]>
Commit | Line | Data |
---|---|---|
1 | .\" Copyright (c) 2017 Apple Inc. All rights reserved. | |
2 | .Dd 12 January, 2018 | |
3 | .Dt style 3 | |
4 | .Os Darwin | |
5 | .Sh NAME | |
6 | .Nm style | |
7 | .Nd C language style guide for Darwin low-level userspace projects | |
8 | .Sh DESCRIPTION | |
9 | This style's primary objective is to be as friendly to the code review process | |
10 | as possible. Therefore, the style aims to ensure that code changes to the | |
11 | project produce diffs that are | |
12 | .Pp | |
13 | .Bl -bullet -compact -offset indent | |
14 | .It | |
15 | small | |
16 | .It | |
17 | unambiguous | |
18 | .It | |
19 | viewable in side-by-side comparison tools | |
20 | .El | |
21 | .Pp | |
22 | As a secondary objective, this style also aims to make code as clear as possible | |
23 | for an uninitiated programmer reading it. "Clever" syntactic shortcuts are | |
24 | actively discouraged in favor of code that is easy to understand, even if it is | |
25 | less concise. Coincidentally, this practice also tends to lend itself to | |
26 | generating more readable diffs. | |
27 | .Pp | |
28 | Like any style, consistent adherence across a project is a virtue in and of | |
29 | itself, resulting in less distraction for the reader. However, these guidelines | |
30 | should be taken as exactly that: guidelines. No style can be completely adhered | |
31 | to all the time. When you have convinced yourself that a deviation from the | |
32 | style is called for, just make sure it is for the greater good and maintains the | |
33 | style's aims of minimizing diffs and code complexity. | |
34 | .Sh GENERAL PRINCIPLES | |
35 | .Ss Vertical space is a commodity | |
36 | Scrolling vertically has always been easier than scrolling horizontally. | |
37 | Computer mouse manufacturers went so far as to dedicate hardware to the task of | |
38 | scrolling vertically when they came up with scroll wheels. Even on modern | |
39 | trackpads, while it is possible to scroll horizontally, it is far easier to | |
40 | scroll vertically. You just flick upwards. Do not be afraid to introduce extra | |
41 | lines of code if it will result in clearer, more human-parseable diffs when | |
42 | those lines are changed. | |
43 | .Ss Horizontal space is precious | |
44 | Scrolling horizontally is typically awkward, imprecise, and does not lend itself | |
45 | well toward reading on computers or even in print. (Academic journals frequently | |
46 | publish with two narrow columns per page to make reading easier, for example.) | |
47 | Lines should be wrapped consciously; this should not be left to the editor. A | |
48 | soft-wrapping scheme that looks good in your editor may not look good in someone | |
49 | else's editor (or with a different configuration for the same editor). | |
50 | .Pp | |
51 | Just as natural language comments are difficult to read in one, long line, | |
52 | so too are lines of code. Both natural languages and programming languages | |
53 | deserve conscious, deliberate wrapping to improve readability. | |
54 | .Pp | |
55 | Wrap at a column width narrow enough to accommodate side-by-side patch | |
56 | review. 80 is more likely to accommodate this, but 120 might be fine too. Pick a | |
57 | reasonable column width and stick to it. Think about the lines you are wrapping. | |
58 | If you have to wrap a line, do so in a way that is clear, and be willing to make | |
59 | changes to accommodate that (e.g. don't be afraid to declare a variable | |
60 | separately if having the declaration and assignment on the same line causes it | |
61 | to wrap in an unclear way). | |
62 | .Ss Indentation is for scope indication and nothing else | |
63 | Indentation's sole purpose is to indicate scope. You should not use indentation | |
64 | to align characters on two lines of source code (beyond, of course, aligning | |
65 | the first characters of each line if they are both in the same scope). | |
66 | .Pp | |
67 | Given this aspect of the style, it does not particularly matter whether the | |
68 | author chooses spaces or tabs for indentation, and therefore the style makes no | |
69 | prescription (other than to pick one and stick with it). | |
70 | .Pp | |
71 | This style also has another implication: tabs and spaces should never appear | |
72 | in sequence. Each line will be a series of tabs followed by the first character | |
73 | of code. Tabs will never appear after the initial indentation of the line. | |
74 | .Ss Don't require leaving the source to understand it | |
75 | Always think of future maintainers and document your thought process for them. | |
76 | Remember, a "future maintainer" might be you after you've forgotten why you did | |
77 | something. For non-trivial changes, you should not rely on linking to a | |
78 | ticket-tracking system to provide context and explanation for a piece of code. | |
79 | You should strive to ensure the reader of your code does not have to | |
80 | context-switch out of reading it in order to understand it. | |
81 | .Pp | |
82 | This is not to say that linking to external resources in code is bad, but | |
83 | if a change's purpose can be reasonably expressed without interrupting how the | |
84 | code reads and flows, just do it. You don't need to publish a whitepaper in | |
85 | comments, but don't just give a link or ticket number with no context. | |
86 | .Ss Each line of code should minimize entropy | |
87 | It is actually very difficult to construct a hash comparison scheme that humans | |
88 | can use without error consistently, and there have been successful social | |
89 | engineering attacks on getting humans to read two hashes that are "close enough" | |
90 | as identical. This means that humans need a lot of help telling the difference | |
91 | between two lines of text. | |
92 | .Pp | |
93 | For any expression, divide it up into fundamental atoms (variable declarations, | |
94 | conditionals, etc.) and then assign each of those atoms to its own line of code. | |
95 | If you do this, when you change a single atom, it is immediately obvious that | |
96 | .Em only | |
97 | that atom changed and nothing else did. The more atoms share lines of code, the | |
98 | more likely it is that changes to them will generate complex diffs that humans | |
99 | will have difficulty understanding. | |
100 | .Ss Don't assume a specific editor | |
101 | Assume people will be reading your code in a tool that you do not control and | |
102 | cannot influence. Try viewing your code in such a tool and make sure that it is | |
103 | understandable. If you follow the guidelines of this style, your code may appear | |
