run-command: dup_devnull(): guard against syscalls failing
[git/git.git] / Documentation / CodingGuidelines
1 Like other projects, we also have some guidelines to keep to the
2 code. For git in general, three rough rules are:
3
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.
7
8 - However, we often say "Let's stay away from that construct,
9 it's not even in POSIX".
10
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
15 let's use it".
16
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.
20
21
22 As for more concrete guidelines, just imitate the existing code
23 (this is a good guideline, no matter which project you are
24 contributing to). It is always preferable to match the _local_
25 convention. New code added to git suite is expected to match
26 the overall style of existing code. Modifications to existing
27 code is expected to match the style the surrounding code already
28 uses (even if it doesn't match the overall style of existing code).
29
30 But if you must have a list of rules, here they are.
31
32 For shell scripts specifically (not exhaustive):
33
34 - We use tabs for indentation.
35
36 - Case arms are indented at the same depth as case and esac lines.
37
38 - Redirection operators should be written with space before, but no
39 space after them. In other words, write 'echo test >"$file"'
40 instead of 'echo test> $file' or 'echo test > $file'. Note that
41 even though it is not required by POSIX to double-quote the
42 redirection target in a variable (as shown above), our code does so
43 because some versions of bash issue a warning without the quotes.
44
45 - We prefer $( ... ) for command substitution; unlike ``, it
46 properly nests. It should have been the way Bourne spelled
47 it from day one, but unfortunately isn't.
48
49 - If you want to find out if a command is available on the user's
50 $PATH, you should use 'type <command>', instead of 'which <command>'.
51 The output of 'which' is not machine parseable and its exit code
52 is not reliable across platforms.
53
54 - We use POSIX compliant parameter substitutions and avoid bashisms;
55 namely:
56
57 - We use ${parameter-word} and its [-=?+] siblings, and their
58 colon'ed "unset or null" form.
59
60 - We use ${parameter#word} and its [#%] siblings, and their
61 doubled "longest matching" form.
62
63 - No "Substring Expansion" ${parameter:offset:length}.
64
65 - No shell arrays.
66
67 - No strlen ${#parameter}.
68
69 - No pattern replacement ${parameter/pattern/string}.
70
71 - We use Arithmetic Expansion $(( ... )).
72
73 - Inside Arithmetic Expansion, spell shell variables with $ in front
74 of them, as some shells do not grok $((x)) while accepting $(($x))
75 just fine (e.g. dash older than 0.5.4).
76
77 - We do not use Process Substitution <(list) or >(list).
78
79 - Do not write control structures on a single line with semicolon.
80 "then" should be on the next line for if statements, and "do"
81 should be on the next line for "while" and "for".
82
83 - We prefer "test" over "[ ... ]".
84
85 - We do not write the noiseword "function" in front of shell
86 functions.
87
88 - We prefer a space between the function name and the parentheses. The
89 opening "{" should also be on the same line.
90 E.g.: my_function () {
91
92 - As to use of grep, stick to a subset of BRE (namely, no \{m,n\},
93 [::], [==], nor [..]) for portability.
94
95 - We do not use \{m,n\};
96
97 - We do not use -E;
98
99 - We do not use ? nor + (which are \{0,1\} and \{1,\}
100 respectively in BRE) but that goes without saying as these
101 are ERE elements not BRE (note that \? and \+ are not even part
102 of BRE -- making them accessible from BRE is a GNU extension).
103
104 - Use Git's gettext wrappers in git-sh-i18n to make the user
105 interface translatable. See "Marking strings for translation" in
106 po/README.
107
108 For C programs:
109
110 - We use tabs to indent, and interpret tabs as taking up to
111 8 spaces.
112
113 - We try to keep to at most 80 characters per line.
114
115 - We try to support a wide range of C compilers to compile git with,
116 including old ones. That means that you should not use C99
117 initializers, even if a lot of compilers grok it.
118
119 - Variables have to be declared at the beginning of the block.
120
121 - NULL pointers shall be written as NULL, not as 0.
