| blob | b5cf508d2f85e824d528a1474c7ba0eb8b0ba088 |
3 <html>
4 <head>
7 </head>
8 <body>
11 LLVM Coding Standards
12 </div>
14 <ol>
17 <ol>
19 <ol>
26 </ol></li>
28 <ol>
30 Errors</a></li>
34 </ol></li>
35 </ol></li>
37 <ol>
39 <ol>
41 Module</a></li>
44 Private</a></li>
46 Code</a></li>
50 Functions</a></li>
51 </ol></li>
53 <ol>
58 classes in headers</a></li>
60 loop</a></li>
65 </ol></li>
68 <ol>
73 </ol></li>
76 </ol></li>
78 </ol>
82 </div>
85 <!-- *********************************************************************** -->
88 </div>
89 <!-- *********************************************************************** -->
93 <p>This document attempts to describe a few coding standards that are being used
94 in the LLVM source tree. Although no coding standards should be regarded as
95 absolute requirements to be followed in all instances, coding standards can be
96 useful.</p>
98 <p>This document intentionally does not prescribe fixed standards for religious
99 issues such as brace placement and space usage. For issues like this, follow
100 the golden rule:</p>
102 <blockquote>
105 project, feel free to use whatever style you are most comfortable with. If you
106 are extending, enhancing, or bug fixing already implemented code, use the style
107 that is already being used so that the source is uniform and easy to
108 follow.</a></b></p>
110 </blockquote>
112 <p>The ultimate goal of these guidelines is the increase readability and
113 maintainability of our common source base. If you have suggestions for topics to
114 be included, please mail them to <a
117 </div>
119 <!-- *********************************************************************** -->
122 </div>
123 <!-- *********************************************************************** -->
125 <!-- ======================================================================= -->
128 </div>
130 <!-- _______________________________________________________________________ -->
133 </div>
137 <p>Comments are one critical part of readability and maintainability. Everyone
138 knows they should comment, so should you. When writing comments, write them as
139 English prose, which means they should use proper capitalization, punctuation,
140 etc. Although we all should probably
141 comment our code more than we do, there are a few very critical places that
142 documentation is very useful:</p>
146 <p>Every source file should have a header on it that describes the basic
147 purpose of the file. If a file does not have a header, it should not be
148 checked into Subversion. Most source trees will probably have a standard
149 file header format. The standard format for the LLVM source tree looks like
150 this:</p>
153 <pre>
154 //===-- llvm/Instruction.h - Instruction class definition -------*- C++ -*-===//
155 //
156 // The LLVM Compiler Infrastructure
157 //
158 // This file is distributed under the University of Illinois Open Source
159 // License. See LICENSE.TXT for details.
160 //
161 //===----------------------------------------------------------------------===//
162 //
163 // This file contains the declaration of the Instruction class, which is the
164 // base class for all of the VM instructions.
165 //
166 //===----------------------------------------------------------------------===//
167 </pre>
168 </div>
171 -*-</tt>" string on the first line is there to tell Emacs that the source file
174 on the first line, along with a very short description of the purpose of the
175 file. This is important when printing out code and flipping though lots of
176 pages.</p>
178 <p>The next section in the file is a concise note that defines the license
179 that the file is released under. This makes it perfectly clear what terms the
180 source code can be distributed under and should not be modified in any way.</p>
182 <p>The main body of the description does not have to be very long in most cases.
