3 perlmodstyle - Perl module style guide
7 This document attempts to describe the Perl Community's "best practice"
8 for writing Perl modules. It extends the recommendations found in
9 L<perlstyle> , which should be considered required reading
10 before reading this document.
12 While this document is intended to be useful to all module authors, it is
13 particularly aimed at authors who wish to publish their modules on CPAN.
15 The focus is on elements of style which are visible to the users of a
16 module, rather than those parts which are only seen by the module's
17 developers. However, many of the guidelines presented in this document
18 can be extrapolated and applied successfully to a module's internals.
20 This document differs from L<perlnewmod> in that it is a style guide
21 rather than a tutorial on creating CPAN modules. It provides a
22 checklist against which modules can be compared to determine whether
23 they conform to best practice, without necessarily describing in detail
26 All the advice contained in this document has been gleaned from
27 extensive conversations with experienced CPAN authors and users. Every
28 piece of advice given here is the result of previous mistakes. This
29 information is here to help you avoid the same mistakes and the extra
30 work that would inevitably be required to fix them.
32 The first section of this document provides an itemized checklist;
33 subsequent sections provide a more detailed discussion of the items on
34 the list. The final section, "Common Pitfalls", describes some of the
35 most popular mistakes made by CPAN authors.
37 =head1 QUICK CHECKLIST
39 For more detail on each item in this checklist, see below.
41 =head2 Before you start
47 Don't re-invent the wheel
51 Patch, extend or subclass an existing module where possible
55 Do one thing and do it well
59 Choose an appropriate name
63 Get feedback before publishing
73 API should be understandable by the average programmer
77 Simple methods for simple tasks
81 Separate functionality from output
85 Consistent naming of subroutines or methods
89 Use named parameters (a hash or hashref) when there are more than two
100 Ensure your module works under C<use strict> and C<-w>
104 Stable modules should maintain backwards compatibility
114 Write documentation in POD
118 Document purpose, scope and target applications
122 Document each publically accessible method or subroutine, including params and return values
126 Give examples of use in your documentation
130 Provide a README file and perhaps also release notes, changelog, etc
134 Provide links to further information (URL, email)
138 =head2 Release considerations
144 Specify pre-requisites in Makefile.PL or Build.PL
148 Specify Perl version requirements with C<use>
152 Include tests with your module
156 Choose a sensible and consistent version numbering scheme (X.YY is the common Perl module numbering scheme)
160 Increment the version number for every change, no matter how small
164 Package the module using "make dist"
168 Choose an appropriate license (GPL/Artistic is a good default)
172 =head1 BEFORE YOU START WRITING A MODULE
174 Try not to launch headlong into developing your module without spending
175 some time thinking first. A little forethought may save you a vast
176 amount of effort later on.
178 =head2 Has it been done before?
180 You may not even need to write the module. Check whether it's already
181 been done in Perl, and avoid re-inventing the wheel unless you have a
184 Good places to look for pre-existing modules include
185 L<http://search.cpan.org/> and L<https://metacpan.org>
186 and asking on C<module-authors@perl.org>
187 (L<http://lists.perl.org/list/module-authors.html>).
189 If an existing module B<almost> does what you want, consider writing a
190 patch, writing a subclass, or otherwise extending the existing module
191 rather than rewriting it.
193 =head2 Do one thing and do it well
195 At the risk of stating the obvious, modules are intended to be modular.
196 A Perl developer should be able to use modules to put together the
197 building blocks of their application. However, it's important that the
198 blocks are the right shape, and that the developer shouldn't have to use
199 a big block when all they need is a small one.
201 Your module should have a clearly defined scope which is no longer than
202 a single sentence. Can your module be broken down into a family of
207 "FooBar.pm provides an implementation of the FOO protocol and the
208 related BAR standard."
212 "Foo.pm provides an implementation of the FOO protocol. Bar.pm
213 implements the related BAR protocol."
215 This means that if a developer only needs a module for the BAR standard,
216 they should not be forced to install libraries for FOO as well.
218 =head2 What's in a name?
220 Make sure you choose an appropriate name for your module early on. This
221 will help people find and remember your module, and make programming
222 with your module more intuitive.
224 When naming your module, consider the following:
230 Be descriptive (i.e. accurately describes the purpose of the module).
234 Be consistent with existing modules.
238 Reflect the functionality of the module, not the implementation.
242 Avoid starting a new top-level hierarchy, especially if a suitable
243 hierarchy already exists under which you could place your module.
