1PERLHACK(1) Perl Programmers Reference Guide PERLHACK(1)
2
6 perlhack - How to hack on Perl
7
9 This document explains how Perl development works. It includes details
10 about the Perl 5 Porters email list, the Perl repository, the Perl bug
11 tracker, patch guidelines, and commentary on Perl development
12 philosophy.
13
15 If you just want to submit a single small patch like a pod fix, a test
16 for a bug, comment fixes, etc., it's easy! Here's how:
17 • Check out the source repository
19 The perl source is in a git repository. You can clone the
21 repository with the following command:
22 % git clone https://github.com/Perl/perl5.git perl
24 • Ensure you're following the latest advice
26 In case the advice in this guide has been updated recently, read
28 the latest version directly from the perl source:
29 % perldoc pod/perlhack.pod
31 • Create a branch for your change
33 Create a branch based on blead to commit your change to, which will
35 later be used to send it to the Perl issue tracker.
36 % git checkout -b mychange
38 • Make your change
40 Hack, hack, hack. Keep in mind that Perl runs on many different
42 platforms, with different operating systems that have different
43 capabilities, different filesystem organizations, and even
44 different character sets. perlhacktips gives advice on this.
45 • Test your change
47 You can run all the tests with the following commands:
49 % ./Configure -des -Dusedevel
51 % make test
52 Keep hacking until the tests pass.
54 • Commit your change
56 Committing your work will save the change on your local system:
58 % git commit -a -m 'Commit message goes here'
60 Make sure the commit message describes your change in a single
62 sentence. For example, "Fixed spelling errors in perlhack.pod".
63 • Send your change to the Perl issue tracker
65 The next step is to submit your patch to the Perl core ticket
67 system.
68 Create a GitHub fork of the perl5 repository and add it as a
70 remote, if you haven't already, as described in the GitHub
71 documentation at
72 <https://help.github.com/en/articles/working-with-forks>.
73 % git remote add fork git@github.com:MyUser/perl5.git
75 For more information, see "Connecting to GitHub with SSH"
77 <https://docs.github.com/en/free-pro-
78 team@latest/github/authenticating-to-github/connecting-to-github-
79 with-ssh>.
80 If you'd rather use an HTTPS URL for your "git push" see "Cloning
82 with HTTPS URLs" <https://docs.github.com/en/free-pro-
83 team@latest/github/using-git/which-remote-url-should-i-use#cloning-
84 with-https-urls>.
85 % git remote add fork https://github.com/MyUser/perl5.git
87 Then, push your new branch to your fork.
89 % git push -u fork mychange
91 Finally, create a Pull Request on GitHub from your branch to blead
93 as described in the GitHub documentation at
94 <https://help.github.com/en/articles/creating-a-pull-request-from-a-fork>.
95 • Thank you
97 The porters appreciate the time you spent helping to make Perl
99 better. Thank you!
100 • Acknowledgement
102 All contributors are credited (by name and email address) in the
104 AUTHORS file, which is part of the perl distribution, as well as
105 the Git commit history.
106 If you donXt want to be included in the AUTHORS file, just let us
108 know. Otherwise we will take your submission of a patch as
109 permission to credit you in the AUTHORS file.
110 • Next time
112 The next time you wish to make a patch, you need to start from the
114 latest perl in a pristine state. Check you don't have any local
115 changes or added files in your perl check-out which you wish to
116 keep, then run these commands:
117 % git checkout blead
119 % git pull
120 % git reset --hard origin/blead
121 % git clean -dxf
122
124 If you want to report a bug in Perl, or browse existing Perl bugs and
125 patches, use the GitHub issue tracker at
126 <https://github.com/perl/perl5/issues>.
127 Please check the archive of the perl5-porters list (see below) and/or
129 the bug tracking system before submitting a bug report. Often, you'll
130 find that the bug has been reported already.
131 You can log in to the bug tracking system and comment on existing bug
133 reports. If you have additional information regarding an existing bug,
134 please add it. This will help the porters fix the bug.
135
137 The perl5-porters (p5p) mailing list is where the Perl standard
138 distribution is maintained and developed. The people who maintain Perl
139 are also referred to as the "Perl 5 Porters", "p5p" or just the
140 "porters".
141 A searchable archive of the list is available at
143 <https://markmail.org/search/?q=perl5-porters>. There is also an
144 archive at <https://archive.develooper.com/perl5-porters@perl.org/>.
145 perl-changes mailing list
147 The perl5-changes mailing list receives a copy of each patch that gets
148 submitted to the maintenance and development branches of the perl
149 repository. See <https://lists.perl.org/list/perl5-changes.html> for
150 subscription and archive information.
