1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
|
Cairo coding style.
This document is intended to be a short description of the preferred
coding style for the cairo source code. Good style requires good
taste, which means this can't all be reduced to automated rules, and
there are exceptions.
We want the code to be easy to understand and maintain, and consistent
style plays an important part in that, even if some of the specific
details seem trivial. If nothing else, this document gives a place to
put consistent answers for issues that would otherwise be arbitrary.
Most of the guidelines here are demonstrated by examples, (which means
this document is quicker to read than it might appear given its
length). Most of the examples are positive examples that you should
imitate. The few negative examples are clearly marked with a comment
of /* Yuck! */. Please don't submit code to cairo that looks like any
of these.
Indentation
-----------
Each new level is indented 4 more spaces than the previous level:
if (condition)
do_something ();
This may be achieved with space characters or a combination of tab
characters and space characters. It may not be achieved with tab
characters exclusively (see below).
Tab characters
--------------
The tab character must always be interpreted according to its
traditional meaning:
Advance to the next column which is a multiple of 8.
With this definition, even levels of indentation can be achieved with
a sequence of tab characters, while odd levels of indentation may
begin with a sequence of tab character but must end with 4 space
characters.
Some programmers have been misled by certain text editors into
thinking that 4-space indentation can be achieved with tab characters
exclusively by changing the meaning of tab character to be "advance to
the next column which is a multiple of 4". Code formatted in this way,
making an assumption of a fictitious 4-character-tab will not be
accepted into cairo.
The rationale here is that tabs are used in the code for lining things
up other than indentation, (see the Whitespace section below), and
changing the interpretation of tab from its traditional meaning will
break this alignment.
Braces
------
Most of the code in cairo uses bracing in the style of K&R:
if (condition) {
do_this ();
do_that ();
} else {
do_the_other ();
}
but some of the code uses an alternate style:
if (condition)
{
do_this ();
do_that ();
}
else
{
do_the_other ();
}
and that seems just fine. We won't lay down any strict rule on this
point, (though there should be some local consistency). If you came
here hoping to find some guidance, then use the first form above.
If all of the substatements of an if statement are single statements,
the optional braces should not usually appear:
if (condition)
do_this ();
else
do_that ();
But the braces are mandatory when mixing single statement and compound
statements in the various clauses. For example, do not do this:
if (condition) {
do_this ();
do_that ();
} else /* Yuck! */
do_the_other ();
And of course, there are exceptions for when the code just looks
better with the braces:
if (condition) {
/* Note that we have to be careful here. */
do_something_dangerous (with_care);
}
if (condition &&
other_condition &&
yet_another)
{
do_something ();
}
And note that this last example also shows a situation in which the
opening brace really needs to be on its own line. The following looks awful:
if (condition &&
other_condition &&
yet_another) { /* Yuck! */
do_something ();
}
As we said above, legible code that is easy to understand and maintain
is the goal, not adherence to strict rules.
Whitespace
----------
Separate logically distinct chunks with a single newline. This
obviously applies between functions, but also applies within a
function or block and can even be used to good effect within a
structure definition:
struct _cairo_gstate {
cairo_operator_t op;
double tolerance;
/* stroke style */
double line_width;
cairo_line_cap_t line_cap;
cairo_line_join_t line_join;
double miter_limit;
cairo_fill_rule_t fill_rule;
double *dash;
int num_dashes;
double dash_offset;
...
}
Use a single space before a left parenthesis, except where the
standard will not allow it, (eg. when defining a parameterized macro).
Don't eliminate newlines just because things would still fit on one
line. This breaks the expected visual structure of the code making it
much harder to read and understand:
if (condition) foo (); else bar (); /* Yuck! */
Do eliminate trailing whitespace (space or tab characters) on any
line. Also, avoid putting initial or final blank lines into any file,
and never use multiple blank lines instead of a single blank line.
Do enable the default git pre-commit hook that detect trailing
whitespace for you and help you to avoid corrupting cairo's tree with
it. Do that as follows:
chmod a+x .git/hooks/pre-commit
You might also find the git-stripspace utility helpful which acts as a
filter to remove trailing whitespace as well as initial, final, and
duplicate blank lines.
As a special case of the bracing and whitespace guidelines, function
definitions should always take the following form:
void
my_function (argument)
{
do_my_things ();
}
And function prototypes should similarly have the return type (and
associated specifiers and qualifiers) on a line above the function, so
that the function name is flush left.
Break up long lines (> ~80 characters) and use whitespace to align
things nicely. For example the arguments in a long list to a function
call should all be aligned with each other:
align_function_arguments (argument_the_first,
argument_the_second,
argument_the_third);
And as a special rule, in a function prototype, (as well as in the
definition), whitespace should be inserted between the parameter types
and names so that the names are aligned:
void
align_parameter_names_in_prototypes (const char *char_star_arg,
int int_arg,
double *double_star_arg,
double double_arg);
Note that parameters with a * prefix are aligned one character to the
left so that the actual names are aligned.
Managing nested blocks
----------------------
Long blocks that are deeply nested make the code very hard to
read. Fortunately such blocks often indicate logically distinct chunks
of functionality that are begging to be split into their own
functions. Please listen to the blocks when they beg.
In other cases, gratuitous nesting comes about because the primary
functionality gets buried in a nested block rather than living at the
primary level where it belongs. Consider the following:
foo = malloc (sizeof (foo_t));
if (foo) { /* Yuck! */
...
/* lots of code to initialize foo */
...
return SUCCESS;
}
return FAILURE;
This kind of gratuitous nesting can be avoided by following a pattern
of handling exceptional cases early and returning:
foo = malloc (sizeof (foo_t));
if (foo == NULL)
return FAILURE;
...
/* lots of code to initialize foo */
...
return SUCCESS;
The return statement is often the best thing to use in a pattern like
this. If it's not available due to additional nesting above which
require some cleanup after the current block, then consider splitting
the current block into a new function before using goto.
Memory allocation
-----------------
Because much of cairo's data consists of dynamically allocated arrays,
it's very easy to introduce integer overflow issues whenever malloc()
is called. Use the _cairo_malloc2(), _cairo_malloc3(), and
_cairo_malloc2_add1 macros to avoid these cases; these macros check
for overflow and will return NULL in that case.
malloc (n * size) => _cairo_malloc_ab (n, size)
e.g. malloc (num_elts * sizeof(some_type)) =>
_cairo_malloc2 (num_elts, sizeof(some_type))
malloc (a * b * size) => _cairo_malloc_abc (a, b, size)
e.g. malloc (width * height * 4) =>
_cairo_malloc3 (width, height, 4)
malloc (n * size + k) => _cairo_malloc_ab_plus_c (n, size, k)
e.g. malloc (num * sizeof(entry) + sizeof(header)) =>
_cairo_malloc2k (num, sizeof(entry), sizeof(header))
In general, be wary of performing any arithmetic operations in an
argument to malloc. You should explicitly check for integer overflow
yourself in any more complex situations.
Mode lines
----------
So given the rules above, what is the best way to simplify one's life as
a code monkey? Get your editor to do most of the tedious work of
beautifying your code!
As a reward for reading this far, here are some mode lines for the more
popular editors:
/*
* vim:sw=4:sts=4:ts=8:tw=78:fo=tcroq:cindent:cino=\:0,(0
* vim:isk=a-z,A-Z,48-57,_,.,-,>
*/
TODO
----
Write rules for common editors to use this style. Also cleanup/unify
the modelines in the source files.
|