183 Here it's only two lines. If an algorithm is being implemented or something
184 tricky is going on, a reference to the paper where it is published should be
189 <p>Classes are one fundamental part of a good object oriented design. As such,
190 a class definition should have a comment block that explains what the class is
191 used for... if it's not obvious. If it's so completely obvious your grandma
192 could figure it out, it's probably safe to leave it out. Naming classes
193 something sane goes a long ways towards avoiding writing documentation.</p>
198 <p>Methods defined in a class (as well as any global functions) should also be
199 documented properly. A quick note about what it does and a description of the
200 borderline behaviour is all that is necessary here (unless something
201 particularly tricky or insidious is going on). The hope is that people can
202 figure out how to use your interfaces without reading the code itself... that is
203 the goal metric.</p>
205 <p>Good things to talk about here are what happens when something unexpected
206 happens: does the method return null? Abort? Format your hard disk?</p>
208 </div>
210 <!-- _______________________________________________________________________ -->
213 </div>
218 require less typing, don't have nesting problems, etc. There are a few cases
221 <ol>
222 <li>When writing C code: Obviously if you are writing C code, use C style
223 comments.</li>
225 file.</li>
226 <li>When writing a source file that is used by a tool that only accepts C
227 style comments.</li>
228 </ol>
231 These nest properly and are better behaved in general than C style comments.</p>
233 </div>
235 <!-- _______________________________________________________________________ -->
238 </div>
243 include guards if working on a header file), the <a
246 order:</p>
248 <ol>
260 </ol>
268 interfaces, we ensure that the header does not have any hidden dependencies
269 which are not explicitly #included in the header, but should be. It is also a
271 implements are defined.</p>
273 </div>
275 <!-- _______________________________________________________________________ -->
278 </div>
283 like to print out code and look at your code in an xterm without resizing
284 it.</p>
286 <p>The longer answer is that there must be some limit to the width of the code
287 in order to reasonably allow developers to have multiple files side-by-side in
288 windows on a modest display. If you are going to pick a width limit, it is
289 somewhat arbitrary but you might as well pick something standard. Going with
291 value and would be detrimental to printing out code. Also many other projects
292 have standardized on 80 columns, so some people have already configured their
295 <p>This is one of many contentious issues in coding standards, but it is not up
296 for debate.</p>
298 </div>
300 <!-- _______________________________________________________________________ -->
303 </div>
307 <p>In all cases, prefer spaces to tabs in source files. People have different
308 preferred indentation levels, and different styles of indentation that they
309 like; this is fine. What isn't fine is that different editors/viewers expand
310 tabs out to different tab stops. This can cause your code to look completely
311 unreadable, and it is not worth dealing with.</p>
314 style of existing code if you are modifying and extending it. If you like four
316 with two spaces of indentation. Also, do not reindent a whole source file: it
317 makes for incredible diffs that are absolutely worthless.</p>
319 </div>
321 <!-- _______________________________________________________________________ -->
324 </div>
328 <p>Okay, in your first year of programming you were told that indentation is
329 important. If you didn't believe and internalize this then, now is the time.
330 Just do it.</p>
332 </div>
335 <!-- ======================================================================= -->
338 </div>
341 <!-- _______________________________________________________________________ -->
344 </div>
349 aren't casting values correctly, your have "questionable" constructs in your
350 code, or you are doing something legitimately wrong. Compiler warnings can
351 cover up legitimate errors in output and make dealing with a translation unit
352 difficult.</p>
354 <p>It is not possible to prevent all warnings from all compilers, nor is it
356 a good thorough set of warnings, and stick to it. At least in the case of
358 syntax of the code slightly. For example, a warning that annoys me occurs when
359 I write code like this:</p>
362 <pre>
363 if (V = getValue()) {
364 ...
365 }
366 </pre>
367 </div>
370 operator, and that I probably mistyped it. In most cases, I haven't, and I
371 really don't want the spurious errors. To fix this particular problem, I
372 rewrite the code like this:</p>
375 <pre>
376 if ((V = getValue())) {
377 ...
378 }
379 </pre>
380 </div>
383 be fixed by massaging the code appropriately.</p>
388 <pre>
389 -Wall -Winline -W -Wwrite-strings -Wno-unused
390 </pre>
391 </div>
393 </div>
395 <!-- _______________________________________________________________________ -->
398 </div>
402 <p>In almost all cases, it is possible and within reason to write completely
403 portable code. If there are cases where it isn't possible to write portable
404 code, isolate it behind a well defined (and well documented) interface.</p>
406 <p>In practice, this means that you shouldn't assume much about the host
407 compiler, and Visual Studio tends to be the lowest common denominator.
