1 Like other projects, we also have some guidelines to keep to the
2 code. For Git in general, a few rough rules are:
4 - Most importantly, we never say "It's in POSIX; we'll happily
5 ignore your needs should your system not conform to it."
6 We live in the real world.
8 - However, we often say "Let's stay away from that construct,
9 it's not even in POSIX".
11 - In spite of the above two rules, we sometimes say "Although
12 this is not in POSIX, it (is so convenient | makes the code
13 much more readable | has other good characteristics) and
14 practically all the platforms we care about support it, so
17 Again, we live in the real world, and it is sometimes a
18 judgement call, the decision based more on real world
19 constraints people face than what the paper standard says.
21 - Fixing style violations while working on a real change as a
22 preparatory clean-up step is good, but otherwise avoid useless code
23 churn for the sake of conforming to the style.
25 "Once it _is_ in the tree, it's not really worth the patch noise to
27 Cf. http://lkml.iu.edu/hypermail/linux/kernel/1001.3/01069.html
29 Make your code readable and sensible, and don't try to be clever.
31 As for more concrete guidelines, just imitate the existing code
32 (this is a good guideline, no matter which project you are
33 contributing to). It is always preferable to match the _local_
34 convention. New code added to Git suite is expected to match
35 the overall style of existing code. Modifications to existing
36 code is expected to match the style the surrounding code already
37 uses (even if it doesn't match the overall style of existing code).
39 But if you must have a list of rules, here they are.
41 For shell scripts specifically (not exhaustive):
43 - We use tabs for indentation.
45 - Case arms are indented at the same depth as case and esac lines,
57 - Redirection operators should be written with space before, but no
58 space after them. In other words, write 'echo test >"$file"'
59 instead of 'echo test> $file' or 'echo test > $file'. Note that
60 even though it is not required by POSIX to double-quote the
61 redirection target in a variable (as shown above), our code does so
62 because some versions of bash issue a warning without the quotes.
65 cat hello > world < universe
69 cat hello >world <universe
72 - We prefer $( ... ) for command substitution; unlike ``, it
73 properly nests. It should have been the way Bourne spelled
74 it from day one, but unfortunately isn't.
76 - If you want to find out if a command is available on the user's
77 $PATH, you should use 'type <command>', instead of 'which <command>'.
78 The output of 'which' is not machine parsable and its exit code
79 is not reliable across platforms.
81 - We use POSIX compliant parameter substitutions and avoid bashisms;
84 - We use ${parameter-word} and its [-=?+] siblings, and their
85 colon'ed "unset or null" form.
87 - We use ${parameter#word} and its [#%] siblings, and their
88 doubled "longest matching" form.
90 - No "Substring Expansion" ${parameter:offset:length}.
94 - No pattern replacement ${parameter/pattern/string}.
96 - We use Arithmetic Expansion $(( ... )).
98 - We do not use Process Substitution <(list) or >(list).
100 - Do not write control structures on a single line with semicolon.
101 "then" should be on the next line for if statements, and "do"
102 should be on the next line for "while" and "for".
105 if test -f hello; then
115 - If a command sequence joined with && or || or | spans multiple
116 lines, put each command on a separate line and put && and || and |
117 operators at the end of each line, rather than the start. This
118 means you don't need to use \ to join lines, since the above
119 operators imply the sequence isn't finished.
122 grep blob verify_pack_result \
123 | awk -f print_1.awk \
128 grep blob verify_pack_result |
133 - We prefer "test" over "[ ... ]".
135 - We do not write the noiseword "function" in front of shell
138 - We prefer a space between the function name and the parentheses,
139 and no space inside the parentheses. The opening "{" should also
150 - As to use of grep, stick to a subset of BRE (namely, no \{m,n\},
151 [::], [==], or [..]) for portability.
153 - We do not use \{m,n\};
157 - We do not use ? or + (which are \{0,1\} and \{1,\}
158 respectively in BRE) but that goes without saying as these
159 are ERE elements not BRE (note that \? and \+ are not even part
160 of BRE -- making them accessible from BRE is a GNU extension).