104 | different in another viewer (in terms of how many columns are required to | |
105 | display a single line), but its structure will appear identical. | |
106 | .Sh SPECIFIC GUIDELINES | |
107 | .Ss Column Width and Line Wrap | |
108 | 80 columns opens the door for a three-way, side-by-side comparison, but it could | |
109 | be impractical for a number of reasons. 120 columns should provide a nice | |
110 | balance, but really all that matters is that you pick a width and stick to it. | |
111 | .Pp | |
112 | When indenting a continuation line, indent over by two additional tabs. This | |
113 | visually separates the indented line from the next line, which may itself be | |
114 | indented. If there is an operator in the middle of the line, the operator should | |
115 | .Em not | |
116 | be wrapped to the continuation line. | |
117 | .Pp | |
118 | .Em Good | |
119 | .Bd -literal -offset indent | |
120 | if (some_condition && some_other_condition && | |
121 | yet_another_condition) { | |
122 | exit(0); | |
123 | } | |
124 | .Ed | |
125 | .Pp | |
126 | .Em Bad | |
127 | .Bd -literal -offset indent | |
128 | if (some_condition && some_other_condition && | |
129 | yet_another_condition) { | |
130 | exit(0); | |
131 | } | |
132 | ||
133 | if (some_condition && some_other_condition | |
134 | && yet_another_condition) { | |
135 | exit(0); | |
136 | } | |
137 | .Ed | |
138 | .Pp | |
139 | Notice on the good example that the | |
140 | .Ic exit(0) | |
141 | line is made obviously distinct from the indented conditions above it. It's very | |
142 | clear on quick visual inspection that it's not a part of the conditional checks. | |
143 | The | |
144 | .Ic && | |
145 | is left on the first line because, when reviewing a patch to this area, it will | |
146 | be immediately clear to the reviewer that that line continues to the next one. | |
147 | .Pp | |
148 | .Ss Avoid prettifying alignment | |
149 | Indentation is used only for indicating scope, so no consideration is given to | |
150 | visual alignment of equal signs, colons, etc. across multiple lines. | |
151 | .Pp | |
152 | .Em Good | |
153 | .Bd -literal -offset indent | |
154 | typedef enum { | |
155 | THING0 = 0, | |
156 | THING1 = 1, | |
157 | THING_THAT_IS_REALLY_LONG = 2, | |
158 | } thing_t; | |
159 | .Ed | |
160 | .Pp | |
161 | .Em Bad | |
162 | .Bd -literal -offset indent | |
163 | enum { | |
164 | THING0 = 0, | |
165 | THING1 = 1, | |
166 | THING_THAT_IS_REALLY_LONG = 2, | |
167 | }; | |
168 | .Ed | |
169 | .Pp | |
170 | This creates bloated diffs. If you have to re-align a ton of lines after you've | |
171 | added something longer, you get a bunch of whitespace diffs. So for variable | |
172 | declarations, enumerations, assignments, etc. just keep every line independent. | |
173 | .Pp | |
174 | There is one exception to this rule, and that is if you choose to define a | |
175 | flagset in terms of its raw hexadecimal values and wish to align them. In this | |
176 | case, it is a significant benefit to have these values aligned, and you may do | |
177 | so with spaces. | |
178 | .Pp | |
179 | .Em Example | |
180 | .Bd -literal -offset indent | |
181 | typedef enum { | |
182 | F_INIT = 0x00, | |
183 | F_FOO = 0x01, | |
184 | F_BARBAZ = 0x02, | |
185 | F_CAD = 0x04, | |
186 | F_FAD = 0x08, | |
187 | F_FUD = 0x10, | |
188 | F_FLAME = 0x20, | |
189 | F_FOOD = 0x40, | |
190 | } flag_t; | |
191 | .Ed | |
192 | .Ss Only one blank line at a time | |
193 | Use blank lines to separate logical chunks of code. Do not use more than one. | |
194 | .Ss Initialization | |
195 | C99 has named initializers for structures. Prefer those to initializing members | |
196 | one-by-one later on. Both structures and arrays should be initialized in the | |
197 | same style, with each element of the initializer being on its own line. This is | |
198 | so that when an element is added to or removed from the initialization list, | |
199 | that change gets its own line of diff. | |
200 | .Pp | |
201 | The exception to this is the string literal. | |
202 | .Pp | |
203 | .Em Good | |
204 | .Bd -literal -offset indent | |
205 | struct my_struct baz = { | |
206 | .ms_foo = 1, | |
207 | .ms_bar = NULL, | |
208 | }; | |
209 | ||
210 | char *strings[] = { | |
211 | "string", | |
212 | "other string", | |
213 | }; | |
214 | .Ed | |
215 | .Em Bad | |
216 | .Bd -literal -offset indent | |
217 | struct my_struct baz = { 1, NULL }; | |
218 | ||
219 | struct my_struct baz = { | |
220 | 1, | |
221 | NULL | |
222 | }; | |
223 | ||
224 | struct my_struct baz = { .ms_foo = 1, .ms_bar = NULL, }; | |
225 | .Ed | |
226 | .Pp | |
227 | The last element of an initializer list should be followed by a comma. This is | |
228 | so that when you add a new element to that list, it's a one-line diff rather | |
229 | rather than a two-line diff (one line of diff to add the | |
230 | .Ic , | |
231 | to the previous-last element, and another line of diff to add the new-last | |
232 | element). | |
233 | .Pp | |
234 | .Em Good | |
235 | .Bd -literal -offset indent | |
236 | enum { | |
237 | THING0, | |
238 | THING1, | |
239 | }; | |
240 | ||
241 | struct my_point p = { | |
242 | .x = 1, | |
243 | .y = 0, | |
244 | .z = 1, | |
245 | }; | |
246 | .Ed | |
247 | .Pp | |
248 | .Em Bad | |
249 | .Bd -literal -offset indent | |
250 | enum { | |
251 | THING0, THING1, | |
252 | }; | |
253 | ||
254 | enum { | |
255 | THING0, | |
256 | THING1 | |
257 | }; | |
258 | ||
259 | struct my_point p = { .x = 1, .y = 0, .z = 1 }; | |
260 | .Ed | |
261 | .Pp | |
262 | Note that, if your project requires ANSI C compliance, you should disregard this | |
263 | guideline, as it will not work under C89. | |
264 | .Ss Avoid function name overloading | |
265 | The | |
266 | .Xr clang 1 | |
267 | compiler supports extensions to the C language which allow for function name | |
268 | overloading. Name overloading generally leads to code which is difficult to | |
269 | read and introspect and should be avoided. | |
270 | .Ss Prefix `struct` members | |
271 | Any | |
272 | .Ic struct | |
273 | which is shared or exported should have a common prefix for each member. This | |
274 | helps avoid collisions with preprocessor macros. | |
275 | .Pp | |
276 | .Em Good | |
277 | .Bd -literal -offset indent | |
278 | struct foobar { | |
279 | int64_t fb_baz; | |
280 | char *fb_string; | |
281 | }; | |
282 | .Ed | |
283 | .Pp | |
284 | .Em Bad | |
285 | .Bd -literal -offset indent | |
286 | struct foobar { | |
287 | int64_t baz; | |
288 | char *string; | |
289 | }; | |
290 | .Ed | |
291 | .Pp | |
292 | .Ss Types end with `_t` | |
293 | A type is indicated with | |
294 | .Ic _t | |
295 | at the end of the | |
296 | .Ic typedef , | |