408 If advanced features are used, they should only be an implementation detail of
409 a library which has a simple exposed API, and preferably be buried in
410 libSystem.</p>
412 </div>
414 <!-- _______________________________________________________________________ -->
417 </div>
420 <p>In an effort to reduce code and executable size, LLVM does not use RTTI
423 causing executable bloat even if exceptions are never used in the code base, or
424 if RTTI is never used for a class. Because of this, we turn them off globally
425 in the code.</p>
427 <p>That said, LLVM does make extensive use of a hand-rolled form of RTTI that
430 opt-in and can be added to any class. It is also substantially more efficient
433 </div>
435 <!-- _______________________________________________________________________ -->
438 </div>
442 interchangeably. The only difference is when they are used to declare a class:
444 all members public by default.</p>
446 <p>Unfortunately, not all compilers follow the rules and some will generate
448 declare the symbol. This can lead to problems at link time.</p>
455 </div>
457 <!-- *********************************************************************** -->
460 </div>
461 <!-- *********************************************************************** -->
464 <!-- ======================================================================= -->
467 </div>
468 <!-- ======================================================================= -->
471 <!-- _______________________________________________________________________ -->
474 </div>
478 <p>C++ doesn't do too well in the modularity department. There is no real
479 encapsulation or data hiding (unless you use expensive protocol classes), but it
480 is what we have to work with. When you write a public header file (in the LLVM
481 source tree, they live in the top level "<tt>include</tt>" directory), you are
482 defining a module of functionality.</p>
484 <p>Ideally, modules should be completely independent of each other, and their
486 headers possible. A module is not just a class, a function, or a
488 a collection of these</a> that defines an interface. This interface may be
489 several functions, classes, or data structures, but the important issue is how
490 they work together.</p>
494 their interface first. This ensures that all of the dependences of the module
495 header have been properly added to the module header itself, and are not
496 implicit. System headers should be included after user headers for a
497 translation unit.</p>
499 </div>
501 <!-- _______________________________________________________________________ -->
504 </div>
509 have to, especially in header files.</p>
511 <p>But wait! Sometimes you need to have the definition of a class to use it, or
513 file. Be aware however that there are many cases where you don't need to have
514 the full definition of a class. If you are using a pointer or reference to a
515 class, you don't need the header file. If you are simply returning a class
516 instance from a prototyped function or method, you don't need it. In fact, for
517 most cases, you simply don't need the definition of a class. And not
520 <p>It is easy to try to go too overboard on this recommendation, however. You
522 include them either directly or indirectly (through another header file). To
523 make sure that you don't accidentally forget to include a header file in your
525 implementation file (as mentioned above). This way there won't be any hidden
526 dependencies that you'll find out about later.</p>
528 </div>
530 <!-- _______________________________________________________________________ -->
533 </div>
537 <p>Many modules have a complex implementation that causes them to use more than
539 internal communication interface (helper classes, extra functions, etc) in the
540 public module header file. Don't do this!</p>
542 <p>If you really need to do something like this, put a private header file in
543 the same directory as the source files, and include it locally. This ensures
544 that your private interface remains private and undisturbed by outsiders.</p>
546 <p>Note however, that it's okay to put extra implementation methods in a public
547 class itself. Just make them private (or protected) and all is well.</p>
549 </div>
551 <!-- _______________________________________________________________________ -->
554 </div>
558 <p>When reading code, keep in mind how much state and how many previous
559 decisions have to be remembered by the reader to understand a block of code.
560 Aim to reduce indentation where possible when it doesn't make it more difficult
561 to understand the code. One great way to do this is by making use of early
566 <pre>
567 Value *DoSomething(Instruction *I) {
570 ... some long code ....
571 }
573 return 0;
574 }
575 </pre>
576 </div>
579 When you're looking at the top of the function, it isn't immediately clear that
581 only applies to things with the other predicates. Second, it is relatively
582 difficult to describe (in comments) why these predicates are important because
584 when you're deep within the body of the code, it is indented an extra level.
