13 This document attempts to describe a few coding standards that are being used in
14 the LLVM source tree. Although no coding standards should be regarded as
15 absolute requirements to be followed in all instances, coding standards are
16 particularly important for large-scale code bases that follow a library-based
19 This document intentionally does not prescribe fixed standards for religious
20 issues such as brace placement and space usage. For issues like this, follow
25 **If you are extending, enhancing, or bug fixing already implemented code,
26 use the style that is already being used so that the source is uniform and
29 Note that some code bases (e.g. ``libc++``) have really good reasons to deviate
30 from the coding standards. In the case of ``libc++``, this is because the
31 naming and other conventions are dictated by the C++ standard. If you think
32 there is a specific good reason to deviate from the standards here, please bring
33 it up on the LLVMdev mailing list.
35 There are some conventions that are not uniformly followed in the code base
36 (e.g. the naming convention). This is because they are relatively new, and a
37 lot of code was written before they were put in place. Our long term goal is
38 for the entire codebase to follow the convention, but we explicitly *do not*
39 want patches that do large-scale reformating of existing code. On the other
40 hand, it is reasonable to rename the methods of a class if you're about to
41 change it in some other way. Just do the reformating as a separate commit from
42 the functionality change.
44 The ultimate goal of these guidelines is the increase readability and
45 maintainability of our common source base. If you have suggestions for topics to
46 be included, please mail them to `Chris <mailto:sabre@nondot.org>`_.
48 Mechanical Source Issues
49 ========================
51 Source Code Formatting
52 ----------------------
57 Comments are one critical part of readability and maintainability. Everyone
58 knows they should comment their code, and so should you. When writing comments,
59 write them as English prose, which means they should use proper capitalization,
60 punctuation, etc. Aim to describe what the code is trying to do and why, not
61 *how* it does it at a micro level. Here are a few critical things to document:
63 .. _header file comment:
68 Every source file should have a header on it that describes the basic purpose of
69 the file. If a file does not have a header, it should not be checked into the
70 tree. The standard header looks like this:
74 //===-- llvm/Instruction.h - Instruction class definition -------*- C++ -*-===//
76 // The LLVM Compiler Infrastructure
78 // This file is distributed under the University of Illinois Open Source
79 // License. See LICENSE.TXT for details.
81 //===----------------------------------------------------------------------===//
84 /// \brief This file contains the declaration of the Instruction class, which is
85 /// the base class for all of the VM instructions.
87 //===----------------------------------------------------------------------===//
89 A few things to note about this particular format: The "``-*- C++ -*-``" string
90 on the first line is there to tell Emacs that the source file is a C++ file, not
91 a C file (Emacs assumes ``.h`` files are C files by default).
95 This tag is not necessary in ``.cpp`` files. The name of the file is also
96 on the first line, along with a very short description of the purpose of the
97 file. This is important when printing out code and flipping though lots of
100 The next section in the file is a concise note that defines the license that the
101 file is released under. This makes it perfectly clear what terms the source
102 code can be distributed under and should not be modified in any way.
104 The main body is a ``doxygen`` comment describing the purpose of the file. It
105 should have a ``\brief`` command that describes the file in one or two
106 sentences. Any additional information should be separated by a blank line. If
107 an algorithm is being implemented or something tricky is going on, a reference
108 to the paper where it is published should be included, as well as any notes or
109 *gotchas* in the code to watch out for.
114 Classes are one fundamental part of a good object oriented design. As such, a
115 class definition should have a comment block that explains what the class is
116 used for and how it works. Every non-trivial class is expected to have a
117 ``doxygen`` comment block.
122 Methods defined in a class (as well as any global functions) should also be
123 documented properly. A quick note about what it does and a description of the
124 borderline behaviour is all that is necessary here (unless something
125 particularly tricky or insidious is going on). The hope is that people can
126 figure out how to use your interfaces without reading the code itself.
128 Good things to talk about here are what happens when something unexpected
129 happens: does the method return null? Abort? Format your hard disk?
134 In general, prefer C++ style (``//``) comments. They take less space, require
135 less typing, don't have nesting problems, etc. There are a few cases when it is
136 useful to use C style (``/* */``) comments however:
138 #. When writing C code: Obviously if you are writing C code, use C style
141 #. When writing a header file that may be ``#include``\d by a C source file.
143 #. When writing a source file that is used by a tool that only accepts C style
146 To comment out a large block of code, use ``#if 0`` and ``#endif``. These nest
147 properly and are better behaved in general than C style comments.
149 Doxygen Use in Documentation Comments
150 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
152 Use the ``\file`` command to turn the standard file header into a file-level
155 Include descriptive ``\brief`` paragraphs for all public interfaces (public
156 classes, member and non-member functions). Explain API use and purpose in
157 ``\brief`` paragraphs, don't just restate the information that can be inferred
158 from the API name. Put detailed discussion into separate paragraphs.
160 To refer to parameter names inside a paragraph, use the ``\p name`` command.
161 Don't use the ``\arg name`` command since it starts a new paragraph that
162 contains documentation for the parameter.
164 Wrap non-inline code examples in ``\code ... \endcode``.
166 To document a function parameter, start a new paragraph with the
167 ``\param name`` command. If the parameter is used as an out or an in/out
168 parameter, use the ``\param [out] name`` or ``\param [in,out] name`` command,
171 To describe function return value, start a new paragraph with the ``\returns``
174 A minimal documentation comment:
178 /// \brief Does foo and bar.