297 | whether the type refers to a | |
298 | .Ic struct , | |
299 | .Ic union , | |
300 | .Ic enum , | |
301 | etc. All types are indicated this way. Types are in all lower-case letters. | |
302 | .Pp | |
303 | .Em Good | |
304 | .Bd -literal -offset indent | |
305 | typedef uint64_t handle_t; | |
306 | typedef enum foo foo_t; | |
307 | typedef union bar bar_t; | |
308 | .Ed | |
309 | .Pp | |
310 | .Em Bad | |
311 | .Bd -literal -offset indent | |
312 | typedef uint64_t Handle; | |
313 | typedef enum foo foo_e; | |
314 | typedef union bar bar_u; | |
315 | .Ed | |
316 | .Ss Use explicitly-sized integer types | |
317 | Avoid integer types whose names do not indicate size, such as | |
318 | .Ic int | |
319 | or | |
320 | .Ic long . | |
321 | Instead, use the types from | |
322 | .Ic stdint.h | |
323 | (e.g. | |
324 | .Ic int64_t , | |
325 | .Ic uint32_t , | |
326 | etc.), which explicitly indicate size. You may use size-ambiguous integer types | |
327 | if an API requires it. | |
328 | .Ss Use `sizeof()` on variables rather than types where appropriate | |
329 | The | |
330 | .Ic sizeof() | |
331 | operator can take both types and variables as arguments. Where possible and | |
332 | relevant, always pass a variable. This ensures that if the variable's type | |
333 | changes, the proper size is used automatically. | |
334 | .Pp | |
335 | .Em Good | |
336 | .Bd -literal -offset indent | |
337 | uuid_t ident; | |
338 | memcpy(ident, buff, sizeof(ident)); | |
339 | .Ed | |
340 | .Pp | |
341 | .Em Bad | |
342 | .Bd -literal -offset indent | |
343 | uuid_t ident; | |
344 | memcpy(ident, buff, sizeof(uuid_t)); | |
345 | .Ed | |
346 | .Pp | |
347 | .Em IMPORTANT : | |
348 | When applied to a | |
349 | .Ic char * , | |
350 | .Ic sizeof() | |
351 | will return the width of a pointer, | |
352 | .Em not | |
353 | the size of the string literal it points to, so take care to only use | |
354 | .Xr strlen 3 | |
355 | for such cases. | |
356 | .Pp | |
357 | Relatedly, when applied to an array variable that is a parameter in a function's | |
358 | parameter list, | |
359 | .Ic sizeof() | |
360 | will return the width of a pointer, | |
361 | .Em not | |
362 | the size of the type. | |
363 | .Pp | |
364 | .Em Good | |
365 | .Bd -literal -offset indent | |
366 | char *string = "the quick brown fox"; | |
367 | size_t len = strlen(string); | |
368 | ||
369 | void | |
370 | foo(uuid_t u) | |
371 | { | |
372 | uuid_t u2; | |
373 | memcpy(u2, u, sizeof(uuid_t)); | |
374 | } | |
375 | .Ed | |
376 | .Pp | |
377 | .Em Bad | |
378 | .Bd -literal -offset indent | |
379 | char *string = "the quick brown fox"; | |
380 | size_t len = sizeof(string) - 1; | |
381 | ||
382 | void | |
383 | foo(uuid_t u) | |
384 | { | |
385 | uuid_t u2; | |
386 | ||
387 | // sizeof(u) == sizeof(void *) in this context. | |
388 | memcpy(u2, u, sizeof(u)); | |
389 | } | |
390 | .Ed | |
391 | .Ss Functions which take no parameters have a parameter list of `void` | |
392 | In C, an empty function parameter list means that | |
393 | .Em any | |
394 | set of parameters is acceptable. In virtually all cases where you do this, you | |
395 | mean to have a parameter list of | |
396 | .Ic void . | |
397 | .Pp | |
398 | .Em Good | |
399 | .Bd -literal -offset indent | |
400 | void | |
401 | foo(void) | |
402 | { | |
403 | do_a_thing_without_arguments(); | |
404 | } | |
405 | .Ed | |
406 | .Pp | |
407 | .Em Bad | |
408 | .Bd -literal -offset indent | |
409 | void | |
410 | foo() | |
411 | { | |
412 | do_a_thing_without_arguments(); | |
413 | } | |
414 | .Ed | |
415 | .Ss Preprocessor macros | |
416 | Preprocessor definitions are written in all-caps. Macros which are function-like | |
417 | may be lower-case provided they do not double-evaluate their arguments. | |
418 | Function-like macros that do double-evaluate their arguments should be in | |
419 | all-caps. | |
420 | .Pp | |
421 | .Em Good | |
422 | .Bd -literal -offset indent | |
423 | #define FOO 1 | |
424 | #define halt() abort() | |
425 | ||
426 | // Does not double-evaluate _a and _b such that max(i++, j) is safe. | |
427 | #define max(_a, _b) ({ \\ | |
428 | typeof(_a) a1 = (_a); \\ | |
429 | typeof(_b) b1 = (_b); \\ | |
430 | (a1 < b1 ? b1 : a1); \\ | |
431 | }) | |
432 | ||
433 | // Double-evaluates _a and _b, so MAX(i++, j) is not safe. | |
434 | #define MAX(_a, _b) ((_a) < (_b) ? (_b) : (_a)) | |
435 | .Ed | |
436 | .Pp | |
437 | .Em Bad | |
438 | .Bd -literal -offset indent | |
439 | #define kFoo 1 | |
440 | ||
441 | // Double-evaluates _a and _b. | |
442 | #define max(_a, _b) ((_a) < (_b) ? (_b) : (_a)) | |
443 | .Ed | |
444 | .Pp | |
445 | Where possible, you should prefer inline functions to preprocessor macros, or | |
446 | split a macro into a preprocessor piece and an inline function piece. | |
447 | .Pp | |
448 | .Em Example | |
449 | .Bd -literal -offset indent | |
450 | static inline void | |
451 | _debug_uint64_impl(const char *name, uint64_t val) | |
452 | { | |
453 | fprintf(stderr, "%s = %llu\\n", name, val); | |
454 | } | |
455 | ||
456 | #define debug_uint64(_v) do { \\ | |
457 | _debug_uint64_impl(#_v, _v); \\ | |
458 | } while (0) | |
459 | .Ed | |
460 | .Pp | |
461 | In this example, the preprocessor is used to do something that only the | |
462 | preprocessor can do: stringify the input variable's name. But once that work is | |
463 | done, the actual logging of the value is done by an inline function. This keeps | |
464 | the code much more readable. | |
465 | .Ss Preprocessor macro parameters should be distinguished | |
466 | Preprocessor macro expansion can run afoul of naming collisions with other | |
467 | variables that are in the same scope as the macro being expanded. To help avoid | |
468 | such collisions, parameters to preprocessor macros should have a prefix, suffix | |
469 | or both. Another good option is to use a | |
470 | .Ic _[A-Z] | |
471 | prefix, since it is reserved by the C standard and will not collide with | |
472 | preprocessor evaluation. | |
473 | .Pp | |
474 | .Em Example | |
475 | .Bd -literal -offset indent | |
476 | #define foo2(_x_) ((_x_) * 2) | |
477 | #define foo4(_x) ((_x) * 4) | |
478 | #define foo8(_X) ((_X) * 8) | |
479 | .Ed | |
480 | .Ss Preprocessor macro parameters should always be evaluated | |
481 | When passing a parameter to a preprocessor macro, it should always be referred | |
482 | to within parentheses to force evaluation. The exception is for parameters | |
483 | intended to be string literals. | |
484 | .Pp | |
485 | .Em Good | |
486 | .Bd -literal -offset indent | |
487 | #define add2(__x) ((__x) + 2) | |
488 | #define println(__fmt, ...) printf(__fmt "\\n", ## __VA_ARGS__) | |