162 - Use Git's gettext wrappers in git-sh-i18n to make the user
163 interface translatable. See "Marking strings for translation" in
166 - We do not write our "test" command with "-a" and "-o" and use "&&"
167 or "||" to concatenate multiple "test" commands instead, because
168 the use of "-a/-o" is often error-prone. E.g.
170 test -n "$x" -a "$a" = "$b"
172 is buggy and breaks when $x is "=", but
174 test -n "$x" && test "$a" = "$b"
176 does not have such a problem.
178 - Even though "local" is not part of POSIX, we make heavy use of it
179 in our test suite. We do not use it in scripted Porcelains, and
180 hopefully nobody starts using "local" before they are reimplemented
186 - We use tabs to indent, and interpret tabs as taking up to
189 - We try to keep to at most 80 characters per line.
191 - As a Git developer we assume you have a reasonably modern compiler
192 and we recommend you to enable the DEVELOPER makefile knob to
193 ensure your patch is clear of all compiler warnings we care about,
194 by e.g. "echo DEVELOPER=1 >>config.mak".
196 - We try to support a wide range of C compilers to compile Git with,
197 including old ones. You should not use features from newer C
198 standard, even if your compiler groks them.
200 There are a few exceptions to this guideline:
202 . since early 2012 with e1327023ea, we have been using an enum
203 definition whose last element is followed by a comma. This, like
204 an array initializer that ends with a trailing comma, can be used
205 to reduce the patch noise when adding a new identifier at the end.
207 . since mid 2017 with cbc0f81d, we have been using designated
208 initializers for struct (e.g. "struct t v = { .val = 'a' };").
210 . since mid 2017 with 512f41cf, we have been using designated
211 initializers for array (e.g. "int array[10] = { [5] = 2 }").
213 These used to be forbidden, but we have not heard any breakage
214 report, and they are assumed to be safe.
216 - Variables have to be declared at the beginning of the block, before
217 the first statement (i.e. -Wdeclaration-after-statement).
219 - Declaring a variable in the for loop "for (int i = 0; i < 10; i++)"
220 is still not allowed in this codebase.
222 - NULL pointers shall be written as NULL, not as 0.
224 - When declaring pointers, the star sides with the variable
225 name, i.e. "char *string", not "char* string" or
226 "char * string". This makes it easier to understand code
227 like "char *string, c;".
229 - Use whitespace around operators and keywords, but not inside
230 parentheses and not around functions. So:
240 - Do not explicitly compare an integral value with constant 0 or '\0',
241 or a pointer value with constant NULL. For instance, to validate that
242 counted array <ptr, cnt> is initialized but has no elements, write:
245 BUG("empty array expected");
249 if (ptr == NULL || cnt != 0);
250 BUG("empty array expected");
252 - We avoid using braces unnecessarily. I.e.
258 is frowned upon. But there are a few exceptions:
260 - When the statement extends over a few lines (e.g., a while loop
261 with an embedded conditional, or a comment). E.g.:
272 * This one requires some explanation,
273 * so we're better off with braces to make
274 * it obvious that the indentation is correct.
279 - When there are multiple arms to a conditional and some of them
280 require braces, enclose even a single line block in braces for
291 - We try to avoid assignments in the condition of an "if" statement.
293 - Try to make your code understandable. You may put comments
294 in, but comments invariably tend to stale out when the code
295 they were describing changes. Often splitting a function
296 into two makes the intention of the code much clearer.
298 - Multi-line comments include their delimiters on separate lines from
303 * multi-line comment.
306 Note however that a comment that explains a translatable string to
307 translators uses a convention of starting with a magic token
308 "TRANSLATORS: ", e.g.
311 * TRANSLATORS: here is a comment that explains the string to
312 * be translated, that follows immediately after it.
314 _("Here is a translatable string explained by the above.");
316 - Double negation is often harder to understand than no negation
319 - There are two schools of thought when it comes to comparison,
320 especially inside a loop. Some people prefer to have the less stable
321 value on the left hand side and the more stable value on the right hand
322 side, e.g. if you have a loop that counts variable i down to the
325 while (i > lower_bound) {
330 Other people prefer to have the textual order of values match the
331 actual order of values in their comparison, so that they can
332 mentally draw a number line from left to right and place these
333 values in order, i.e.