179 void fooBar(bool Baz);
181 A documentation comment that uses all Doxygen features in a preferred way:
185 /// \brief Does foo and bar.
187 /// Does not do foo the usual way if \p Baz is true.
191 /// fooBar(false, "quux", Res);
194 /// \param Quux kind of foo to do.
195 /// \param [out] Result filled with bar sequence on foo success.
197 /// \returns true on success.
198 bool fooBar(bool Baz, StringRef Quux, std::vector<int> &Result);
200 Don't duplicate the documentation comment in the header file and in the
201 implementation file. Put the documentation comments for public APIs into the
202 header file. Documentation comments for private APIs can go to the
203 implementation file. In any case, implementation files can include additional
204 comments (not necessarily in Doxygen markup) to explain implementation details
207 Don't duplicate function or class name at the beginning of the comment.
208 For humans it is obvious which function or class is being documented;
209 automatic documentation processing tools are smart enough to bind the comment
210 to the correct declaration.
218 /// Something - An abstraction for some complicated thing.
221 /// fooBar - Does foo and bar.
227 /// fooBar - Does foo and bar.
228 void Something::fooBar() { ... }
236 /// \brief An abstraction for some complicated thing.
239 /// \brief Does foo and bar.
245 // Builds a B-tree in order to do foo. See paper by...
246 void Something::fooBar() { ... }
248 It is not required to use additional Doxygen features, but sometimes it might
249 be a good idea to do so.
253 * adding comments to any narrow namespace containing a collection of
254 related functions or types;
256 * using top-level groups to organize a collection of related functions at
257 namespace scope where the grouping is smaller than the namespace;
259 * using member groups and additional comments attached to member
260 groups to organize within a class.
267 /// \name Functions that do Foo.
278 Immediately after the `header file comment`_ (and include guards if working on a
279 header file), the `minimal list of #includes`_ required by the file should be
280 listed. We prefer these ``#include``\s to be listed in this order:
282 .. _Main Module Header:
283 .. _Local/Private Headers:
285 #. Main Module Header
286 #. Local/Private Headers
288 #. ``llvm/Analysis/*``
289 #. ``llvm/Assembly/*``
290 #. ``llvm/Bitcode/*``
291 #. ``llvm/CodeGen/*``
293 #. ``llvm/Support/*``
295 #. System ``#include``\s
297 and each category should be sorted by name.
299 The `Main Module Header`_ file applies to ``.cpp`` files which implement an
300 interface defined by a ``.h`` file. This ``#include`` should always be included
301 **first** regardless of where it lives on the file system. By including a
302 header file first in the ``.cpp`` files that implement the interfaces, we ensure
303 that the header does not have any hidden dependencies which are not explicitly
304 ``#include``\d in the header, but should be. It is also a form of documentation
305 in the ``.cpp`` file to indicate where the interfaces it implements are defined.
307 .. _fit into 80 columns:
312 Write your code to fit within 80 columns of text. This helps those of us who
313 like to print out code and look at your code in an ``xterm`` without resizing
316 The longer answer is that there must be some limit to the width of the code in
317 order to reasonably allow developers to have multiple files side-by-side in
318 windows on a modest display. If you are going to pick a width limit, it is
319 somewhat arbitrary but you might as well pick something standard. Going with 90
320 columns (for example) instead of 80 columns wouldn't add any significant value
321 and would be detrimental to printing out code. Also many other projects have
322 standardized on 80 columns, so some people have already configured their editors
323 for it (vs something else, like 90 columns).
325 This is one of many contentious issues in coding standards, but it is not up for
328 Use Spaces Instead of Tabs
329 ^^^^^^^^^^^^^^^^^^^^^^^^^^
331 In all cases, prefer spaces to tabs in source files. People have different
332 preferred indentation levels, and different styles of indentation that they
333 like; this is fine. What isn't fine is that different editors/viewers expand
334 tabs out to different tab stops. This can cause your code to look completely
335 unreadable, and it is not worth dealing with.
337 As always, follow the `Golden Rule`_ above: follow the style of
338 existing code if you are modifying and extending it. If you like four spaces of
339 indentation, **DO NOT** do that in the middle of a chunk of code with two spaces
340 of indentation. Also, do not reindent a whole source file: it makes for
341 incredible diffs that are absolutely worthless.
343 Indent Code Consistently
344 ^^^^^^^^^^^^^^^^^^^^^^^^
346 Okay, in your first year of programming you were told that indentation is
347 important. If you didn't believe and internalize this then, now is the time.
353 Treat Compiler Warnings Like Errors
354 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
356 If your code has compiler warnings in it, something is wrong --- you aren't
357 casting values correctly, you have "questionable" constructs in your code, or
358 you are doing something legitimately wrong. Compiler warnings can cover up
359 legitimate errors in output and make dealing with a translation unit difficult.