489 | .Ed | |
490 | .Pp | |
491 | .Em Bad | |
492 | .Bd -literal -offset indent | |
493 | #define add2(__x) x + 2 | |
494 | .Ed | |
495 | .Ss Preprocessor directives always start at column 0 | |
496 | Preprocessor directives do not have scope, and therefore they always start at | |
497 | column zero. | |
498 | .Pp | |
499 | .Em Good | |
500 | .Bd -literal -offset indent | |
501 | if (do_loop) { | |
502 | for (i = 0; i < 10; i++) { | |
503 | #if CONFIG_FOOBAR | |
504 | foobar(i); | |
505 | #else | |
506 | foobaz(i); | |
507 | #endif | |
508 | } | |
509 | } | |
510 | .Ed | |
511 | .Pp | |
512 | .Em Bad | |
513 | .Bd -literal -offset indent | |
514 | if (do_loop) { | |
515 | for (i = 0; i < 10; i++) { | |
516 | #if CONFIG_FOOBAR | |
517 | foobar(i); | |
518 | #else | |
519 | foobaz(i); | |
520 | #endif | |
521 | } | |
522 | } | |
523 | .Ed | |
524 | .Ss Always reference string length directly | |
525 | Do not hard-code the size of a string. Use either | |
526 | .Ic sizeof(str) - 1 | |
527 | or | |
528 | .Ic strlen(str) . | |
529 | In the latter case, | |
530 | .Xr clang 1 | |
531 | is smart enough to recognize when a constant string is being passed to | |
532 | .Xr strlen(3) | |
533 | and replace the function call with the string's actual length. | |
534 | .Pp | |
535 | .Em Good | |
536 | .Bd -literal -offset indent | |
537 | char str[] = "string"; | |
538 | frob_string(str, sizeof(str) - 1); | |
539 | frob_string(str, strlen(str)); | |
540 | .Ed | |
541 | .Pp | |
542 | .Em Bad | |
543 | .Bd -literal -offset indent | |
544 | char str[] = "string"; | |
545 | frob_string(str, 6); | |
546 | .Ed | |
547 | .Ss Don't pointlessly validate inputs | |
548 | If you control all call sites for a function, then there is no point to | |
549 | validating the inputs to that function. If none of your call sites pass | |
550 | .Ic NULL , | |
551 | to a pointer parameter, for example, then the a | |
552 | .Ic NULL | |
553 | input indicates that there is state corruption in your address space. You may | |
554 | think that it's good to catch such corruption, but | |
555 | .Ic NULL | |
556 | is just | |
557 | .Em one | |
558 | possible invalid pointer value. What if the invalid input is | |
559 | .Ic 0x1 ? | |
560 | What if it is | |
561 | .Ic 0x2 ? | |
562 | Should you also check for those? | |
563 | .Pp | |
564 | This kind of input checking complicates code. Because it indicates state | |
565 | corruption, the only sensible thing to do in that situation would be to abort. | |
566 | But the operating system has mechanisms in place to detect the reference of an | |
567 | invalid resource, such as virtual memory and use-after-free detection. There is | |
568 | no point to you duplicating these mechanisms. | |
569 | .Pp | |
570 | Of course, you should always validate inputs | |
571 | .Em when they come from untrusted external sources | |
572 | (such as a file or IPC message), but if the inputs only ever comes from your | |
573 | program, you should trust them. | |
574 | .Pp | |
575 | .Em Good | |
576 | .Bd -literal -offset indent | |
577 | static foo_t * | |
578 | get_item(foo_t *arr, size_t idx) | |
579 | { | |
580 | return &arr[idx]; | |
581 | } | |
582 | ||
583 | int | |
584 | only_call_site(foo_t *f) | |
585 | { | |
586 | foo_t *arr = calloc(10, sizeof(arr[0])); | |
587 | if (!arr) { | |
588 | return errno; | |
589 | } | |
590 | ||
591 | *f = get_item(arr, 0); | |
592 | return 0; | |
593 | } | |
594 | .Ed | |
595 | .Pp | |
596 | .Em Bad | |
597 | .Bd -literal -offset indent | |
598 | static foo_t * | |
599 | get_item(foo_t *arr, size_t idx) | |
600 | { | |
601 | if (!arr) { | |
602 | // No point to this check since we'll abort immediately below when we | |
603 | // try to dereference `arr`. The crash report will have more than enough | |
604 | // information to diagnose the NULL pointer dereference if it ever | |
605 | // happens. | |
606 | abort(); | |
607 | } | |
608 | return &arr[idx]; | |
609 | } | |
610 | ||
611 | int | |
612 | only_call_site(foo_t *f) | |
613 | { | |
614 | foo_t *arr = calloc(10, sizeof(arr[0])); | |
615 | if (!arr) { | |
616 | return errno; | |
617 | } | |
618 | ||
619 | *f = get_item(arr, 0); | |
620 | return 0; | |
621 | } | |
622 | .Ed | |
623 | .Ss Abort on bad API inputs | |
624 | The C language provides precious few compile-time validation mechanisms, and so | |
625 | in many cases it is not possible to fully describe to the compiler the range of | |
626 | expected inputs for an API. So your API should validate input from its caller | |
627 | and abort on invalid input. Returning an error in such a case is pointless, | |
628 | since the caller probably isn't checking the return code anyway. The only sure | |
629 | way to get the programmer's attention is to abort the calling process with a | |
630 | helpful message. The | |
631 | .Ic os_crash | |
632 | routine allows you to supply such a message that the crash reporter on Darwin | |
633 | will display in its crash report. | |
634 | .Pp | |
635 | .Em Good | |
636 | .Bd -literal -offset indent | |
637 | uint8_t | |
638 | foo_a_bar(uint8_t number) | |
639 | { | |
640 | if (number > (UINT8_MAX / 2)) { | |
641 | os_crash("number given to foo_a_bar() too large"); | |
642 | } | |
643 | return (number * 2); | |
644 | } | |
645 | .Ed | |
646 | .Pp | |
647 | .Em Bad | |
648 | .Bd -literal -offset indent | |
649 | int | |
650 | foo_a_bar(uint8_t number, uint8_t *new_number) | |
651 | { | |
652 | if (number > (UINT8_MAX / 2)) { | |
653 | return EINVAL; | |
654 | } | |
655 | *new_number = (number * 2); | |
656 | return 0; | |
657 | } | |
658 | .Ed | |
659 | .Ss Don't mingle POSIX return codes and errors | |
660 | Some POSIX routines have return values that indicate whether you should check | |
661 | .Ic errno , | |
662 | and others just return the error directly. While POSIX generally documents what | |
663 | does what pretty well, there are lots of SPIs scattered around the system that | |
664 | use both conventions and aren't documented at all, leaving you to spelunk | |
665 | through the implementation to find out what's what. | |
666 | .Pp | |
667 | To avoid confusion, do not re-use the same variable for the return codes from | |
668 | these functions. If an API returns a code indicating that you should check | |
669 | .Ic errno , | |
670 | name it | |
671 | .Ic ret | |
672 | or similar. If it returns the error directly, name it | |
673 | .Ic error | |
674 | or similar and make it of type | |
675 | .Ic errno_t . | |
676 | This makes it very clear to the person reading the code that you did the work to | |
677 | find out how the API worked. By naming the variable you store the return value | |