247 =head2 Get feedback before publishing
249 If you have never uploaded a module to CPAN before (and even if you have),
250 you are strongly encouraged to get feedback on L<PrePAN|http://prepan.org>.
251 PrePAN is a site dedicated to discussing ideas for CPAN modules with other
252 Perl developers and is a great resource for new (and experienced) Perl
255 You should also try to get feedback from people who are already familiar
256 with the module's application domain and the CPAN naming system. Authors
257 of similar modules, or modules with similar names, may be a good place to
258 start, as are community sites like L<Perl Monks|http://www.perlmonks.org>.
260 =head1 DESIGNING AND WRITING YOUR MODULE
262 Considerations for module design and coding:
264 =head2 To OO or not to OO?
266 Your module may be object oriented (OO) or not, or it may have both kinds
267 of interfaces available. There are pros and cons of each technique, which
268 should be considered when you design your API.
270 In I<Perl Best Practices> (copyright 2004, Published by O'Reilly Media, Inc.),
271 Damian Conway provides a list of criteria to use when deciding if OO is the
272 right fit for your problem:
278 The system being designed is large, or is likely to become large.
282 The data can be aggregated into obvious structures, especially if
283 there's a large amount of data in each aggregate.
287 The various types of data aggregate form a natural hierarchy that
288 facilitates the use of inheritance and polymorphism.
292 You have a piece of data on which many different operations are
297 You need to perform the same general operations on related types of
298 data, but with slight variations depending on the specific type of data
299 the operations are applied to.
303 It's likely you'll have to add new data types later.
307 The typical interactions between pieces of data are best represented by
312 The implementation of individual components of the system is likely to
317 The system design is already object-oriented.
321 Large numbers of other programmers will be using your code modules.
325 Think carefully about whether OO is appropriate for your module.
326 Gratuitous object orientation results in complex APIs which are
327 difficult for the average module user to understand or use.
329 =head2 Designing your API
331 Your interfaces should be understandable by an average Perl programmer.
332 The following guidelines may help you judge whether your API is
333 sufficiently straightforward:
337 =item Write simple routines to do simple things.
339 It's better to have numerous simple routines than a few monolithic ones.
340 If your routine changes its behaviour significantly based on its
341 arguments, it's a sign that you should have two (or more) separate
344 =item Separate functionality from output.
346 Return your results in the most generic form possible and allow the user
347 to choose how to use them. The most generic form possible is usually a
348 Perl data structure which can then be used to generate a text report,
349 HTML, XML, a database query, or whatever else your users require.
351 If your routine iterates through some kind of list (such as a list of
352 files, or records in a database) you may consider providing a callback
353 so that users can manipulate each element of the list in turn.
354 File::Find provides an example of this with its
355 C<find(\&wanted, $dir)> syntax.
357 =item Provide sensible shortcuts and defaults.
359 Don't require every module user to jump through the same hoops to achieve a
360 simple result. You can always include optional parameters or routines for
361 more complex or non-standard behaviour. If most of your users have to
362 type a few almost identical lines of code when they start using your
363 module, it's a sign that you should have made that behaviour a default.
364 Another good indicator that you should use defaults is if most of your
365 users call your routines with the same arguments.
367 =item Naming conventions
369 Your naming should be consistent. For instance, it's better to have:
381 This applies equally to method names, parameter names, and anything else
382 which is visible to the user (and most things that aren't!)
384 =item Parameter passing
386 Use named parameters. It's easier to use a hash like this:
394 ... than to have a long list of unnamed parameters like this:
396 $obj->do_something("wibble", "text", 1024);
398 While the list of arguments might work fine for one, two or even three
399 arguments, any more arguments become hard for the module user to
400 remember, and hard for the module author to manage. If you want to add
401 a new parameter you will have to add it to the end of the list for
402 backward compatibility, and this will probably make your list order
403 unintuitive. Also, if many elements may be undefined you may see the
404 following unattractive method calls:
406 $obj->do_something(undef, undef, undef, undef, undef, 1024);
408 Provide sensible defaults for parameters which have them. Don't make
409 your users specify parameters which will almost always be the same.
411 The issue of whether to pass the arguments in a hash or a hashref is
412 largely a matter of personal style.