361 It is not possible to prevent all warnings from all compilers, nor is it
362 desirable. Instead, pick a standard compiler (like ``gcc``) that provides a
363 good thorough set of warnings, and stick to it. At least in the case of
364 ``gcc``, it is possible to work around any spurious errors by changing the
365 syntax of the code slightly. For example, a warning that annoys me occurs when
366 I write code like this:
370 if (V = getValue()) {
374 ``gcc`` will warn me that I probably want to use the ``==`` operator, and that I
375 probably mistyped it. In most cases, I haven't, and I really don't want the
376 spurious errors. To fix this particular problem, I rewrite the code like
381 if ((V = getValue())) {
385 which shuts ``gcc`` up. Any ``gcc`` warning that annoys you can be fixed by
386 massaging the code appropriately.
391 In almost all cases, it is possible and within reason to write completely
392 portable code. If there are cases where it isn't possible to write portable
393 code, isolate it behind a well defined (and well documented) interface.
395 In practice, this means that you shouldn't assume much about the host compiler
396 (and Visual Studio tends to be the lowest common denominator). If advanced
397 features are used, they should only be an implementation detail of a library
398 which has a simple exposed API, and preferably be buried in ``libSystem``.
400 Do not use RTTI or Exceptions
401 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
403 In an effort to reduce code and executable size, LLVM does not use RTTI
404 (e.g. ``dynamic_cast<>;``) or exceptions. These two language features violate
405 the general C++ principle of *"you only pay for what you use"*, causing
406 executable bloat even if exceptions are never used in the code base, or if RTTI
407 is never used for a class. Because of this, we turn them off globally in the
410 That said, LLVM does make extensive use of a hand-rolled form of RTTI that use
411 templates like `isa<>, cast<>, and dyn_cast<> <ProgrammersManual.html#isa>`_.
412 This form of RTTI is opt-in and can be
413 :doc:`added to any class <HowToSetUpLLVMStyleRTTI>`. It is also
414 substantially more efficient than ``dynamic_cast<>``.
416 .. _static constructor:
418 Do not use Static Constructors
419 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
421 Static constructors and destructors (e.g. global variables whose types have a
422 constructor or destructor) should not be added to the code base, and should be
423 removed wherever possible. Besides `well known problems
424 <http://yosefk.com/c++fqa/ctors.html#fqa-10.12>`_ where the order of
425 initialization is undefined between globals in different source files, the
426 entire concept of static constructors is at odds with the common use case of
427 LLVM as a library linked into a larger application.
429 Consider the use of LLVM as a JIT linked into another application (perhaps for
430 `OpenGL, custom languages <http://llvm.org/Users.html>`_, `shaders in movies
431 <http://llvm.org/devmtg/2010-11/Gritz-OpenShadingLang.pdf>`_, etc). Due to the
432 design of static constructors, they must be executed at startup time of the
433 entire application, regardless of whether or how LLVM is used in that larger
434 application. There are two problems with this:
436 * The time to run the static constructors impacts startup time of applications
437 --- a critical time for GUI apps, among others.
439 * The static constructors cause the app to pull many extra pages of memory off
440 the disk: both the code for the constructor in each ``.o`` file and the small
441 amount of data that gets touched. In addition, touched/dirty pages put more
442 pressure on the VM system on low-memory machines.
444 We would really like for there to be zero cost for linking in an additional LLVM
445 target or other library into an application, but static constructors violate
448 That said, LLVM unfortunately does contain static constructors. It would be a
449 `great project <http://llvm.org/PR11944>`_ for someone to purge all static
450 constructors from LLVM, and then enable the ``-Wglobal-constructors`` warning
451 flag (when building with Clang) to ensure we do not regress in the future.
453 Use of ``class`` and ``struct`` Keywords
454 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
456 In C++, the ``class`` and ``struct`` keywords can be used almost
457 interchangeably. The only difference is when they are used to declare a class:
458 ``class`` makes all members private by default while ``struct`` makes all
459 members public by default.
461 Unfortunately, not all compilers follow the rules and some will generate
462 different symbols based on whether ``class`` or ``struct`` was used to declare
463 the symbol. This can lead to problems at link time.
465 So, the rule for LLVM is to always use the ``class`` keyword, unless **all**
466 members are public and the type is a C++ `POD
467 <http://en.wikipedia.org/wiki/Plain_old_data_structure>`_ type, in which case
468 ``struct`` is allowed.
473 The High-Level Issues
474 ---------------------
476 A Public Header File **is** a Module
477 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
479 C++ doesn't do too well in the modularity department. There is no real
480 encapsulation or data hiding (unless you use expensive protocol classes), but it
481 is what we have to work with. When you write a public header file (in the LLVM
482 source tree, they live in the top level "``include``" directory), you are
483 defining a module of functionality.
485 Ideally, modules should be completely independent of each other, and their
486 header files should only ``#include`` the absolute minimum number of headers
487 possible. A module is not just a class, a function, or a namespace: it's a
488 collection of these that defines an interface. This interface may be several
489 functions, classes, or data structures, but the important issue is how they work
492 In general, a module should be implemented by one or more ``.cpp`` files. Each
493 of these ``.cpp`` files should include the header that defines their interface
494 first. This ensures that all of the dependences of the module header have been
495 properly added to the module header itself, and are not implicit. System
496 headers should be included after user headers for a translation unit.
498 .. _minimal list of #includes:
500 ``#include`` as Little as Possible
501 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
503 ``#include`` hurts compile time performance. Don't do it unless you have to,
504 especially in header files.