678 | in appropriately, a reader of your code (possibly Future You) can immediately | |
679 | know what's going on. | |
680 | .Pp | |
681 | If you are making new API or SPI that returns an error code, make it return | |
682 | .Ic errno_t | |
683 | and do not use the global | |
684 | .Ic errno | |
685 | for communicating error information. | |
686 | .Pp | |
687 | .Em Good | |
688 | .Bd -literal -offset indent | |
689 | #include <sys/types.h> | |
690 | ||
691 | errno_t error = posix_spawn(NULL, "ls", NULL, NULL, argv, envp); | |
692 | switch (error) { | |
693 | case 0: | |
694 | // Handle success. | |
695 | break; | |
696 | case EACCES: | |
697 | // Handle "permission denied". | |
698 | break; | |
699 | } | |
700 | ||
701 | int ret = reboot(RB_AUTOBOOT); | |
702 | if (ret == -1) { | |
703 | switch (errno) { | |
704 | case EPERM: | |
705 | // Handle "permission denied". | |
706 | break; | |
707 | case EBUSY: | |
708 | // Handle "reboot already in progress". | |
709 | break; | |
710 | } | |
711 | } | |
712 | .Ed | |
713 | .Pp | |
714 | .Em Bad | |
715 | .Bd -literal -offset indent | |
716 | int ret = posix_spawn(NULL, "ls", NULL, NULL, argv, envp); | |
717 | switch (error) { | |
718 | case 0: | |
719 | // Handle success. | |
720 | break; | |
721 | case EACCES: | |
722 | // Handle "permission denied". | |
723 | break; | |
724 | } | |
725 | ||
726 | int error = reboot(RB_AUTOBOOT); | |
727 | if (error == -1) { | |
728 | switch (errno) { | |
729 | case EPERM: | |
730 | // Handle "permission denied". | |
731 | break; | |
732 | case EBUSY: | |
733 | // Handle "reboot already in progress". | |
734 | break; | |
735 | } | |
736 | } | |
737 | .Ed | |
738 | .Ss Avoid complex `if` statements and return distinct error codes | |
739 | Breaking up a single complex | |
740 | .Ic if | |
741 | statement | |
742 | into multiple distinct checks is both more readable and makes it possible to be | |
743 | more granular about handling failure cases. It also leads to smaller diffs if | |
744 | one of those conditions turns out to require special handling. | |
745 | .Pp | |
746 | Complex | |
747 | .Ic if | |
748 | statements are often associated with input validation and just returning an | |
749 | error code (usually | |
750 | .Ic EINVAL ) | |
751 | if any input is invalid. While deciding which error to return in which case is | |
752 | more of an art than a science, that doesn't mean you should just give up and | |
753 | return a single error every time there isn't an immediately obvious fit to the | |
754 | case you've encountered. | |
755 | .Pp | |
756 | Ideally, every case where your routine may fail should be represented by a | |
757 | distinct error code, but this is often not practical. Still, you should attempt | |
758 | to distinguish each | |
759 | .Em likely | |
760 | failure case with its own error code. The POSIX error space is fairly rich, and | |
761 | error descriptions are brief enough that they can be liberally interpreted. For | |
762 | example, | |
763 | .Ic ESRCH | |
764 | can be used to apply to any situation where a resource could not be located, not | |
765 | just conditions where there is literally "No such process". | |
766 | .Pp | |
767 | This isn't to say that you should never have compound conditions in an | |
768 | .Ic if | |
769 | statement, but the groupings should almost always be small, and the grouped | |
770 | checks should be highly likely to require change as a group when change is | |
771 | needed. | |
772 | .Pp | |
773 | .Em Good | |
774 | .Bd -literal -offset indent | |
775 | if (foo->f_int > 10 || foo->f_int < 5) | |
776 | return ERANGE; | |
777 | } | |
778 | ||
779 | if (!foo->f_uaddr) { | |
780 | return EFAULT; | |
781 | } | |
782 | ||
783 | if (foo->f_has_idx && foo->f_idx > 100) { | |
784 | return ERANGE; | |
785 | } | |
786 | ||
787 | if (foo->f_state != FS_INITIALIZED) { | |
788 | return EBUSY; | |
789 | } | |
790 | .Ed | |
791 | .Pp | |
792 | .Em Bad | |
793 | .Bd -literal -offset indent | |
794 | if (foo->f_int > 10 || foo->f_int < 5 || !foo->f_uaddr || (foo->f_has_idx && foo->f_idx > 100) || | |
795 | foo->f_state != FS_INITIALIZED) { | |
796 | return EINVAL; | |
797 | } | |
798 | .Ed | |
799 | .Pp | |
800 | See | |
801 | .Xr intro 2 , | |
802 | .Ic <sys/errno.h> , | |
803 | and | |
804 | .Ic <os/error.h> | |
805 | for the error codes supported on Darwin. | |
806 | .Ss Don't NULL-check when calling `free(3)` | |
807 | .Ic NULL | |
808 | is valid input to | |
809 | .Xr free 3 . | |
810 | It's part of the API contract. Armed with this knowledge, you can do things like | |
811 | avoid conditional memory calls, which are always weird. | |
812 | .Pp | |
813 | .Em Good | |
814 | .Bd -literal -offset indent | |
815 | char buff[1024]; | |
816 | char *ptr = buff; | |
817 | char *what2free = NULL; | |
818 | ||
819 | if (condition) { | |
820 | ptr = malloc(8); | |
821 | what2free = ptr; | |
822 | } | |
823 | ||
824 | free(what2free); | |
825 | .Ed | |
826 | .Pp | |
827 | .Em Bad | |
828 | .Bd -literal -offset indent | |
829 | char buff[1024]; | |
830 | char *ptr = buff; | |
831 | bool did_malloc = false; | |
832 | ||
833 | if (condition) { | |
834 | ptr = malloc(8); | |
835 | did_malloc = true; | |
836 | } | |
837 | ||
838 | if (did_malloc) { | |
839 | free(ptr); | |
840 | } | |
841 | .Ed | |
842 | .Ss Distinguish exported and non-exported symbols | |
843 | Any non-exported symbols should be prefixed with a | |
844 | .Ic _ . | |
845 | Thus any | |
846 | .Ic static | |
847 | functions, project-local interfaces, etc. should have this prefix. Exported | |
848 | symbols (API or SPI) should | |
849 | .Em not | |
850 | have such a prefix. | |
851 | .Pp | |
852 | .Em Good | |
853 | .Bd -literal -offset indent | |
854 | static const char *_thing = "thing"; | |
855 | static void _foo(void); | |
856 | ||
857 | void | |
858 | _project_local_interface(void); | |
859 | .Ed | |
860 | .Em Bad | |
861 | .Bd -literal -offset indent | |
862 | static const char *thing = "thing"; | |
863 | static void foo(void); | |
864 | ||
865 | void | |
866 | project_local_interface(void); | |
867 | .Ed | |
868 | .Pp | |
869 | Global variables should have a sensible prefix, preferably related to the | |
870 | project name -- e.g. globals in the | |
871 | .Xr libxpc 3 | |
872 | project are prefixed with | |
873 | .Ic xpc_ . | |
874 | .Pp | |
875 | You may also consider declaring a global structure which contains all of your | |
876 | project's shared, unexported global state. This makes it very clear when code is | |
877 | referencing that state. Also, if your project is a library at the libSystem | |
878 | layer, this is required if you are ever to adopt | |
879 | .Xr os_alloc_once 3 . | |