414 The use of hash keys starting with a hyphen (C<-name>) or entirely in
415 upper case (C<NAME>) is a relic of older versions of Perl in which
416 ordinary lower case strings were not handled correctly by the C<=E<gt>>
417 operator. While some modules retain uppercase or hyphenated argument
418 keys for historical reasons or as a matter of personal style, most new
419 modules should use simple lower case keys. Whatever you choose, be
424 =head2 Strictness and warnings
426 Your module should run successfully under the strict pragma and should
427 run without generating any warnings. Your module should also handle
428 taint-checking where appropriate, though this can cause difficulties in
431 =head2 Backwards compatibility
433 Modules which are "stable" should not break backwards compatibility
434 without at least a long transition phase and a major change in version
437 =head2 Error handling and messages
439 When your module encounters an error it should do one or more of:
445 Return an undefined value.
449 set C<$Module::errstr> or similar (C<errstr> is a common name used by
450 DBI and other popular modules; if you choose something else, be sure to
451 document it clearly).
455 C<warn()> or C<carp()> a message to STDERR.
459 C<croak()> only when your module absolutely cannot figure out what to
460 do. (C<croak()> is a better version of C<die()> for use within
461 modules, which reports its errors from the perspective of the caller.
462 See L<Carp> for details of C<croak()>, C<carp()> and other useful
467 As an alternative to the above, you may prefer to throw exceptions using
472 Configurable error handling can be very useful to your users. Consider
473 offering a choice of levels for warning and debug messages, an option to
474 send messages to a separate file, a way to specify an error-handling
475 routine, or other such features. Be sure to default all these options
476 to the commonest use.
478 =head1 DOCUMENTING YOUR MODULE
482 Your module should include documentation aimed at Perl developers.
483 You should use Perl's "plain old documentation" (POD) for your general
484 technical documentation, though you may wish to write additional
485 documentation (white papers, tutorials, etc) in some other format.
486 You need to cover the following subjects:
492 A synopsis of the common uses of the module
496 The purpose, scope and target applications of your module
500 Use of each publically accessible method or subroutine, including
501 parameters and return values
509 Sources of further information
513 A contact email address for the author/maintainer
517 The level of detail in Perl module documentation generally goes from
518 less detailed to more detailed. Your SYNOPSIS section should contain a
519 minimal example of use (perhaps as little as one line of code; skip the
520 unusual use cases or anything not needed by most users); the
521 DESCRIPTION should describe your module in broad terms, generally in
522 just a few paragraphs; more detail of the module's routines or methods,
523 lengthy code examples, or other in-depth material should be given in
526 Ideally, someone who's slightly familiar with your module should be able
527 to refresh their memory without hitting "page down". As your reader
528 continues through the document, they should receive a progressively
529 greater amount of knowledge.
531 The recommended order of sections in Perl module documentation is:
549 One or more sections or subsections giving greater detail of available
550 methods and routines and any other relevant information.
566 COPYRIGHT and LICENSE
570 Keep your documentation near the code it documents ("inline"
571 documentation). Include POD for a given method right above that
572 method's subroutine. This makes it easier to keep the documentation up
573 to date, and avoids having to document each piece of code twice (once in
574 POD and once in comments).
576 =head2 README, INSTALL, release notes, changelogs
578 Your module should also include a README file describing the module and
579 giving pointers to further information (website, author email).
581 An INSTALL file should be included, and should contain simple installation
582 instructions. When using ExtUtils::MakeMaker this will usually be:
586 =item perl Makefile.PL
596 When using Module::Build, this will usually be:
604 =item perl Build test
606 =item perl Build install
610 Release notes or changelogs should be produced for each release of your
611 software describing user-visible changes to your module, in terms
612 relevant to the user.
614 Unless you have good reasons for using some other format
615 (for example, a format used within your company),
616 the convention is to name your changelog file C<Changes>,
617 and to follow the simple format described in L<CPAN::Changes::Spec>.
619 =head1 RELEASE CONSIDERATIONS
621 =head2 Version numbering
623 Version numbers should indicate at least major and minor releases, and
624 possibly sub-minor releases. A major release is one in which most of
625 the functionality has changed, or in which major new functionality is
626 added. A minor release is one in which a small amount of functionality
627 has been added or changed. Sub-minor version numbers are usually used
628 for changes which do not affect functionality, such as documentation
631 The most common CPAN version numbering scheme looks like this:
633 1.00, 1.10, 1.11, 1.20, 1.30, 1.31, 1.32
635 A correct CPAN version number is a floating point number with at least
636 2 digits after the decimal. You can test whether it conforms to CPAN by
639 perl -MExtUtils::MakeMaker -le 'print MM->parse_version(shift)' \
642 If you want to release a 'beta' or 'alpha' version of a module but
643 don't want CPAN.pm to list it as most recent use an '_' after the
644 regular version number followed by at least 2 digits, eg. 1.20_01. If
645 you do this, the following idiom is recommended:
647 our $VERSION = "1.12_01"; # so CPAN distribution will have
649 our $XS_VERSION = $VERSION; # only needed if you have XS code
650 $VERSION = eval $VERSION; # so "use Module 0.002" won't warn on
653 With that trick MakeMaker will only read the first line and thus read
654 the underscore, while the perl interpreter will evaluate the $VERSION
655 and convert the string into a number. Later operations that treat
656 $VERSION as a number will then be able to do so without provoking a
657 warning about $VERSION not being a number.