506 But wait! Sometimes you need to have the definition of a class to use it, or to
507 inherit from it. In these cases go ahead and ``#include`` that header file. Be
508 aware however that there are many cases where you don't need to have the full
509 definition of a class. If you are using a pointer or reference to a class, you
510 don't need the header file. If you are simply returning a class instance from a
511 prototyped function or method, you don't need it. In fact, for most cases, you
512 simply don't need the definition of a class. And not ``#include``\ing speeds up
515 It is easy to try to go too overboard on this recommendation, however. You
516 **must** include all of the header files that you are using --- you can include
517 them either directly or indirectly through another header file. To make sure
518 that you don't accidentally forget to include a header file in your module
519 header, make sure to include your module header **first** in the implementation
520 file (as mentioned above). This way there won't be any hidden dependencies that
521 you'll find out about later.
523 Keep "Internal" Headers Private
524 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
526 Many modules have a complex implementation that causes them to use more than one
527 implementation (``.cpp``) file. It is often tempting to put the internal
528 communication interface (helper classes, extra functions, etc) in the public
529 module header file. Don't do this!
531 If you really need to do something like this, put a private header file in the
532 same directory as the source files, and include it locally. This ensures that
533 your private interface remains private and undisturbed by outsiders.
537 It's okay to put extra implementation methods in a public class itself. Just
538 make them private (or protected) and all is well.
542 Use Early Exits and ``continue`` to Simplify Code
543 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
545 When reading code, keep in mind how much state and how many previous decisions
546 have to be remembered by the reader to understand a block of code. Aim to
547 reduce indentation where possible when it doesn't make it more difficult to
548 understand the code. One great way to do this is by making use of early exits
549 and the ``continue`` keyword in long loops. As an example of using an early
550 exit from a function, consider this "bad" code:
554 Value *doSomething(Instruction *I) {
555 if (!isa<TerminatorInst>(I) &&
556 I->hasOneUse() && doOtherThing(I)) {
557 ... some long code ....
563 This code has several problems if the body of the ``'if'`` is large. When
564 you're looking at the top of the function, it isn't immediately clear that this
565 *only* does interesting things with non-terminator instructions, and only
566 applies to things with the other predicates. Second, it is relatively difficult
567 to describe (in comments) why these predicates are important because the ``if``
568 statement makes it difficult to lay out the comments. Third, when you're deep
569 within the body of the code, it is indented an extra level. Finally, when
570 reading the top of the function, it isn't clear what the result is if the
571 predicate isn't true; you have to read to the end of the function to know that
574 It is much preferred to format the code like this:
578 Value *doSomething(Instruction *I) {
579 // Terminators never need 'something' done to them because ...
580 if (isa<TerminatorInst>(I))
583 // We conservatively avoid transforming instructions with multiple uses
584 // because goats like cheese.
588 // This is really just here for example.
589 if (!doOtherThing(I))
592 ... some long code ....
595 This fixes these problems. A similar problem frequently happens in ``for``
596 loops. A silly example is something like this:
600 for (BasicBlock::iterator II = BB->begin(), E = BB->end(); II != E; ++II) {
601 if (BinaryOperator *BO = dyn_cast<BinaryOperator>(II)) {
602 Value *LHS = BO->getOperand(0);
603 Value *RHS = BO->getOperand(1);
610 When you have very, very small loops, this sort of structure is fine. But if it
611 exceeds more than 10-15 lines, it becomes difficult for people to read and
612 understand at a glance. The problem with this sort of code is that it gets very
613 nested very quickly. Meaning that the reader of the code has to keep a lot of
614 context in their brain to remember what is going immediately on in the loop,
615 because they don't know if/when the ``if`` conditions will have ``else``\s etc.
616 It is strongly preferred to structure the loop like this:
620 for (BasicBlock::iterator II = BB->begin(), E = BB->end(); II != E; ++II) {
621 BinaryOperator *BO = dyn_cast<BinaryOperator>(II);
624 Value *LHS = BO->getOperand(0);
625 Value *RHS = BO->getOperand(1);
626 if (LHS == RHS) continue;
631 This has all the benefits of using early exits for functions: it reduces nesting
632 of the loop, it makes it easier to describe why the conditions are true, and it
633 makes it obvious to the reader that there is no ``else`` coming up that they
634 have to push context into their brain for. If a loop is large, this can be a
635 big understandability win.
637 Don't use ``else`` after a ``return``
638 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
640 For similar reasons above (reduction of indentation and easier reading), please
641 do not use ``'else'`` or ``'else if'`` after something that interrupts control
642 flow --- like ``return``, ``break``, ``continue``, ``goto``, etc. For
643 example, this is *bad*:
649 Type = Context.getsigjmp_bufType();
651 Error = ASTContext::GE_Missing_sigjmp_buf;
657 Type = Context.getjmp_bufType();
659 Error = ASTContext::GE_Missing_jmp_buf;
667 It is better to write it like this:
673 Type = Context.getsigjmp_bufType();
675 Error = ASTContext::GE_Missing_sigjmp_buf;
679 Type = Context.getjmp_bufType();
681 Error = ASTContext::GE_Missing_jmp_buf;
687 Or better yet (in this case) as:
693 Type = Context.getsigjmp_bufType();
695 Type = Context.getjmp_bufType();
698 Error = Signed ? ASTContext::GE_Missing_sigjmp_buf :
699 ASTContext::GE_Missing_jmp_buf;
704 The idea is to reduce indentation and the amount of code you have to keep track
705 of when reading the code.