585 Finally, when reading the top of the function, it isn't clear what the result is
586 if the predicate isn't true; you have to read to the end of the function to know
587 that it returns null.</p>
592 <pre>
593 Value *DoSomething(Instruction *I) {
594 // Terminators never need 'something' done to them because ...
596 return 0;
598 // We conservatively avoid transforming instructions with multiple uses
599 // because goats like cheese.
600 if (!I->hasOneUse())
601 return 0;
603 // This is really just here for example.
604 if (!SomeOtherThing(I))
605 return 0;
607 ... some long code ....
608 }
609 </pre>
610 </div>
613 loops. A silly example is something like this:</p>
616 <pre>
621 if (LHS != RHS) {
622 ...
623 }
624 }
625 }
626 </pre>
627 </div>
629 <p>When you have very, very small loops, this sort of structure is fine. But if
631 understand at a glance. The problem with this sort of code is that it gets very
632 nested very quickly. Meaning that the reader of the code has to keep a lot of
633 context in their brain to remember what is going immediately on in the loop,
635 It is strongly preferred to structure the loop like this:</p>
638 <pre>
641 if (!BO) continue;
645 if (LHS == RHS) continue;
647 ...
648 }
649 </pre>
650 </div>
652 <p>This has all the benefits of using early exits for functions: it reduces
653 nesting of the loop, it makes it easier to describe why the conditions are true,
655 that they have to push context into their brain for. If a loop is large, this
656 can be a big understandability win.</p>
658 </div>
660 <!-- _______________________________________________________________________ -->
663 </div>
667 <p>For similar reasons above (reduction of indentation and easier reading),
673 <pre>
674 case 'J': {
675 if (Signed) {
676 Type = Context.getsigjmp_bufType();
677 if (Type.isNull()) {
678 Error = ASTContext::GE_Missing_sigjmp_buf;
679 return QualType();
680 <b>} else {
681 break;
682 }</b>
683 } else {
684 Type = Context.getjmp_bufType();
685 if (Type.isNull()) {
686 Error = ASTContext::GE_Missing_jmp_buf;
687 return QualType();
688 <b>} else {
689 break;
690 }</b>
691 }
692 }
693 }
694 </pre>
695 </div>
700 <pre>
701 case 'J':
702 if (Signed) {
703 Type = Context.getsigjmp_bufType();
704 if (Type.isNull()) {
705 Error = ASTContext::GE_Missing_sigjmp_buf;
706 return QualType();
707 }
708 } else {
709 Type = Context.getjmp_bufType();
710 if (Type.isNull()) {
711 Error = ASTContext::GE_Missing_jmp_buf;
712 return QualType();
713 }
714 }
716 </pre>
717 </div>
722 <pre>
723 case 'J':
724 if (Signed)
725 Type = Context.getsigjmp_bufType();
726 else
727 Type = Context.getjmp_bufType();
729 if (Type.isNull()) {
730 Error = Signed ? ASTContext::GE_Missing_sigjmp_buf :
731 ASTContext::GE_Missing_jmp_buf;
732 return QualType();
733 }
735 </pre>
736 </div>
738 <p>The idea is to reduce indentation and the amount of code you have to keep
739 track of when reading the code.</p>
741 </div>
743 <!-- _______________________________________________________________________ -->
746 </div>
750 <p>It is very common to write small loops that just compute a boolean value.
751 There are a number of ways that people commonly write these, but an example of
752 this sort of thing is:</p>
755 <pre>
757 for (unsigned i = 0, e = BarList.size(); i != e; ++i)
758 if (BarList[i]->isFoo()) {
760 break;
761 }
764 ...
765 }
766 </pre>
767 </div>
769 <p>This sort of code is awkward to write, and is almost always a bad sign.
770 Instead of this sort of loop, we strongly prefer to use a predicate function
773 the code to be structured like this:</p>
776 <pre>
777 /// ListContainsFoo - Return true if the specified list has an element that is
778 /// a foo.
780 for (unsigned i = 0, e = List.size(); i != e; ++i)
781 if (List[i]->isFoo())
782 return true;
783 return false;
784 }
785 ...
788 ...