122
123 - When declaring pointers, the star sides with the variable
124 name, i.e. "char *string", not "char* string" or
125 "char * string". This makes it easier to understand code
126 like "char *string, c;".
127
128 - We avoid using braces unnecessarily. I.e.
129
130 if (bla) {
131 x = 1;
132 }
133
134 is frowned upon. A gray area is when the statement extends
135 over a few lines, and/or you have a lengthy comment atop of
136 it. Also, like in the Linux kernel, if there is a long list
137 of "else if" statements, it can make sense to add braces to
138 single line blocks.
139
140 - We try to avoid assignments inside if().
141
142 - Try to make your code understandable. You may put comments
143 in, but comments invariably tend to stale out when the code
144 they were describing changes. Often splitting a function
145 into two makes the intention of the code much clearer.
146
147 - Double negation is often harder to understand than no negation
148 at all.
149
150 - Some clever tricks, like using the !! operator with arithmetic
151 constructs, can be extremely confusing to others. Avoid them,
152 unless there is a compelling reason to use them.
153
154 - Use the API. No, really. We have a strbuf (variable length
155 string), several arrays with the ALLOC_GROW() macro, a
156 string_list for sorted string lists, a hash map (mapping struct
157 objects) named "struct decorate", amongst other things.
158
159 - When you come up with an API, document it.
160
161 - The first #include in C files, except in platform specific
162 compat/ implementations, should be git-compat-util.h or another
163 header file that includes it, such as cache.h or builtin.h.
164
165 - If you are planning a new command, consider writing it in shell
166 or perl first, so that changes in semantics can be easily
167 changed and discussed. Many git commands started out like
168 that, and a few are still scripts.
169
170 - Avoid introducing a new dependency into git. This means you
171 usually should stay away from scripting languages not already
172 used in the git core command set (unless your command is clearly
173 separate from it, such as an importer to convert random-scm-X
174 repositories to git).
175
176 - When we pass <string, length> pair to functions, we should try to
177 pass them in that order.
178
179 - Use Git's gettext wrappers to make the user interface
180 translatable. See "Marking strings for translation" in po/README.
181
182 Writing Documentation:
183
184 Every user-visible change should be reflected in the documentation.
185 The same general rule as for code applies -- imitate the existing
186 conventions. A few commented examples follow to provide reference
187 when writing or modifying command usage strings and synopsis sections
188 in the manual pages:
189
190 Placeholders are spelled in lowercase and enclosed in angle brackets:
191 <file>
192 --sort=<key>
193 --abbrev[=<n>]
194
195 Possibility of multiple occurrences is indicated by three dots:
196 <file>...
197 (One or more of <file>.)
198
199 Optional parts are enclosed in square brackets:
200 [<extra>]
201 (Zero or one <extra>.)
202
203 --exec-path[=<path>]
204 (Option with an optional argument. Note that the "=" is inside the
205 brackets.)
206
207 [<patch>...]
208 (Zero or more of <patch>. Note that the dots are inside, not
209 outside the brackets.)
210
211 Multiple alternatives are indicated with vertical bar:
212 [-q | --quiet]
213 [--utf8 | --no-utf8]
214
215 Parentheses are used for grouping:
216 [(<rev>|<range>)...]
217 (Any number of either <rev> or <range>. Parens are needed to make
218 it clear that "..." pertains to both <rev> and <range>.)
219
220 [(-p <parent>)...]
221 (Any number of option -p, each with one <parent> argument.)
222
223 git remote set-head <name> (-a | -d | <branch>)
224 (One and only one of "-a", "-d" or "<branch>" _must_ (no square
225 brackets) be provided.)
226
227 And a somewhat more contrived example:
228 --diff-filter=[(A|C|D|M|R|T|U|X|B)...[*]]
229 Here "=" is outside the brackets, because "--diff-filter=" is a
230 valid usage. "*" has its own pair of brackets, because it can
231 (optionally) be specified only when one or more of the letters is
232 also provided.