707 Turn Predicate Loops into Predicate Functions
708 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
710 It is very common to write small loops that just compute a boolean value. There
711 are a number of ways that people commonly write these, but an example of this
716 bool FoundFoo = false;
717 for (unsigned I = 0, E = BarList.size(); I != E; ++I)
718 if (BarList[I]->isFoo()) {
727 This sort of code is awkward to write, and is almost always a bad sign. Instead
728 of this sort of loop, we strongly prefer to use a predicate function (which may
729 be `static`_) that uses `early exits`_ to compute the predicate. We prefer the
730 code to be structured like this:
734 /// \returns true if the specified list has an element that is a foo.
735 static bool containsFoo(const std::vector<Bar*> &List) {
736 for (unsigned I = 0, E = List.size(); I != E; ++I)
737 if (List[I]->isFoo())
743 if (containsFoo(BarList)) {
747 There are many reasons for doing this: it reduces indentation and factors out
748 code which can often be shared by other code that checks for the same predicate.
749 More importantly, it *forces you to pick a name* for the function, and forces
750 you to write a comment for it. In this silly example, this doesn't add much
751 value. However, if the condition is complex, this can make it a lot easier for
752 the reader to understand the code that queries for this predicate. Instead of
753 being faced with the in-line details of how we check to see if the BarList
754 contains a foo, we can trust the function name and continue reading with better
760 Name Types, Functions, Variables, and Enumerators Properly
761 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
763 Poorly-chosen names can mislead the reader and cause bugs. We cannot stress
764 enough how important it is to use *descriptive* names. Pick names that match
765 the semantics and role of the underlying entities, within reason. Avoid
766 abbreviations unless they are well known. After picking a good name, make sure
767 to use consistent capitalization for the name, as inconsistency requires clients
768 to either memorize the APIs or to look it up to find the exact spelling.
770 In general, names should be in camel case (e.g. ``TextFileReader`` and
771 ``isLValue()``). Different kinds of declarations have different rules:
773 * **Type names** (including classes, structs, enums, typedefs, etc) should be
774 nouns and start with an upper-case letter (e.g. ``TextFileReader``).
776 * **Variable names** should be nouns (as they represent state). The name should
777 be camel case, and start with an upper case letter (e.g. ``Leader`` or
780 * **Function names** should be verb phrases (as they represent actions), and
781 command-like function should be imperative. The name should be camel case,
782 and start with a lower case letter (e.g. ``openFile()`` or ``isFoo()``).
784 * **Enum declarations** (e.g. ``enum Foo {...}``) are types, so they should
785 follow the naming conventions for types. A common use for enums is as a
786 discriminator for a union, or an indicator of a subclass. When an enum is
787 used for something like this, it should have a ``Kind`` suffix
788 (e.g. ``ValueKind``).
790 * **Enumerators** (e.g. ``enum { Foo, Bar }``) and **public member variables**
791 should start with an upper-case letter, just like types. Unless the
792 enumerators are defined in their own small namespace or inside a class,
793 enumerators should have a prefix corresponding to the enum declaration name.
794 For example, ``enum ValueKind { ... };`` may contain enumerators like
795 ``VK_Argument``, ``VK_BasicBlock``, etc. Enumerators that are just
796 convenience constants are exempt from the requirement for a prefix. For
806 As an exception, classes that mimic STL classes can have member names in STL's
807 style of lower-case words separated by underscores (e.g. ``begin()``,
808 ``push_back()``, and ``empty()``).
810 Here are some examples of good and bad names:
816 Factory<Tire> F; // Bad -- abbreviation and non-descriptive.
817 Factory<Tire> Factory; // Better.
818 Factory<Tire> TireFactory; // Even better -- if VehicleMaker has more than one
819 // kind of factories.
822 Vehicle MakeVehicle(VehicleType Type) {
823 VehicleMaker M; // Might be OK if having a short life-span.
824 Tire Tmp1 = M.makeTire(); // Bad -- 'Tmp1' provides no information.
825 Light Headlight = M.makeLight("head"); // Good -- descriptive.
832 Use the "``assert``" macro to its fullest. Check all of your preconditions and
833 assumptions, you never know when a bug (not necessarily even yours) might be
834 caught early by an assertion, which reduces debugging time dramatically. The
835 "``<cassert>``" header file is probably already included by the header files you
836 are using, so it doesn't cost anything to use it.
838 To further assist with debugging, make sure to put some kind of error message in
839 the assertion statement, which is printed if the assertion is tripped. This
840 helps the poor debugger make sense of why an assertion is being made and
841 enforced, and hopefully what to do about it. Here is one complete example:
845 inline Value *getOperand(unsigned I) {
846 assert(I < Operands.size() && "getOperand() out of range!");
850 Here are more examples:
854 assert(Ty->isPointerType() && "Can't allocate a non pointer type!");
856 assert((Opcode == Shl || Opcode == Shr) && "ShiftInst Opcode invalid!");
858 assert(idx < getNumSuccessors() && "Successor # out of range!");
860 assert(V1.getType() == V2.getType() && "Constant types must be identical!");
862 assert(isa<PHINode>(Succ->front()) && "Only works on PHId BBs!");
866 In the past, asserts were used to indicate a piece of code that should not be
867 reached. These were typically of the form:
871 assert(0 && "Invalid radix for integer literal");
873 This has a few issues, the main one being that some compilers might not
874 understand the assertion, or warn about a missing return in builds where
875 assertions are compiled out.