789 }
790 </pre>
791 </div>
793 <p>There are many reasons for doing this: it reduces indentation and factors out
794 code which can often be shared by other code that checks for the same predicate.
796 forces you to write a comment for it. In this silly example, this doesn't add
797 much value. However, if the condition is complex, this can make it a lot easier
798 for the reader to understand the code that queries for this predicate. Instead
799 of being faced with the in-line details of how we check to see if the BarList
800 contains a foo, we can trust the function name and continue reading with better
801 locality.</p>
803 </div>
806 <!-- ======================================================================= -->
809 </div>
810 <!-- ======================================================================= -->
813 <!-- _______________________________________________________________________ -->
816 </div>
820 <p>Poorly-chosen names can mislead the reader and cause bugs. We cannot stress
822 match the semantics and role of the underlying entities, within reason. Avoid
823 abbreviations unless they are well known. After picking a good name, make sure
824 to use consistent capitalization for the name, as inconsistency requires clients
825 to either memorize the APIs or to look it up to find the exact spelling.</p>
829 rules:</p>
831 <ul>
833 should be nouns and start with an upper-case letter (e.g.
837 actions), and command-like function should be imperative. The name should
842 they should follow the naming conventions for types. A common use for enums
843 is as a discriminator for a union, or an indicator of a subclass. When an
848 variables</b> should start with an upper-case letter, just like types.
849 Unless the enumerators are defined in their own small namespace or inside a
850 class, enumerators should have a prefix corresponding to the enum
853 Enumerators that are just convenience constants are exempt from the
854 requirement for a prefix. For instance:</p>
857 <pre>
858 enum {
859 MaxSize = 42,
860 Density = 12
861 };
862 </pre>
863 </div>
864 </li>
866 </ul>
868 <p>As an exception, classes that mimic STL classes can have member names in
875 <pre>
876 class VehicleMaker {
877 ...
881 // kind of factories.
882 };
884 Vehicle MakeVehicle(VehicleType Type) {
885 VehicleMaker M; // Might be OK if having a short life-span.
886 Tire tmp1 = M.makeTire(); // Bad -- 'tmp1' provides no information.
887 Light headlight = M.makeLight("head"); // Good -- descriptive.
888 ...
889 }
890 </pre>
891 </div>
893 </div>
896 <!-- _______________________________________________________________________ -->
899 </div>
904 preconditions and assumptions, you never know when a bug (not necessarily even
905 yours) might be caught early by an assertion, which reduces debugging time
906 dramatically. The "<tt><cassert></tt>" header file is probably already
907 included by the header files you are using, so it doesn't cost anything to use
908 it.</p>
910 <p>To further assist with debugging, make sure to put some kind of error message
911 in the assertion statement, which is printed if the assertion is tripped. This
912 helps the poor debugger make sense of why an assertion is being made and
913 enforced, and hopefully what to do about it. Here is one complete example:</p>
916 <pre>
917 inline Value *getOperand(unsigned i) {
919 return Operands[i];
920 }
921 </pre>
922 </div>
927 <pre>
937 </pre>
938 </div>
942 <p>Please be aware that, when adding assert statements, not all compilers are aware of
943 the semantics of the assert. In some places, asserts are used to indicate a piece of
944 code that should not be reached. These are typically of the form:</p>
947 <pre>
949 </pre>
950 </div>
952 <p>When used in a function that returns a value, they should be followed with a return
953 statement and a comment indicating that this line is never reached. This will prevent
954 a compiler which is unable to deduce that the assert statement never returns from
955 generating a warning.</p>
958 <pre>
960 // Not reached
961 return 0;
962 </pre>
963 </div>
966 value" warning when assertions are disabled. For example, this code will
967 warn:</p>
970 <pre>
971 unsigned Size = V.size();
974 bool NewToSet = Myset.insert(Value);
976 </pre>
977 </div>
979 <p>These are two interesting different cases. In the first case, the call to
980 V.size() is only useful for the assert, and we don't want it executed when
981 assertions are disabled. Code like this should move the call into the assert
982 itself. In the second case, the side effects of the call must happen whether
983 the assert is enabled or not. In this case, the value should be cast to void to
984 disable the warning. To be specific, it is preferred to write the code like
985 this:</p>
988 <pre>
991 bool NewToSet = Myset.insert(Value); (void)NewToSet;
993 </pre>
994 </div>
997 </div>
999 <!-- _______________________________________________________________________ -->
1002 </div>
1006 <p>In LLVM, we prefer to explicitly prefix all identifiers from the standard
1007 namespace with an "<tt>std::</tt>" prefix, rather than rely on
1012 clearly a bad thing.</p>
1015 rule, but is still important. Basically, using explicit namespace prefixes
1018 namespace clashes cannot occur between LLVM code and other namespaces. The
1019 portability rule is important because different standard library implementations
1020 expose different symbols (potentially ones they shouldn't), and future revisions
1024 <p>The exception to the general rule (i.e. it's not an exception for
1026 the code in the LLVM project implements code that lives in the 'llvm' namespace.