335 while (lower_bound < i) {
340 Both are valid, and we use both. However, the more "stable" the
341 stable side becomes, the more we tend to prefer the former
342 (comparison with a constant, "i > 0", is an extreme example).
343 Just do not mix styles in the same part of the code and mimic
344 existing styles in the neighbourhood.
346 - There are two schools of thought when it comes to splitting a long
347 logical line into multiple lines. Some people push the second and
348 subsequent lines far enough to the right with tabs and align them:
350 if (the_beginning_of_a_very_long_expression_that_has_to ||
351 span_more_than_a_single_line_of ||
355 while other people prefer to align the second and the subsequent
356 lines with the column immediately inside the opening parenthesis,
357 with tabs and spaces, following our "tabstop is always a multiple
360 if (the_beginning_of_a_very_long_expression_that_has_to ||
361 span_more_than_a_single_line_of ||
365 Both are valid, and we use both. Again, just do not mix styles in
366 the same part of the code and mimic existing styles in the
369 - When splitting a long logical line, some people change line before
370 a binary operator, so that the result looks like a parse tree when
371 you turn your head 90-degrees counterclockwise:
373 if (the_beginning_of_a_very_long_expression_that_has_to
374 || span_more_than_a_single_line_of_the_source_text) {
376 while other people prefer to leave the operator at the end of the
379 if (the_beginning_of_a_very_long_expression_that_has_to ||
380 span_more_than_a_single_line_of_the_source_text) {
382 Both are valid, but we tend to use the latter more, unless the
383 expression gets fairly complex, in which case the former tends to
384 be easier to read. Again, just do not mix styles in the same part
385 of the code and mimic existing styles in the neighbourhood.
387 - When splitting a long logical line, with everything else being
388 equal, it is preferable to split after the operator at higher
389 level in the parse tree. That is, this is more preferable:
391 if (a_very_long_variable * that_is_used_in +
392 a_very_long_expression) {
397 if (a_very_long_variable *
398 that_is_used_in + a_very_long_expression) {
401 - Some clever tricks, like using the !! operator with arithmetic
402 constructs, can be extremely confusing to others. Avoid them,
403 unless there is a compelling reason to use them.
405 - Use the API. No, really. We have a strbuf (variable length
406 string), several arrays with the ALLOC_GROW() macro, a
407 string_list for sorted string lists, a hash map (mapping struct
408 objects) named "struct decorate", amongst other things.
410 - When you come up with an API, document its functions and structures
411 in the header file that exposes the API to its callers. Use what is
412 in "strbuf.h" as a model for the appropriate tone and level of
415 - The first #include in C files, except in platform specific compat/
416 implementations, must be either "git-compat-util.h", "cache.h" or
417 "builtin.h". You do not have to include more than one of these.
419 - A C file must directly include the header files that declare the
420 functions and the types it uses, except for the functions and types
421 that are made available to it by including one of the header files
422 it must include by the previous rule.
424 - If you are planning a new command, consider writing it in shell
425 or perl first, so that changes in semantics can be easily
426 changed and discussed. Many Git commands started out like
427 that, and a few are still scripts.
429 - Avoid introducing a new dependency into Git. This means you
430 usually should stay away from scripting languages not already
431 used in the Git core command set (unless your command is clearly
432 separate from it, such as an importer to convert random-scm-X
433 repositories to Git).
435 - When we pass <string, length> pair to functions, we should try to
436 pass them in that order.
438 - Use Git's gettext wrappers to make the user interface
439 translatable. See "Marking strings for translation" in po/README.
441 - Variables and functions local to a given source file should be marked
442 with "static". Variables that are visible to other source files
443 must be declared with "extern" in header files. However, function
444 declarations should not use "extern", as that is already the default.
446 - You can launch gdb around your program using the shorthand GIT_DEBUGGER.