877 Today, we have something much better: ``llvm_unreachable``:
881 llvm_unreachable("Invalid radix for integer literal");
883 When assertions are enabled, this will print the message if it's ever reached
884 and then exit the program. When assertions are disabled (i.e. in release
885 builds), ``llvm_unreachable`` becomes a hint to compilers to skip generating
886 code for this branch. If the compiler does not support this, it will fall back
887 to the "abort" implementation.
889 Another issue is that values used only by assertions will produce an "unused
890 value" warning when assertions are disabled. For example, this code will warn:
894 unsigned Size = V.size();
895 assert(Size > 42 && "Vector smaller than it should be");
897 bool NewToSet = Myset.insert(Value);
898 assert(NewToSet && "The value shouldn't be in the set yet");
900 These are two interesting different cases. In the first case, the call to
901 ``V.size()`` is only useful for the assert, and we don't want it executed when
902 assertions are disabled. Code like this should move the call into the assert
903 itself. In the second case, the side effects of the call must happen whether
904 the assert is enabled or not. In this case, the value should be cast to void to
905 disable the warning. To be specific, it is preferred to write the code like
910 assert(V.size() > 42 && "Vector smaller than it should be");
912 bool NewToSet = Myset.insert(Value); (void)NewToSet;
913 assert(NewToSet && "The value shouldn't be in the set yet");
915 Do Not Use ``using namespace std``
916 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
918 In LLVM, we prefer to explicitly prefix all identifiers from the standard
919 namespace with an "``std::``" prefix, rather than rely on "``using namespace
922 In header files, adding a ``'using namespace XXX'`` directive pollutes the
923 namespace of any source file that ``#include``\s the header. This is clearly a
926 In implementation files (e.g. ``.cpp`` files), the rule is more of a stylistic
927 rule, but is still important. Basically, using explicit namespace prefixes
928 makes the code **clearer**, because it is immediately obvious what facilities
929 are being used and where they are coming from. And **more portable**, because
930 namespace clashes cannot occur between LLVM code and other namespaces. The
931 portability rule is important because different standard library implementations
932 expose different symbols (potentially ones they shouldn't), and future revisions
933 to the C++ standard will add more symbols to the ``std`` namespace. As such, we
934 never use ``'using namespace std;'`` in LLVM.
936 The exception to the general rule (i.e. it's not an exception for the ``std``
937 namespace) is for implementation files. For example, all of the code in the
938 LLVM project implements code that lives in the 'llvm' namespace. As such, it is
939 ok, and actually clearer, for the ``.cpp`` files to have a ``'using namespace
940 llvm;'`` directive at the top, after the ``#include``\s. This reduces
941 indentation in the body of the file for source editors that indent based on
942 braces, and keeps the conceptual context cleaner. The general form of this rule
943 is that any ``.cpp`` file that implements code in any namespace may use that
944 namespace (and its parents'), but should not use any others.
946 Provide a Virtual Method Anchor for Classes in Headers
947 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
949 If a class is defined in a header file and has a vtable (either it has virtual
950 methods or it derives from classes with virtual methods), it must always have at
951 least one out-of-line virtual method in the class. Without this, the compiler
952 will copy the vtable and RTTI into every ``.o`` file that ``#include``\s the
953 header, bloating ``.o`` file sizes and increasing link times.
955 Don't use default labels in fully covered switches over enumerations
956 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
958 ``-Wswitch`` warns if a switch, without a default label, over an enumeration
959 does not cover every enumeration value. If you write a default label on a fully
960 covered switch over an enumeration then the ``-Wswitch`` warning won't fire
961 when new elements are added to that enumeration. To help avoid adding these
962 kinds of defaults, Clang has the warning ``-Wcovered-switch-default`` which is
963 off by default but turned on when building LLVM with a version of Clang that
964 supports the warning.
966 A knock-on effect of this stylistic requirement is that when building LLVM with
967 GCC you may get warnings related to "control may reach end of non-void function"
968 if you return from each case of a covered switch-over-enum because GCC assumes
969 that the enum expression may take any representable value, not just those of
970 individual enumerators. To suppress this warning, use ``llvm_unreachable`` after
973 Use ``LLVM_DELETED_FUNCTION`` to mark uncallable methods
974 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
976 Prior to C++11, a common pattern to make a class uncopyable was to declare an
977 unimplemented copy constructor and copy assignment operator and make them
978 private. This would give a compiler error for accessing a private method or a
979 linker error because it wasn't implemented.