880 | .Pp | |
881 | .Em Example | |
882 | .Bd -literal -offset indent | |
883 | typedef struct _foobar_globals { | |
884 | uint64_t fg_global_int; | |
885 | char *fg_global_string; | |
886 | } foobar_globals_t; | |
887 | ||
888 | foobar_globals_t _g; | |
889 | foobar_globals_t *g = &_g; | |
890 | .Ed | |
891 | .Ss Distinguish SPIs meant for one caller | |
892 | Sometimes projects must create bespoke SPIs for one particular caller, and these | |
893 | SPIs are not considered suitable for general use. Append a suffix to these SPIs | |
894 | to indicate their bespokeness and the intended caller with | |
895 | .Ic _4caller . | |
896 | For example, if you add an SPI specifically for IOKit, your suffix would likely | |
897 | be | |
898 | .Ic _4IOKit . | |
899 | .Ss Use `#if` instead of `#ifdef` where appropriate | |
900 | .Ic #ifdef | |
901 | is to check if a token is | |
902 | .Em defined at all to anything. | |
903 | .Ic #if | |
904 | is to check the token's value. The C standard specifies that when a token is | |
905 | undefined, | |
906 | .Ic #if | |
907 | will evaluate it as | |
908 | .Ic 0 . | |
909 | When checking for features, it's almost always more appropriate to use | |
910 | .Ic #if | |
911 | since the lack of a feature could still be communicated by setting the token's | |
912 | value to | |
913 | .Ic 0 , | |
914 | which would pass the | |
915 | .Ic #ifdef | |
916 | check. | |
917 | .Ss Use Function Attributes from `<os/base.h>` | |
918 | If you're on Darwin, | |
919 | .Ic libplatform | |
920 | defines a lot of nice macros for compiler attributes. Use them to decorate your | |
921 | functions. This gives the compiler lots more information so it can do fancy | |
922 | optimizations. Things like | |
923 | .Ic OS_NONNULL | |
924 | let the compiler know that a parameter should never be | |
925 | .Ic NULL . | |
926 | .Ic OS_WARN_RESULT | |
927 | is great for enforcing that a caller always check the return value of a | |
928 | function. | |
929 | .Pp | |
930 | .Ic OS_MALLOC | |
931 | lets the compiler know that the function returns a heap allocation, and | |
932 | .Ic OS_OBJECT_RETURNS_RETAINED | |
933 | lets ARC know that the function returns an object with a reference that the | |
934 | caller is responsible for releasing. | |
935 | .Pp | |
936 | You can avoid having to decorate all your pointer parameters by using | |
937 | .Ic OS_ASSUME_NONNULL_BEGIN | |
938 | and | |
939 | .Ic OS_ASSUME_NONNULL_END | |
940 | and specifically annotating variables which | |
941 | .Em can | |
942 | be | |
943 | .Ic NULL | |
944 | with the | |
945 | .Ic _Nullable | |
946 | keyword. Either approach is acceptable. | |
947 | .Pp | |
948 | Generally, use these attributes on functions which will have callers who cannot | |
949 | view the implementation. Putting many of these attributes on (for example) an | |
950 | inline function is harmless, but the compiler can reason about things like | |
951 | .Ic OS_NONNULL | |
952 | and infer it when it can view the implementation at all call sites. | |
953 | .Pp | |
954 | So as a rule of thumb, if it's in a header, decorate it appropriately. These | |
955 | attributes can also serve as nice implicit documentation around API and SPI. For | |
956 | example, if you have a decoration of | |
957 | .Ic OS_NONNULL1 , | |
958 | you don't have to spell out in the HeaderDoc that you can't pass | |
959 | .Ic NULL | |
960 | for that parameter; it'll be right there in the declaration, and the compiler | |
961 | will catch attempts to do so. | |
962 | .Ss Distinguish C function definitions from declarations | |
963 | In C, make the definition of a function findable and distinguishable from its | |
964 | declaration (if any) through regular expressions. This way, you can find the | |
965 | implementation of | |
966 | .Ic foo | |
967 | by doing a regex search for | |
968 | .Ic ^foo , | |
969 | and you won't get the declaration as a result. | |
970 | .Pp | |
971 | .Em Good | |
972 | .Bd -literal -offset indent | |
973 | static int foo(int bar); | |
974 | ||
975 | int | |
976 | foo(int bar) | |
977 | { | |
978 | return bar; | |
979 | } | |
980 | .Ed | |
981 | .Pp | |
982 | .Em Bad | |
983 | .Bd -literal -offset indent | |
984 | static int foo(int bar); | |
985 | ||
986 | int foo(int bar) | |
987 | { | |
988 | return bar; | |
989 | } | |
990 | .Ed | |
991 | .Pp | |
992 | This has the additional benefit of allowing you to change the name/parameter | |
993 | list of a function independently of the return type. A diff of either will not | |
994 | be confused with the rest of the function signature. | |
995 | .Ss Use HeaderDoc for API declarations | |
996 | Make them look nice. Include the appropriate decorations (including an explicit | |
997 | export attribute such as | |
998 | .Ic OS_EXPORT | |
999 | so it's very, very clear that it's intended to be API), availability attributes, | |
1000 | and HeaderDoc. Put this stuff before the function. | |
1001 | .Pp | |
1002 | .Em Example | |
1003 | .Bd -literal -offset indent | |
1004 | /*! | |
1005 | * @function foo | |
1006 | * Returns `bar` and ignores another parameter. | |
1007 | * | |
1008 | * @param bar | |
1009 | * The value to return. | |
1010 | * | |
1011 | * @param baz | |
1012 | * The value to ignore. | |
1013 | * | |
1014 | * @result | |
1015 | * The value of `bar`. This routine cannot fail. | |
1016 | */ | |
1017 | __API_AVAILABLE(macos(10.14), ios(12.0), tvos(12.0), watchos(5.0)) | |
1018 | OS_EXPORT OS_WARN_RESULT OS_NONNULL2 | |
1019 | int | |
1020 | foo(int bar, char *baz); | |
1021 | .Ed | |
1022 | .Ss Comments | |
1023 | In general, use C++/C99-style comments. But there may be good reasons to use the | |
1024 | classic C-style comments, such as for HeaderDoc, which requires them, e.g. | |
1025 | .Bd -literal -offset indent | |
1026 | /*! | |
1027 | * Documentation | |
1028 | */ | |
1029 | .Ed | |
1030 | .Pp | |
1031 | Long, top-level comments may also use classic C-style comments. | |
1032 | .Pp | |
1033 | C++/C99-style comments may directly follow code on the same line only if they | |
1034 | are extremely brief. Otherwise, in general, comments and code should not share | |
1035 | a line. | |
1036 | .Pp | |
1037 | Also, do not get cute with | |
1038 | .Ic /* */ | |
1039 | comments and embed them within code. | |
1040 | .Pp | |
1041 | .Em Good | |
1042 | .Bd -literal -offset indent | |
1043 | // Comment on what the loop does. | |
1044 | for (i = 0; i < cnt; i++) { | |
1045 | // some code... | |
1046 | } | |
1047 | ||
1048 | /* | |
1049 | * A top-level or very long comment. | |
1050 | */ | |
1051 | ||
1052 | int ret = esoteric_spi(); // returns -1 on failure, does not set errno | |
1053 | .Ed | |