447 Run `GIT_DEBUGGER=1 ./bin-wrappers/git foo` to simply use gdb as is, or
448 run `GIT_DEBUGGER="<debugger> <debugger-args>" ./bin-wrappers/git foo` to
449 use your own debugger and arguments. Example: `GIT_DEBUGGER="ddd --gdb"
450 ./bin-wrappers/git log` (See `wrap-for-bin.sh`.)
454 - Most of the C guidelines above apply.
456 - We try to support Perl 5.8 and later ("use Perl 5.008").
458 - use strict and use warnings are strongly preferred.
460 - Don't overuse statement modifiers unless using them makes the
461 result easier to follow.
464 do_this() unless (condition);
465 ... do something else ...
467 is more readable than:
473 ... do something else ...
475 *only* when the condition is so rare that do_this() will be almost
478 - We try to avoid assignments inside "if ()" conditions.
480 - Learn and use Git.pm if you need that functionality.
482 - For Emacs, it's useful to put the following in
483 GIT_CHECKOUT/.dir-locals.el, assuming you use cperl-mode:
485 ;; note the first part is useful for C editing, too
486 ((nil . ((indent-tabs-mode . t)
489 (cperl-mode . ((cperl-indent-level . 8)
490 (cperl-extra-newline-before-brace . nil)
491 (cperl-merge-trailing-else . t))))
495 - We follow PEP-8 (http://www.python.org/dev/peps/pep-0008/).
497 - As a minimum, we aim to be compatible with Python 2.7.
499 - Where required libraries do not restrict us to Python 2, we try to
500 also be compatible with Python 3.1 and later.
505 We make a distinction between a Git command's primary output and
506 output which is merely chatty feedback (for instance, status
507 messages, running transcript, or progress display), as well as error
508 messages. Roughly speaking, a Git command's primary output is that
509 which one might want to capture to a file or send down a pipe; its
510 chatty output should not interfere with these use-cases.
512 As such, primary output should be sent to the standard output stream
513 (stdout), and chatty output should be sent to the standard error
514 stream (stderr). Examples of commands which produce primary output
515 include `git log`, `git show`, and `git branch --list` which generate
516 output on the stdout stream.
518 Not all Git commands have primary output; this is often true of
519 commands whose main function is to perform an action. Some action
520 commands are silent, whereas others are chatty. An example of a
521 chatty action commands is `git clone` with its "Cloning into
522 '<path>'..." and "Checking connectivity..." status messages which it
523 sends to the stderr stream.
525 Error messages from Git commands should always be sent to the stderr
531 - Do not end error messages with a full stop.
533 - Do not capitalize the first word, only because it is the first word
534 in the message ("unable to open %s", not "Unable to open %s"). But
535 "SHA-3 not supported" is fine, because the reason the first word is
536 capitalized is not because it is at the beginning of the sentence,
537 but because the word would be spelled in capital letters even when
538 it appeared in the middle of the sentence.
540 - Say what the error is first ("cannot open %s", not "%s: cannot open")
543 Externally Visible Names
545 - For configuration variable names, follow the existing convention:
547 . The section name indicates the affected subsystem.
549 . The subsection name, if any, indicates which of an unbounded set
550 of things to set the value for.
552 . The variable name describes the effect of tweaking this knob.
554 The section and variable names that consist of multiple words are
555 formed by concatenating the words without punctuations (e.g. `-`),
556 and are broken using bumpyCaps in documentation as a hint to the
559 When choosing the variable namespace, do not use variable name for
560 specifying possibly unbounded set of things, most notably anything
561 an end user can freely come up with (e.g. branch names). Instead,
562 use subsection names or variable values, like the existing variable
563 branch.<name>.description does.
566 Writing Documentation:
568 Most (if not all) of the documentation pages are written in the
569 AsciiDoc format in *.txt files (e.g. Documentation/git.txt), and
570 processed into HTML and manpages (e.g. git.html and git.1 in the
573 The documentation liberally mixes US and UK English (en_US/UK)
574 norms for spelling and grammar, which is somewhat unfortunate.