981 With C++11, we can mark methods that won't be implemented with ``= delete``.
982 This will trigger a much better error message and tell the compiler that the
983 method will never be implemented. This enables other checks like
984 ``-Wunused-private-field`` to run correctly on classes that contain these
987 To maintain compatibility with C++03, ``LLVM_DELETED_FUNCTION`` should be used
988 which will expand to ``= delete`` if the compiler supports it. These methods
989 should still be declared private. Example of the uncopyable pattern:
995 DontCopy(const DontCopy&) LLVM_DELETED_FUNCTION;
996 DontCopy &operator =(const DontCopy&) LLVM_DELETED_FUNCTION;
1001 Don't evaluate ``end()`` every time through a loop
1002 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1004 Because C++ doesn't have a standard "``foreach``" loop (though it can be
1005 emulated with macros and may be coming in C++'0x) we end up writing a lot of
1006 loops that manually iterate from begin to end on a variety of containers or
1007 through other data structures. One common mistake is to write a loop in this
1012 BasicBlock *BB = ...
1013 for (BasicBlock::iterator I = BB->begin(); I != BB->end(); ++I)
1016 The problem with this construct is that it evaluates "``BB->end()``" every time
1017 through the loop. Instead of writing the loop like this, we strongly prefer
1018 loops to be written so that they evaluate it once before the loop starts. A
1019 convenient way to do this is like so:
1023 BasicBlock *BB = ...
1024 for (BasicBlock::iterator I = BB->begin(), E = BB->end(); I != E; ++I)
1027 The observant may quickly point out that these two loops may have different
1028 semantics: if the container (a basic block in this case) is being mutated, then
1029 "``BB->end()``" may change its value every time through the loop and the second
1030 loop may not in fact be correct. If you actually do depend on this behavior,
1031 please write the loop in the first form and add a comment indicating that you
1032 did it intentionally.
1034 Why do we prefer the second form (when correct)? Writing the loop in the first
1035 form has two problems. First it may be less efficient than evaluating it at the
1036 start of the loop. In this case, the cost is probably minor --- a few extra
1037 loads every time through the loop. However, if the base expression is more
1038 complex, then the cost can rise quickly. I've seen loops where the end
1039 expression was actually something like: "``SomeMap[X]->end()``" and map lookups
1040 really aren't cheap. By writing it in the second form consistently, you
1041 eliminate the issue entirely and don't even have to think about it.
1043 The second (even bigger) issue is that writing the loop in the first form hints
1044 to the reader that the loop is mutating the container (a fact that a comment
1045 would handily confirm!). If you write the loop in the second form, it is
1046 immediately obvious without even looking at the body of the loop that the
1047 container isn't being modified, which makes it easier to read the code and
1048 understand what it does.
1050 While the second form of the loop is a few extra keystrokes, we do strongly
1053 ``#include <iostream>`` is Forbidden
1054 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1056 The use of ``#include <iostream>`` in library files is hereby **forbidden**,
1057 because many common implementations transparently inject a `static constructor`_
1058 into every translation unit that includes it.
1060 Note that using the other stream headers (``<sstream>`` for example) is not
1061 problematic in this regard --- just ``<iostream>``. However, ``raw_ostream``
1062 provides various APIs that are better performing for almost every use than
1063 ``std::ostream`` style APIs.
1067 New code should always use `raw_ostream`_ for writing, or the
1068 ``llvm::MemoryBuffer`` API for reading files.
1075 LLVM includes a lightweight, simple, and efficient stream implementation in
1076 ``llvm/Support/raw_ostream.h``, which provides all of the common features of
1077 ``std::ostream``. All new code should use ``raw_ostream`` instead of
1080 Unlike ``std::ostream``, ``raw_ostream`` is not a template and can be forward
1081 declared as ``class raw_ostream``. Public headers should generally not include
1082 the ``raw_ostream`` header, but use forward declarations and constant references
1083 to ``raw_ostream`` instances.
1088 The ``std::endl`` modifier, when used with ``iostreams`` outputs a newline to
1089 the output stream specified. In addition to doing this, however, it also
1090 flushes the output stream. In other words, these are equivalent:
1094 std::cout << std::endl;
1095 std::cout << '\n' << std::flush;
1097 Most of the time, you probably have no reason to flush the output stream, so
1098 it's better to use a literal ``'\n'``.
1103 This section describes preferred low-level formatting guidelines along with
1104 reasoning on why we prefer them.
1106 Spaces Before Parentheses
1107 ^^^^^^^^^^^^^^^^^^^^^^^^^
1109 We prefer to put a space before an open parenthesis only in control flow
1110 statements, but not in normal function call expressions and function-like
1111 macros. For example, this is good:
1116 for (I = 0; I != 100; ++I) ...
1117 while (LLVMRocks) ...
1120 assert(3 != 4 && "laws of math are failing me");
1122 A = foo(42, 92) + bar(X);
1129 for(I = 0; I != 100; ++I) ...
1130 while(LLVMRocks) ...
1133 assert (3 != 4 && "laws of math are failing me");
1135 A = foo (42, 92) + bar (X);
1137 The reason for doing this is not completely arbitrary. This style makes control
1138 flow operators stand out more, and makes expressions flow better. The function
1139 call operator binds very tightly as a postfix operator. Putting a space after a
1140 function name (as in the last example) makes it appear that the code might bind
1141 the arguments of the left-hand-side of a binary operator with the argument list
1142 of a function and the name of the right side. More specifically, it is easy to
1143 misread the "``A``" example as:
1147 A = foo ((42, 92) + bar) (X);
1149 when skimming through the code. By avoiding a space in a function, we avoid
1150 this misinterpretation.