1030 editors that indent based on braces, and keeps the conceptual context cleaner.
1032 code in any namespace may use that namespace (and its parents'), but should not
1033 use any others.</p>
1035 </div>
1037 <!-- _______________________________________________________________________ -->
1040 in Headers</a>
1041 </div>
1045 <p>If a class is defined in a header file and has a v-table (either it has
1046 virtual methods or it derives from classes with virtual methods), it must
1047 always have at least one out-of-line virtual method in the class. Without
1050 increasing link times.</p>
1052 </div>
1054 <!-- _______________________________________________________________________ -->
1057 </div>
1062 emulated with macros and may be coming in C++'0x) we end up writing a lot of
1063 loops that manually iterate from begin to end on a variety of containers or
1064 through other data structures. One common mistake is to write a loop in this
1065 style:</p>
1068 <pre>
1069 BasicBlock *BB = ...
1071 ... use I ...
1072 </pre>
1073 </div>
1076 every time through the loop. Instead of writing the loop like this, we strongly
1077 prefer loops to be written so that they evaluate it once before the loop starts.
1078 A convenient way to do this is like so:</p>
1081 <pre>
1082 BasicBlock *BB = ...
1084 ... use I ...
1085 </pre>
1086 </div>
1088 <p>The observant may quickly point out that these two loops may have different
1089 semantics: if the container (a basic block in this case) is being mutated, then
1090 "<tt>BB->end()</tt>" may change its value every time through the loop and the
1091 second loop may not in fact be correct. If you actually do depend on this
1092 behavior, please write the loop in the first form and add a comment indicating
1093 that you did it intentionally.</p>
1095 <p>Why do we prefer the second form (when correct)? Writing the loop in the
1096 first form has two problems. First it may be less efficient than evaluating it
1097 at the start of the loop. In this case, the cost is probably minor — a
1098 few extra loads every time through the loop. However, if the base expression is
1099 more complex, then the cost can rise quickly. I've seen loops where the end
1100 expression was actually something like: "<tt>SomeMap[x]->end()</tt>" and map
1101 lookups really aren't cheap. By writing it in the second form consistently, you
1102 eliminate the issue entirely and don't even have to think about it.</p>
1104 <p>The second (even bigger) issue is that writing the loop in the first form
1105 hints to the reader that the loop is mutating the container (a fact that a
1106 comment would handily confirm!). If you write the loop in the second form, it
1107 is immediately obvious without even looking at the body of the loop that the
1108 container isn't being modified, which makes it easier to read the code and
1109 understand what it does.</p>
1111 <p>While the second form of the loop is a few extra keystrokes, we do strongly
1112 prefer it.</p>
1114 </div>
1116 <!-- _______________________________________________________________________ -->
1119 </div>
1125 support clients using LLVM libraries as part of larger systems. In particular,
1126 we statically link LLVM into some dynamic libraries. Even if LLVM isn't used,
1127 the static constructors are run whenever an application starts up that uses the
1128 dynamic library. There are two problems with this:</p>
1130 <ol>
1131 <li>The time to run the static c'tors impacts startup time of applications
1134 <li>The static c'tors cause the app to pull many extra pages of memory off the
1136 small amount of data that gets touched. In addition, touched/dirty pages
1137 put more pressure on the VM system on low-memory machines.</li>
1138 </ol>
1141 example) is not problematic in this regard —
1144 style APIs. <b>Therefore new code should always
1148 </div>
1151 <!-- _______________________________________________________________________ -->
1154 </div>