1054 | .Pp | |
1055 | .Em Bad | |
1056 | .Bd -literal -offset indent | |
1057 | //Comment | |
1058 | ||
1059 | int ret = esoteric_spi(); // This SPI returns -1 on failure but does not set | |
1060 | // errno, so here is a comment explaining that that really should be above | |
1061 | // the line of code rather than immediately following it. | |
1062 | ||
1063 | foo(arg1, /* first argument */, arg2 /* second argument */); | |
1064 | .Ed | |
1065 | .Ss `case` and `switch` are indented at the same level | |
1066 | .Ic case | |
1067 | and | |
1068 | .Ic switch | |
1069 | belong at the same column indent because indentation indicates scope, and due to | |
1070 | case fall-through, all cases are in the same scope -- one lower than the | |
1071 | previous. (Unless you scope them explicitly with braces, but you should avoid | |
1072 | doing that if at all possible.) | |
1073 | .Pp | |
1074 | .Em Good | |
1075 | .Bd -literal -offset indent | |
1076 | switch (thing) { | |
1077 | case THING1: | |
1078 | exit(0); | |
1079 | break; | |
1080 | case THING2: | |
1081 | exit(1); | |
1082 | break; | |
1083 | default: | |
1084 | __builtin_unreachable(); | |
1085 | } | |
1086 | .Ed | |
1087 | .Pp | |
1088 | .Em Bad | |
1089 | .Bd -literal -offset indent | |
1090 | switch (thing) { | |
1091 | case THING1: { | |
1092 | exit(0); | |
1093 | break; | |
1094 | } | |
1095 | case THING2: { | |
1096 | exit(1); | |
1097 | break; | |
1098 | } | |
1099 | default: | |
1100 | __builtin_unreachable(); | |
1101 | } | |
1102 | ||
1103 | switch (thing) { | |
1104 | case THING1: | |
1105 | exit(0); | |
1106 | break; | |
1107 | case THING2: | |
1108 | exit(1); | |
1109 | break; | |
1110 | default: { | |
1111 | __builtin_unreachable(); | |
1112 | } | |
1113 | } | |
1114 | .Ed | |
1115 | .Ss Use typed `enum`s | |
1116 | If you're declaring an | |
1117 | .Ic enum , | |
1118 | you should | |
1119 | .Ic typedef | |
1120 | it so the compiler can reason about valid values and know the width of the | |
1121 | .Ic enum | |
1122 | type if possible. The | |
1123 | .Ic OS_ENUM | |
1124 | macro provides the correct behavior for C, C++, and Objective-C. | |
1125 | .Ss Initialize all variables and fail closed | |
1126 | If you pre-declare a variable before using it, initialize it to a sane value. If | |
1127 | this value is something like the return value of the function, initialize it to | |
1128 | a value which indicates failure of the operation. You should | |
1129 | .Em always | |
1130 | do this even if there are no code paths which fail to initialize the variable | |
1131 | later. It's just good practice, and it gives the person reading your code an | |
1132 | indication of what ranges of values the variable is expected to hold. | |
1133 | .Pp | |
1134 | .Em Good | |
1135 | .Bd -literal -offset indent | |
1136 | int result = -1; | |
1137 | ||
1138 | if (success) { | |
1139 | result = 0; | |
1140 | } | |
1141 | .Ed | |
1142 | .Pp | |
1143 | .Em Bad | |
1144 | .Bd -literal -offset indent | |
1145 | int result; | |
1146 | ||
1147 | if (success) { | |
1148 | result = 0; | |
1149 | } | |
1150 | .Ed | |
1151 | .Pp | |
1152 | Any error variable should always be initialized to a non-success condition. In | |
1153 | general, consider success as something that your code must | |
1154 | .Em explicitly declare | |
1155 | and that the absence of such a declaration indicates failure. | |
1156 | .Pp | |
1157 | .Em Good | |
1158 | .Bd -literal -offset indent | |
1159 | int error = -1; | |
1160 | ||
1161 | if (is_root()) { | |
1162 | error = 0; | |
1163 | } else { | |
1164 | error = EPERM; | |
1165 | } | |
1166 | .Ed | |
1167 | .Pp | |
1168 | .Em Bad | |
1169 | .Bd -literal -offset indent | |
1170 | int error = 0; | |
1171 | ||
1172 | if (!is_root()) { | |
1173 | error = EPERM; | |
1174 | } | |
1175 | .Ed | |
1176 | .Pp | |
1177 | Note that you may omit an initializer for a complex | |
1178 | .Ic struct | |
1179 | type (such as the | |
1180 | .Xr stat 2 | |
1181 | .Ic struct ) | |
1182 | but then it is incumbent upon you to ensure that that variable is not used | |
1183 | uninitialized except to populate it. For many | |
1184 | .Ic struct | |
1185 | types, you can initialize them with | |
1186 | .Ic {0} . | |
1187 | This will not work for structures with nested structures though. For those you | |
1188 | can use | |
1189 | .Xr bzero 3 | |
1190 | or similar. | |
1191 | .Ss Using `goto` is fine | |
1192 | .Ic goto | |
1193 | has gotten a bad rap, but it's probably the best way in C to do lots of | |
1194 | sequential error handling. You don't | |
1195 | .Em have | |
1196 | to use | |
1197 | .Ic goto | |
1198 | if you don't want to, but if you do, just keep a a couple things in mind. | |
1199 | .Pp | |
1200 | .Bl -bullet -compact -offset indent | |
1201 | .It | |
1202 | Compile with | |
1203 | .Ic -Wsometimes-uninitialized . | |
1204 | With this warning, | |
1205 | .Xr clang 1 | |
1206 | will catch cases where a variable may be used uninitialized because a | |
1207 | .Ic goto | |
1208 | skipped the initialization. | |
1209 | .It | |
1210 | Never use | |
1211 | .Ic goto | |
1212 | as a looping construct. The C language has a few different control statements | |
1213 | for looping and iteration. Use one of those; it's not the 70's anymore. | |
1214 | .El | |
1215 | .Pp | |
1216 | These guidelines make it simple to use | |
1217 | .Ic goto | |
1218 | effectively while avoiding the | |
1219 | most common pitfalls. | |
1220 | .Ss Avoid magic Booleans | |
1221 | Sometimes you have to pass a parameter to a function to trigger some sort of | |
1222 | behavior. Avoid using a magic Boolean for these cases. Instead, use an invariant | |
1223 | that describes the behavior you are triggering. | |
1224 | .Pp | |
1225 | .Em Good | |
1226 | .Bd -literal -offset indent | |
1227 | replace_spaces(string, REPLACE_TABS_TOO); | |
1228 | replace_spaces(string, REPLACE_ONLY_SPACES); | |
1229 | .Ed | |
1230 | .Pp | |
1231 | .Em Bad | |
1232 | .Bd -literal -offset indent | |
1233 | replace_spaces(string, true); | |
1234 | replace_spaces(string, false); | |
1235 | .Ed | |
1236 | .Pp | |
1237 | If you find yourself creating many such Boolean values for function parameters, | |
1238 | you should seriously considering defining a set of flags and passing that as one | |
1239 | parameter instead. | |
1240 | .Ss Spaces around binary operators | |
1241 | In general, avoid code that looks crunched together, especially around | |
1242 | operators. Specifically: | |
1243 | .Bl -bullet -compact -offset indent | |
1244 | .It | |
1245 | Unary operators should | |
1246 | .Em not | |
1247 | have spaces around them. | |
1248 | .It | |
1249 | Binary operators | |
