3 perlstyle - Perl style guide
7 Each programmer will, of course, have his or her own preferences in
8 regards to formatting, but there are some general guidelines that will
9 make your programs easier to read, understand, and maintain.
11 The most important thing is to use L<strict> and L<warnings> in all your
12 code or know the reason why not to. You may turn them off explicitly for
13 particular portions of code via C<no warnings> or C<no strict>, and this
14 can be limited to the specific warnings or strict features you wish to
15 disable. The B<-w> flag and C<$^W> variable should not be used for this
16 purpose since they can affect code you use but did not write, such as
17 modules from core or CPAN.
19 Regarding aesthetics of code lay out, about the only thing Larry
20 cares strongly about is that the closing curly bracket of
21 a multi-line BLOCK should line up with the keyword that started the construct.
22 Beyond that, he has other preferences that aren't so strong:
32 Opening curly on same line as keyword, if possible, otherwise line up.
36 Space before the opening curly of a multi-line BLOCK.
40 One-line BLOCK may be put on one line, including curlies.
44 No space before the semicolon.
48 Semicolon omitted in "short" one-line BLOCK.
52 Space around most operators.
56 Space around a "complex" subscript (inside brackets).
60 Blank lines between chunks that do different things.
68 No space between function name and its opening parenthesis.
72 Space after each comma.
76 Long lines broken after an operator (except C<and> and C<or>).
80 Space after last parenthesis matching on current line.
84 Line up corresponding items vertically.
88 Omit redundant punctuation as long as clarity doesn't suffer.
92 Larry has his reasons for each of these things, but he doesn't claim that
93 everyone else's mind works the same as his does.
95 Here are some other more substantive style issues to think about:
101 Just because you I<CAN> do something a particular way doesn't mean that
102 you I<SHOULD> do it that way. Perl is designed to give you several
103 ways to do anything, so consider picking the most readable one. For
106 open(my $fh, '<', $foo) || die "Can't open $foo: $!";
110 die "Can't open $foo: $!" unless open(my $fh, '<', $foo);
112 because the second way hides the main point of the statement in a
113 modifier. On the other hand
115 print "Starting analysis\n" if $verbose;
119 $verbose && print "Starting analysis\n";
121 because the main point isn't whether the user typed B<-v> or not.
123 Similarly, just because an operator lets you assume default arguments
124 doesn't mean that you have to make use of the defaults. The defaults
125 are there for lazy systems programmers writing one-shot programs. If
126 you want your program to be readable, consider supplying the argument.
128 Along the same lines, just because you I<CAN> omit parentheses in many
129 places doesn't mean that you ought to:
131 return print reverse sort num values %array;
132 return print(reverse(sort num (values(%array))));
134 When in doubt, parenthesize. At the very least it will let some poor
135 schmuck bounce on the % key in B<vi>.
137 Even if you aren't in doubt, consider the mental welfare of the person
138 who has to maintain the code after you, and who will probably put
139 parentheses in the wrong place.
143 Don't go through silly contortions to exit a loop at the top or the
144 bottom, when Perl provides the C<last> operator so you can exit in
145 the middle. Just "outdent" it a little to make it more visible:
157 Don't be afraid to use loop labels--they're there to enhance
158 readability as well as to allow multilevel loop breaks. See the
163 Avoid using C<grep()> (or C<map()>) or `backticks` in a void context, that is,
164 when you just throw away their return values. Those functions all
165 have return values, so use them. Otherwise use a C<foreach()> loop or
166 the C<system()> function instead.
170 For portability, when using features that may not be implemented on
171 every machine, test the construct in an eval to see if it fails. If
172 you know what version or patchlevel a particular feature was
173 implemented, you can test C<$]> (C<$PERL_VERSION> in C<English>) to see if it
174 will be there. The C<Config> module will also let you interrogate values
175 determined by the B<Configure> program when Perl was installed.
179 Choose mnemonic identifiers. If you can't remember what mnemonic means,
180 you've got a problem.
184 While short identifiers like C<$gotit> are probably ok, use underscores to
185 separate words in longer identifiers. It is generally easier to read
186 C<$var_names_like_this> than C<$VarNamesLikeThis>, especially for
187 non-native speakers of English. It's also a simple rule that works
188 consistently with C<VAR_NAMES_LIKE_THIS>.
190 Package names are sometimes an exception to this rule. Perl informally
191 reserves lowercase module names for "pragma" modules like C<integer> and
192 C<strict>. Other modules should begin with a capital letter and use mixed
193 case, but probably without underscores due to limitations in primitive
194 file systems' representations of module names as files that must fit into a
199 You may find it helpful to use letter case to indicate the scope
200 or nature of a variable. For example:
202 $ALL_CAPS_HERE constants only (beware clashes with perl vars!)
203 $Some_Caps_Here package-wide global/static
204 $no_caps_here function scope my() or local() variables
206 Function and method names seem to work best as all lowercase.
207 E.g., C<$obj-E<gt>as_string()>.
209 You can use a leading underscore to indicate that a variable or
210 function should not be used outside the package that defined it.
214 If you have a really hairy regular expression, use the C</x> or C</xx>
215 modifiers and put in some whitespace to make it look a little less like
217 Don't use slash as a delimiter when your regexp has slashes or backslashes.
221 Use the new C<and> and C<or> operators to avoid having to parenthesize
222 list operators so much, and to reduce the incidence of punctuation
223 operators like C<&&> and C<||>. Call your subroutines as if they were
224 functions or list operators to avoid excessive ampersands and parentheses.
228 Use here documents instead of repeated C<print()> statements.
232 Line up corresponding things vertically, especially if it'd be too long
233 to fit on one line anyway.
236 $IDX = $ST_ATIME if $opt_u;
237 $IDX = $ST_CTIME if $opt_c;
238 $IDX = $ST_SIZE if $opt_s;
240 mkdir $tmpdir, 0700 or die "can't mkdir $tmpdir: $!";
241 chdir($tmpdir) or die "can't chdir $tmpdir: $!";
242 mkdir 'tmp', 0777 or die "can't mkdir $tmpdir/tmp: $!";
246 Always check the return codes of system calls. Good error messages should
247 go to C<STDERR>, include which program caused the problem, what the failed
248 system call and arguments were, and (VERY IMPORTANT) should contain the
249 standard system error message for what went wrong. Here's a simple but
252 opendir(my $dh, $dir) or die "can't opendir $dir: $!";
256 Line up your transliterations when it makes sense:
263 Think about reusability. Why waste brainpower on a one-shot when you
264 might want to do something like it again? Consider generalizing your
265 code. Consider writing a module or object class. Consider making your
266 code run cleanly with C<use strict> and C<use warnings> in
267 effect. Consider giving away your code. Consider changing your whole
268 world view. Consider... oh, never mind.
272 Try to document your code and use Pod formatting in a consistent way. Here
273 are commonly expected conventions:
279 use C<CE<lt>E<gt>> for function, variable and module names (and more
280 generally anything that can be considered part of code, like filehandles
281 or specific values). Note that function names are considered more readable
282 with parentheses after their name, that is C<function()>.
286 use C<BE<lt>E<gt>> for commands names like B<cat> or B<grep>.
290 use C<FE<lt>E<gt>> or C<CE<lt>E<gt>> for file names. C<FE<lt>E<gt>> should
291 be the only Pod code for file names, but as most Pod formatters render it
292 as italic, Unix and Windows paths with their slashes and backslashes may
293 be less readable, and better rendered with C<CE<lt>E<gt>>.