can easily
1213understand the entire concept in a glance, and the end of the file (where the
1214namespaces end) may be a long ways away from the place they open. As such,
1215indenting the contents of the namespace doesn't add any value, and detracts from
1216the readability of the class. In these cases it is best to <em>not</em> indent
1217the contents of the namespace.</p>
1218
1219</div>
1220
1221<!-- _______________________________________________________________________ -->
1222<div class="doc_subsubsection">
1223 <a name="micro_anonns">Anonymous Namespaces</a>
1224</div>
1225
1226<div class="doc_text">
1227
Chris Lattner5af04d22009-07-22 16:54:14 +0000[diff] [blame]1228<p>After talking about namespaces in general, you may be wondering about
1229anonymous namespaces in particular.
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]1230Anonymous namespaces are a great language feature that tells the C++ compiler
1231that the contents of the namespace are only visible within the current
1232translation unit, allowing more aggressive optimization and eliminating the
1233possibility of symbol name collisions. Anonymous namespaces are to C++ as
1234"static" is to C functions and global variables. While "static" is available
1235in C++, anonymous namespaces are more general: they can make entire classes
1236private to a file.</p>
1237
1238<p>The problem with anonymous namespaces is that they naturally want to
1239encourage indentation of their body, and they reduce locality of reference: if
1240you see a random function definition in a C++ file, it is easy to see if it is
1241marked static, but seeing if it is in an anonymous namespace requires scanning
1242a big chunk of the file.</p>
1243
1244<p>Because of this, we have a simple guideline: make anonymous namespaces as
1245small as possible, and only use them for class declarations. For example, this
1246is good:</p>
1247
1248<div class="doc_code">
1249<pre>
1250<b>namespace {</b>
1251 class StringSort {
1252 ...
1253 public:
1254 StringSort(...)
1255 bool operator&lt;(const char *RHS) const;
1256 };
1257<b>} // end anonymous namespace</b>
1258
1259static void Helper() {
1260 ...
1261}
1262
1263bool StringSort::operator&lt;(const char *RHS) const {
1264 ...
1265}
1266
1267</pre>
1268</div>
1269
1270<p>This is bad:</p>
1271
1272
1273<div class="doc_code">
1274<pre>
1275<b>namespace {</b>
1276class StringSort {
1277...
1278public:
1279 StringSort(...)
1280 bool operator&lt;(const char *RHS) const;
1281};
1282
1283void Helper() {
1284 ...
1285}
1286
1287bool StringSort::operator&lt;(const char *RHS) const {
1288 ...
1289}
1290
1291<b>} // end anonymous namespace</b>
1292
1293</pre>
1294</div>
1295
1296
1297<p>This is bad specifically because if you're looking at "Helper" in the middle
1298of a large C++ file, that you have no immediate way to tell if it is local to
1299the file. When it is marked static explicitly, this is immediately obvious.
1300Also, there is no reason to enclose the definition of "operator&lt;" in the
Chris Lattner5af04d22009-07-22 16:54:14 +0000[diff] [blame]1301namespace just because it was declared there.
Chris Lattnereb1e8c32009-07-22 05:40:54 +0000[diff] [blame]1302</p>
1303
1304</div>
1305
1306
1307
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1308<!-- *********************************************************************** -->
1309<div class="doc_section">
1310 <a name="seealso">See Also</a>
1311</div>
1312<!-- *********************************************************************** -->
1313
1314<div class="doc_text">
1315
1316<p>A lot of these comments and recommendations have been culled for other
1317sources. Two particularly important books for our work are:</p>
1318
1319<ol>
1320
Chris Lattnerb5955b72007-11-09 21:49:08 +0000[diff] [blame]1321<li><a href="http://www.amazon.com/Effective-Specific-Addison-Wesley-Professional-Computing/dp/0321334876">Effective
1322C++</a> by Scott Meyers. Also
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1323interesting and useful are "More Effective C++" and "Effective STL" by the same
1324author.</li>
1325
Chris Lattnerb5955b72007-11-09 21:49:08 +0000[diff] [blame]1326<li>Large-Scale C++ Software Design by John Lakos</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1327
1328</ol>
1329
1330<p>If you get some free time, and you haven't read them: do so, you might learn
1331something.</p>
1332
1333</div>
1334
1335<!-- *********************************************************************** -->
1336
1337<hr>
1338<address>
1339 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman947321d2008-12-11 17:34:48 +0000[diff] [blame]1340 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]1341 <a href="http://validator.w3.org/check/referer"><img
Misha Brukman947321d2008-12-11 17:34:48 +0000[diff] [blame]1342 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]1343
1344 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
1345 <a href="http://llvm.org">LLVM Compiler Infrastructure</a><br>
1346 Last modified: $Date$
1347</address>
1348
1349</body>
1350</html>