1250 | .Em should | |
1251 | have spaces around them. | |
1252 | .It | |
1253 | The ternary operator | |
1254 | .Em should | |
1255 | have spacing around it. | |
1256 | .El | |
1257 | .Pp | |
1258 | .Em Good | |
1259 | .Bd -literal -offset indent | |
1260 | i++; | |
1261 | j = i + k; | |
1262 | k += condition ? i : j; | |
1263 | .Ed | |
1264 | .Pp | |
1265 | .Em Bad | |
1266 | .Bd -literal -offset indent | |
1267 | i ++; | |
1268 | j=i+k | |
1269 | k+=condition?i:j; | |
1270 | .Ed | |
1271 | .Ss Reserve the ternary operator for trivial cases | |
1272 | Don't use the ternary operator to choose between complex or long expressions. | |
1273 | Reserve it for very trivial cases that are highly unlikely to change. In general | |
1274 | if you've found yourself putting the expressions in your usage of ternary | |
1275 | operator on multiple lines, you should just be using an | |
1276 | .Ic if | |
1277 | statement. | |
1278 | .Pp | |
1279 | .Em Good | |
1280 | .Bd -literal -offset indent | |
1281 | i += condition ? j : k; | |
1282 | .Ed | |
1283 | .Pp | |
1284 | .Em Bad | |
1285 | .Bd -literal -offset indent | |
1286 | i += (i < j && j > k || i == j) ? foo(bar, baz, 0, NULL) : frob(bar, 0, NULL, baz); | |
1287 | .Ed | |
1288 | .Ss Spaces around parentheses | |
1289 | .Bl -bullet -compact -offset indent | |
1290 | .It | |
1291 | Put a space between the control statement and the parenthesis indicating its | |
1292 | condition. | |
1293 | .It | |
1294 | Do | |
1295 | .Em not | |
1296 | put a space between the end of a function name and the parenthesis | |
1297 | indicating its argument list. | |
1298 | .It | |
1299 | Do | |
1300 | .Em not | |
1301 | put spaces between any parenthesis and its following content. | |
1302 | .El | |
1303 | .Pp | |
1304 | .Em Good | |
1305 | .Bd -literal -offset indent | |
1306 | if (condition) { | |
1307 | do_thing(); | |
1308 | } | |
1309 | .Ed | |
1310 | .Pp | |
1311 | .Em Bad | |
1312 | .Bd -literal -offset indent | |
1313 | if(condition) { | |
1314 | do_thing (); | |
1315 | } | |
1316 | ||
1317 | if ( condition ) { | |
1318 | do_thing ( argument ); | |
1319 | } | |
1320 | .Ed | |
1321 | .Pp | |
1322 | .Em Worse | |
1323 | .Bd -literal -offset indent | |
1324 | while( condition) { | |
1325 | do_thing( ); | |
1326 | } | |
1327 | .Ed | |
1328 | .Ss Braces and statements | |
1329 | Always, always, always use braces for your control statements. Lack of braces | |
1330 | can and has led to serious security issues that were missed during code review, | |
1331 | and putting the braces there from the start means that adding new statements to | |
1332 | that clause does not require you to also add the braces. | |
1333 | .Pp | |
1334 | The clause should be indented on the next line with no blank lines in between. | |
1335 | .Pp | |
1336 | .Em Good | |
1337 | .Bd -literal -offset indent | |
1338 | if (condition) { | |
1339 | do_thing(); | |
1340 | } | |
1341 | ||
1342 | while (condition) { | |
1343 | do_thing(); | |
1344 | } | |
1345 | .Ed | |
1346 | .Pp | |
1347 | .Em Bad | |
1348 | .Bd -literal -offset indent | |
1349 | if (condition) do_thing(); | |
1350 | ||
1351 | if (condition) | |
1352 | do_thing(); | |
1353 | ||
1354 | while (condition) do_thing(); | |
1355 | ||
1356 | while (condition) { | |
1357 | ||
1358 | do_thing(); | |
1359 | } | |
1360 | .Ed | |
1361 | .Pp | |
1362 | Even trivial uses of braceless | |
1363 | .Ic if | |
1364 | statements are problematic. Consider the following: | |
1365 | .Pp | |
1366 | .Em Bad | |
1367 | .Bd -literal -offset indent | |
1368 | if (error) i++, | |
1369 | i++; | |
1370 | .Ed | |
1371 | .Pp | |
1372 | This is admittedly contrived, but it would be likely to escape code review | |
1373 | because it's very easy to miss that the first line ends with a | |
1374 | .Ic , | |
1375 | rather than a | |
1376 | .Ic ; . | |
1377 | Braces in | |
1378 | .Ic if | |
1379 | statements are sensitive enough to security that the best policy is to simply | |
1380 | always use them, without exception. | |
1381 | .Pp | |
1382 | Specific rules for braces: | |
1383 | .Bl -bullet -compact -offset indent | |
1384 | .It | |
1385 | .Ic else | |
1386 | goes between two braces on the same line. | |
1387 | .It | |
1388 | The brace which indicates the expression associated with a control flow | |
1389 | statement goes on the same line as that statement or the same line as the last | |
1390 | continuation line of the statement. | |
1391 | .It | |
1392 | The brace which begins the definition of a | |
1393 | .Ic struct , | |
1394 | .Ic union , | |
1395 | .Ic enum , | |
1396 | etc. goes on the same line as the declaration. | |
1397 | .It | |
1398 | The brace concluding the expression associated with a control flow statement | |
1399 | is aligned with the same column as that control flow statement. | |
1400 | .It | |
1401 | The opening brace of a function definition goes on its own line and is | |
1402 | immediately followed by a new line. | |
1403 | .It | |
1404 | Control statements with empty bodies should have empty braces. | |
1405 | .El | |
1406 | .Pp | |
1407 | .Em Good | |
1408 | .Bd -literal -offset indent | |
1409 | if (condition) { | |
1410 | do_thing(); | |
1411 | } else { | |
1412 | do_other_thing(); | |
1413 | } | |
1414 | ||
1415 | void | |
1416 | function(void) | |
1417 | { | |
1418 | return; | |
1419 | } | |
1420 | ||
1421 | struct my_struct { | |
1422 | uint32_t thing; | |
1423 | }; | |
1424 | ||
1425 | for (cur; cur; cur = cur->next) { } | |
1426 | .Ed | |
1427 | .Pp | |
1428 | .Em Bad | |
1429 | .Bd -literal -offset indent | |
1430 | if (condition) | |
1431 | { | |
1432 | do_thing(); | |
1433 | } | |
1434 | else | |
1435 | { | |
1436 | do_other_thing(); | |
1437 | } | |
1438 | ||
1439 | if (condition) | |
1440 | { | |
1441 | do_thing(); | |
1442 | } | |
1443 | else | |
1444 | do_other_thing(); | |
1445 | ||
1446 | void | |
1447 | function(void) { | |
1448 | return; | |
1449 | } | |
1450 | ||
1451 | struct my_struct | |
1452 | { | |
1453 | uint32_t thing; | |
1454 | }; | |
1455 | ||
1456 | for (cur; cur; cur = cur->next) | |
1457 | .Ed | |
1458 | .Pp | |
1459 | .Em Worse | |
1460 | .Bd -literal -offset indent | |
1461 | if (condition) | |
1462 | { | |
1463 | do_thing(); | |
1464 | } | |
1465 | ||
1466 | void | |
1467 | function(void) | |
1468 | { return; | |
1469 | } | |
1470 | .Ed | |
1471 | .Sh SEE ALSO | |
1472 | .Xr style 9 , | |
1473 | .Xr intro 2 , | |
1474 | .Xr errno 3 , | |
1475 | .Xr types 5 | |
1476 | .Sh HISTORY | |
1477 | This style was largely derived from the style that evolved through the | |
1478 | .Xr launchd 8 , | |
1479 | .Xr libdispatch 3 , | |
1480 | and | |
1481 | .Xr libxpc 3 | |
1482 | projects. |