575 In an ideal world, it would have been better if it consistently
576 used only one and not the other, and we would have picked en_US
577 (if you wish to correct the English of some of the existing
578 documentation, please see the documentation-related advice in the
579 Documentation/SubmittingPatches file).
581 In order to ensure the documentation is inclusive, avoid assuming
582 that an unspecified example person is male or female, and think
583 twice before using "he", "him", "she", or "her". Here are some
584 tips to avoid use of gendered pronouns:
586 - Prefer succinctness and matter-of-factly describing functionality
587 in the abstract. E.g.
589 --short:: Emit output in the short-format.
591 and avoid something like these overly verbose alternatives:
593 --short:: Use this to emit output in the short-format.
594 --short:: You can use this to get output in the short-format.
595 --short:: A user who prefers shorter output could....
596 --short:: Should a person and/or program want shorter output, he
599 This practice often eliminates the need to involve human actors in
600 your description, but it is a good practice regardless of the
601 avoidance of gendered pronouns.
603 - When it becomes awkward to stick to this style, prefer "you" when
604 addressing the the hypothetical user, and possibly "we" when
605 discussing how the program might react to the user. E.g.
607 You can use this option instead of --xyz, but we might remove
608 support for it in future versions.
610 while keeping in mind that you can probably be less verbose, e.g.
612 Use this instead of --xyz. This option might be removed in future
615 - If you still need to refer to an example person that is
616 third-person singular, you may resort to "singular they" to avoid
617 "he/she/him/her", e.g.
619 A contributor asks their upstream to pull from them.
621 Note that this sounds ungrammatical and unnatural to those who
622 learned that "they" is only used for third-person plural, e.g.
623 those who learn English as a second language in some parts of the
626 Every user-visible change should be reflected in the documentation.
627 The same general rule as for code applies -- imitate the existing
630 A few commented examples follow to provide reference when writing or
631 modifying command usage strings and synopsis sections in the manual
634 Placeholders are spelled in lowercase and enclosed in angle brackets:
639 If a placeholder has multiple words, they are separated by dashes:
641 --template=<template-directory>
643 Possibility of multiple occurrences is indicated by three dots:
645 (One or more of <file>.)
647 Optional parts are enclosed in square brackets:
649 (Zero or one <extra>.)
652 (Option with an optional argument. Note that the "=" is inside the
656 (Zero or more of <patch>. Note that the dots are inside, not
657 outside the brackets.)
659 Multiple alternatives are indicated with vertical bars:
663 Parentheses are used for grouping:
664 [(<rev> | <range>)...]
665 (Any number of either <rev> or <range>. Parens are needed to make
666 it clear that "..." pertains to both <rev> and <range>.)
669 (Any number of option -p, each with one <parent> argument.)
671 git remote set-head <name> (-a | -d | <branch>)
672 (One and only one of "-a", "-d" or "<branch>" _must_ (no square
673 brackets) be provided.)
675 And a somewhat more contrived example:
676 --diff-filter=[(A|C|D|M|R|T|U|X|B)...[*]]
677 Here "=" is outside the brackets, because "--diff-filter=" is a
678 valid usage. "*" has its own pair of brackets, because it can
679 (optionally) be specified only when one or more of the letters is
683 Use 'git' (all lowercase) when talking about commands i.e. something
684 the user would type into a shell and use 'Git' (uppercase first letter)
685 when talking about the version control system and its properties.
687 A few commented examples follow to provide reference when writing or
688 modifying paragraphs or option/command explanations that contain options
691 Literal examples (e.g. use of command-line options, command names,
692 branch names, URLs, pathnames (files and directories), configuration and
693 environment variables) must be typeset in monospace (i.e. wrapped with
698 `http://git.example.com`
703 An environment variable must be prefixed with "$" only when referring to its
704 value and not when referring to the variable itself, in this case there is
705 nothing to add except the backticks:
706 `GIT_DIR` is specified
707 `$GIT_DIR/hooks/pre-receive`
709 Word phrases enclosed in `backtick characters` are rendered literally
710 and will not be further expanded. The use of `backticks` to achieve the
711 previous rule means that literal examples should not use AsciiDoc
718 If some place in the documentation needs to typeset a command usage
719 example with inline substitutions, it is fine to use +monospaced and
720 inline substituted text+ instead of `monospaced literal text`, and with
721 the former, the part that should not get substituted must be