1158 <p>LLVM includes a lightweight, simple, and efficient stream implementation
1168 </div>
1171 <!-- _______________________________________________________________________ -->
1174 </div>
1179 newline to the output stream specified. In addition to doing this, however, it
1180 also flushes the output stream. In other words, these are equivalent:</p>
1183 <pre>
1184 std::cout << std::endl;
1186 </pre>
1187 </div>
1189 <p>Most of the time, you probably have no reason to flush the output stream, so
1192 </div>
1195 <!-- ======================================================================= -->
1198 </div>
1199 <!-- ======================================================================= -->
1201 <p>This section describes preferred low-level formatting guidelines along with
1202 reasoning on why we prefer them.</p>
1204 <!-- _______________________________________________________________________ -->
1207 </div>
1211 <p>We prefer to put a space before an open parenthesis only in control flow
1212 statements, but not in normal function call expressions and function-like
1213 macros. For example, this is good:</p>
1216 <pre>
1225 </pre>
1226 </div>
1231 <pre>
1240 </pre>
1241 </div>
1243 <p>The reason for doing this is not completely arbitrary. This style makes
1244 control flow operators stand out more, and makes expressions flow better. The
1245 function call operator binds very tightly as a postfix operator. Putting a
1246 space after a function name (as in the last example) makes it appear that the
1247 code might bind the arguments of the left-hand-side of a binary operator with
1248 the argument list of a function and the name of the right side. More
1252 <pre>
1254 </pre>
1255 </div>
1257 <p>when skimming through the code. By avoiding a space in a function, we avoid
1258 this misinterpretation.</p>
1260 </div>
1262 <!-- _______________________________________________________________________ -->
1265 </div>
1271 preincrementation whenever possible.</p>
1273 <p>The semantics of postincrement include making a copy of the value being
1274 incremented, returning it, and then preincrementing the "work value". For
1275 primitive types, this isn't a big deal... but for iterators, it can be a huge
1276 issue (for example, some iterators contains stack and set objects in them...
1277 copying an iterator could invoke the copy ctor's of these as well). In general,
1278 get in the habit of always using preincrement, and you won't have a problem.</p>
1280 </div>
1282 <!-- _______________________________________________________________________ -->
1285 </div>
1289 <p>
1290 In general, we strive to reduce indentation wherever possible. This is useful
1292 wrapping horribly, but also because it makes it easier to understand the code.
1293 Namespaces are a funny thing: they are often large, and we often desire to put
1294 lots of stuff into them (so they can be large). Other times they are tiny,
1295 because they just hold an enum or something similar. In order to balance this,
1296 we use different approaches for small versus large namespaces.
1297 </p>
1299 <p>
1301 less than 35 lines of code), then you should indent its body. Here's an
1302 example:
1303 </p>
1306 <pre>
1307 namespace llvm {
1308 namespace X86 {
1309 /// RelocationType - An enum for the x86 relocation codes. Note that
1310 /// the terminology here doesn't follow x86 convention - word means
1312 enum RelocationType {
1313 /// reloc_pcrel_word - PC relative relocation, add the relocated value to
1314 /// the value already in memory, after we adjust it for where the PC is.
1315 reloc_pcrel_word = 0,
1317 /// reloc_picrel_word - PIC base relative relocation, add the relocated
1318 /// value to the value already in memory, after we adjust it for where the
1319 /// PIC base is.
1320 reloc_picrel_word = 1,
1322 /// reloc_absolute_word, reloc_absolute_dword - Absolute relocation, just
1323 /// add the relocated value to the value already in memory.