659 Never release anything (even a one-word documentation patch) without
660 incrementing the number. Even a one-word documentation patch should
661 result in a change in version at the sub-minor level.
663 Once picked, it is important to stick to your version scheme, without
664 reducing the number of digits. This is because "downstream" packagers,
665 such as the FreeBSD ports system, interpret the version numbers in
666 various ways. If you change the number of digits in your version scheme,
667 you can confuse these systems so they get the versions of your module
668 out of order, which is obviously bad.
670 =head2 Pre-requisites
672 Module authors should carefully consider whether to rely on other
673 modules, and which modules to rely on.
675 Most importantly, choose modules which are as stable as possible. In
690 Unstable CPAN modules
694 Modules not available from CPAN
698 Specify version requirements for other Perl modules in the
699 pre-requisites in your Makefile.PL or Build.PL.
701 Be sure to specify Perl version requirements both in Makefile.PL or
702 Build.PL and with C<require 5.6.1> or similar. See the section on
703 C<use VERSION> of L<perlfunc/require> for details.
707 All modules should be tested before distribution (using "make disttest"),
708 and the tests should also be available to people installing the modules
710 For Module::Build you would use the C<make test> equivalent C<perl Build test>.
712 The importance of these tests is proportional to the alleged stability of a
713 module. A module which purports to be
714 stable or which hopes to achieve wide
715 use should adhere to as strict a testing regime as possible.
717 Useful modules to help you write tests (with minimum impact on your
718 development process or your time) include Test::Simple, Carp::Assert
720 For more sophisticated test suites there are Test::More and Test::MockObject.
724 Modules should be packaged using one of the standard packaging tools.
725 Currently you have the choice between ExtUtils::MakeMaker and the
726 more platform independent Module::Build, allowing modules to be installed in a
728 When using ExtUtils::MakeMaker, you can use "make dist" to create your
729 package. Tools exist to help you to build your module in a
730 MakeMaker-friendly style. These include ExtUtils::ModuleMaker and h2xs.
731 See also L<perlnewmod>.
735 Make sure that your module has a license, and that the full text of it
736 is included in the distribution (unless it's a common one and the terms
737 of the license don't require you to include it).
739 If you don't know what license to use, dual licensing under the GPL
740 and Artistic licenses (the same as Perl itself) is a good idea.
741 See L<perlgpl> and L<perlartistic>.
743 =head1 COMMON PITFALLS
745 =head2 Reinventing the wheel
747 There are certain application spaces which are already very, very well
748 served by CPAN. One example is templating systems, another is date and
749 time modules, and there are many more. While it is a rite of passage to
750 write your own version of these things, please consider carefully
751 whether the Perl world really needs you to publish it.
753 =head2 Trying to do too much
755 Your module will be part of a developer's toolkit. It will not, in
756 itself, form the B<entire> toolkit. It's tempting to add extra features
757 until your code is a monolithic system rather than a set of modular
760 =head2 Inappropriate documentation
762 Don't fall into the trap of writing for the wrong audience. Your
763 primary audience is a reasonably experienced developer with at least
764 a moderate understanding of your module's application domain, who's just
765 downloaded your module and wants to start using it as quickly as possible.
767 Tutorials, end-user documentation, research papers, FAQs etc are not
768 appropriate in a module's main documentation. If you really want to
769 write these, include them as sub-documents such as C<My::Module::Tutorial> or
770 C<My::Module::FAQ> and provide a link in the SEE ALSO section of the
779 General Perl style guide
783 How to create a new module
791 Verifies your POD's correctness
793 =item Packaging Tools
795 L<ExtUtils::MakeMaker>, L<Module::Build>
799 L<Test::Simple>, L<Test::Inline>, L<Carp::Assert>, L<Test::More>, L<Test::MockObject>
801 =item L<http://pause.perl.org/>
803 Perl Authors Upload Server. Contains links to information for module
806 =item Any good book on software engineering
812 Kirrily "Skud" Robert <skud@cpan.org>