1155 Hard fast rule: Preincrement (``++X``) may be no slower than postincrement
1156 (``X++``) and could very well be a lot faster than it. Use preincrementation
1159 The semantics of postincrement include making a copy of the value being
1160 incremented, returning it, and then preincrementing the "work value". For
1161 primitive types, this isn't a big deal. But for iterators, it can be a huge
1162 issue (for example, some iterators contains stack and set objects in them...
1163 copying an iterator could invoke the copy ctor's of these as well). In general,
1164 get in the habit of always using preincrement, and you won't have a problem.
1167 Namespace Indentation
1168 ^^^^^^^^^^^^^^^^^^^^^
1170 In general, we strive to reduce indentation wherever possible. This is useful
1171 because we want code to `fit into 80 columns`_ without wrapping horribly, but
1172 also because it makes it easier to understand the code. Namespaces are a funny
1173 thing: they are often large, and we often desire to put lots of stuff into them
1174 (so they can be large). Other times they are tiny, because they just hold an
1175 enum or something similar. In order to balance this, we use different
1176 approaches for small versus large namespaces.
1178 If a namespace definition is small and *easily* fits on a screen (say, less than
1179 35 lines of code), then you should indent its body. Here's an example:
1185 /// \brief An enum for the x86 relocation codes. Note that
1186 /// the terminology here doesn't follow x86 convention - word means
1187 /// 32-bit and dword means 64-bit.
1188 enum RelocationType {
1189 /// \brief PC relative relocation, add the relocated value to
1190 /// the value already in memory, after we adjust it for where the PC is.
1191 reloc_pcrel_word = 0,
1193 /// \brief PIC base relative relocation, add the relocated value to
1194 /// the value already in memory, after we adjust it for where the
1196 reloc_picrel_word = 1,
1198 /// \brief Absolute relocation, just add the relocated value to the
1199 /// value already in memory.
1200 reloc_absolute_word = 2,
1201 reloc_absolute_dword = 3
1206 Since the body is small, indenting adds value because it makes it very clear
1207 where the namespace starts and ends, and it is easy to take the whole thing in
1208 in one "gulp" when reading the code. If the blob of code in the namespace is
1209 larger (as it typically is in a header in the ``llvm`` or ``clang`` namespaces),
1210 do not indent the code, and add a comment indicating what namespace is being
1211 closed. For example:
1216 namespace knowledge {
1218 /// This class represents things that Smith can have an intimate
1219 /// understanding of and contains the data associated with it.
1223 explicit Grokable() { ... }
1224 virtual ~Grokable() = 0;
1230 } // end namespace knowledge
1231 } // end namespace llvm
1233 Because the class is large, we don't expect that the reader can easily
1234 understand the entire concept in a glance, and the end of the file (where the
1235 namespaces end) may be a long ways away from the place they open. As such,
1236 indenting the contents of the namespace doesn't add any value, and detracts from
1237 the readability of the class. In these cases it is best to *not* indent the
1238 contents of the namespace.
1242 Anonymous Namespaces
1243 ^^^^^^^^^^^^^^^^^^^^
1245 After talking about namespaces in general, you may be wondering about anonymous
1246 namespaces in particular. Anonymous namespaces are a great language feature
1247 that tells the C++ compiler that the contents of the namespace are only visible
1248 within the current translation unit, allowing more aggressive optimization and
1249 eliminating the possibility of symbol name collisions. Anonymous namespaces are
1250 to C++ as "static" is to C functions and global variables. While "``static``"
1251 is available in C++, anonymous namespaces are more general: they can make entire
1252 classes private to a file.
1254 The problem with anonymous namespaces is that they naturally want to encourage
1255 indentation of their body, and they reduce locality of reference: if you see a
1256 random function definition in a C++ file, it is easy to see if it is marked
1257 static, but seeing if it is in an anonymous namespace requires scanning a big
1260 Because of this, we have a simple guideline: make anonymous namespaces as small
1261 as possible, and only use them for class declarations. For example, this is
1271 bool operator<(const char *RHS) const;
1273 } // end anonymous namespace
1275 static void runHelper() {
1279 bool StringSort::operator<(const char *RHS) const {
1292 bool operator<(const char *RHS) const;
1299 bool StringSort::operator<(const char *RHS) const {
1303 } // end anonymous namespace
1305 This is bad specifically because if you're looking at "``runHelper``" in the middle
1306 of a large C++ file, that you have no immediate way to tell if it is local to
1307 the file. When it is marked static explicitly, this is immediately obvious.
1308 Also, there is no reason to enclose the definition of "``operator<``" in the
1309 namespace just because it was declared there.
1314 A lot of these comments and recommendations have been culled for other sources.
1315 Two particularly important books for our work are:
1318 <http://www.amazon.com/Effective-Specific-Addison-Wesley-Professional-Computing/dp/0321334876>`_
1319 by Scott Meyers. Also interesting and useful are "More Effective C++" and
1320 "Effective STL" by the same author.
1322 #. `Large-Scale C++ Software Design
1323 <http://www.amazon.com/Large-Scale-Software-Design-John-Lakos/dp/0201633620/ref=sr_1_1>`_
1326 If you get some free time, and you haven't read them: do so, you might learn