1324 reloc_absolute_word = 2,
1325 reloc_absolute_dword = 3
1326 };
1327 }
1328 }
1329 </pre>
1330 </div>
1332 <p>Since the body is small, indenting adds value because it makes it very clear
1333 where the namespace starts and ends, and it is easy to take the whole thing in
1334 in one "gulp" when reading the code. If the blob of code in the namespace is
1335 larger (as it typically is in a header in the <tt>llvm</tt> or <tt>clang</tt> namespaces), do not
1336 indent the code, and add a comment indicating what namespace is being closed.
1337 For example:</p>
1340 <pre>
1341 namespace llvm {
1342 namespace knowledge {
1344 /// Grokable - This class represents things that Smith can have an intimate
1345 /// understanding of and contains the data associated with it.
1346 class Grokable {
1347 ...
1348 public:
1349 explicit Grokable() { ... }
1350 virtual ~Grokable() = 0;
1352 ...
1354 };
1356 } // end namespace knowledge
1357 } // end namespace llvm
1358 </pre>
1359 </div>
1361 <p>Because the class is large, we don't expect that the reader can easily
1362 understand the entire concept in a glance, and the end of the file (where the
1363 namespaces end) may be a long ways away from the place they open. As such,
1364 indenting the contents of the namespace doesn't add any value, and detracts from
1366 the contents of the namespace.</p>
1368 </div>
1370 <!-- _______________________________________________________________________ -->
1373 </div>
1377 <p>After talking about namespaces in general, you may be wondering about
1378 anonymous namespaces in particular.
1379 Anonymous namespaces are a great language feature that tells the C++ compiler
1380 that the contents of the namespace are only visible within the current
1381 translation unit, allowing more aggressive optimization and eliminating the
1382 possibility of symbol name collisions. Anonymous namespaces are to C++ as
1384 in C++, anonymous namespaces are more general: they can make entire classes
1385 private to a file.</p>
1387 <p>The problem with anonymous namespaces is that they naturally want to
1388 encourage indentation of their body, and they reduce locality of reference: if
1389 you see a random function definition in a C++ file, it is easy to see if it is
1390 marked static, but seeing if it is in an anonymous namespace requires scanning
1391 a big chunk of the file.</p>
1393 <p>Because of this, we have a simple guideline: make anonymous namespaces as
1394 small as possible, and only use them for class declarations. For example, this
1395 is good:</p>
1398 <pre>
1400 class StringSort {
1401 ...
1402 public:
1403 StringSort(...)
1404 bool operator<(const char *RHS) const;
1405 };
1408 static void Helper() {
1409 ...
1410 }
1412 bool StringSort::operator<(const char *RHS) const {
1413 ...
1414 }
1416 </pre>
1417 </div>
1423 <pre>
1425 class StringSort {
1426 ...
1427 public:
1428 StringSort(...)
1429 bool operator<(const char *RHS) const;
1430 };
1432 void Helper() {
1433 ...
1434 }
1436 bool StringSort::operator<(const char *RHS) const {
1437 ...
1438 }
1442 </pre>
1443 </div>
1447 of a large C++ file, that you have no immediate way to tell if it is local to
1448 the file. When it is marked static explicitly, this is immediately obvious.
1449 Also, there is no reason to enclose the definition of "operator<" in the
1450 namespace just because it was declared there.
1451 </p>
1453 </div>
1457 <!-- *********************************************************************** -->
1460 </div>
1461 <!-- *********************************************************************** -->
1465 <p>A lot of these comments and recommendations have been culled for other
1466 sources. Two particularly important books for our work are:</p>
1468 <ol>
1470 <li><a href="http://www.amazon.com/Effective-Specific-Addison-Wesley-Professional-Computing/dp/0321334876">Effective
1471 C++</a> by Scott Meyers. Also
1473 author.</li>
1477 </ol>
1479 <p>If you get some free time, and you haven't read them: do so, you might learn
1480 something.</p>
1482 </div>
1484 <!-- *********************************************************************** -->
1486 <hr>
1487 <address>
1495 Last modified: $Date$
1496 </address>
1498 </body>
1499 </html>