151 #p5p on IRC
153 Many porters are also active on the <irc://irc.perl.org/#p5p> channel.
154 Feel free to join the channel and ask questions about hacking on the
155 Perl core.
156
158 All of Perl's source code is kept centrally in a Git repository at
159 github.com. The repository contains many Perl revisions from Perl 1
160 onwards and all the revisions from Perforce, the previous version
161 control system.
162 For much more detail on using git with the Perl repository, please see
164 perlgit.
165 Read access via Git
167 You will need a copy of Git for your computer. You can fetch a copy of
168 the repository using the git protocol:
169 % git clone git@github.com:Perl/perl5.git perl
171 This clones the repository and makes a local copy in the perl
173 directory.
174 If you cannot use the git protocol for firewall reasons, you can also
176 clone via http:
177 % git clone https://github.com/Perl/perl5.git perl
179 Read access via the web
181 You may access the repository over the web. This allows you to browse
182 the tree, see recent commits, subscribe to repository notifications,
183 search for particular commits and more. You may access it at
184 <https://github.com/Perl/perl5>.
185 Write access via git
187 If you have a commit bit, please see perlgit for more details on using
188 git.
189
191 If you're planning to do more extensive work than a single small fix,
192 we encourage you to read the documentation below. This will help you
193 focus your work and make your patches easier to incorporate into the
194 Perl source.
195 Submitting patches
197 If you have a small patch to submit, please submit it via the GitHub
198 Pull Request workflow. You may also send patches to the p5p list.
199 Patches are reviewed and discussed on GitHub or the p5p list. Simple,
201 uncontroversial patches will usually be applied without any discussion.
202 When the patch is applied, the ticket will be updated and you will
203 receive email.
204 In other cases, the patch will need more work or discussion. You are
206 encouraged to participate in the discussion and advocate for your
207 patch. Sometimes your patch may get lost in the shuffle. It's
208 appropriate to send a reminder email to p5p if no action has been taken
209 in a month. Please remember that the Perl 5 developers are all
210 volunteers, and be polite.
211 Changes are always applied directly to the main development branch,
213 called "blead". Some patches may be backported to a maintenance
214 branch. If you think your patch is appropriate for the maintenance
215 branch (see "MAINTENANCE BRANCHES" in perlpolicy), please explain why
216 when you submit it.
217 Getting your patch accepted
219 If you are submitting a code patch there are several things that you
220 can do to help the Perl 5 Porters accept your patch.
221 Patch style
223 Using the GitHub Pull Request workflow, your patch will automatically
225 be available in a suitable format. If you wish to submit a patch to
226 the p5p list for review, make sure to create it appropriately.
227 If you used git to check out the Perl source, then using "git
229 format-patch" will produce a patch in a style suitable for Perl. The
230 "format-patch" command produces one patch file for each commit you
231 made. If you prefer to send a single patch for all commits, you can
232 use "git diff".
233 % git checkout blead
235 % git pull
236 % git diff blead my-branch-name
237 This produces a patch based on the difference between blead and your
239 current branch. It's important to make sure that blead is up to date
240 before producing the diff, that's why we call "git pull" first.
241 We strongly recommend that you use git if possible. It will make your
243 life easier, and ours as well.
244 However, if you're not using git, you can still produce a suitable
246 patch. You'll need a pristine copy of the Perl source to diff against.
247 The porters prefer unified diffs. Using GNU "diff", you can produce a
248 diff like this:
249 % diff -Npurd perl.pristine perl.mine
251 Make sure that you "make realclean" in your copy of Perl to remove any
253 build artifacts, or you may get a confusing result.
254 Commit message
256 As you craft each patch you intend to submit to the Perl core, it's
258 important to write a good commit message. This is especially important
259 if your submission will consist of a series of commits.
260 The first line of the commit message should be a short description
262 without a period. It should be no longer than the subject line of an
263 email, 50 characters being a good rule of thumb.
264 A lot of Git tools (Gitweb, GitHub, git log --pretty=oneline, ...) will
266 only display the first line (cut off at 50 characters) when presenting
267 commit summaries.
268 The commit message should include a description of the problem that the
270 patch corrects or new functionality that the patch adds.
271 As a general rule of thumb, your commit message should help a
273 programmer who knows the Perl core quickly understand what you were
274 trying to do, how you were trying to do it, and why the change matters
275 to Perl.
276 • Why
278 Your commit message should describe why the change you are making
280 is important. When someone looks at your change in six months or
281 six years, your intent should be clear.
282 If you're deprecating a feature with the intent of later
284 simplifying another bit of code, say so. If you're fixing a
285 performance problem or adding a new feature to support some other
286 bit of the core, mention that.
287 • What
289 Your commit message should describe what part of the Perl core
291 you're changing and what you expect your patch to do.
292 • How
294 While it's not necessary for documentation changes, new tests or
296 trivial patches, it's often worth explaining how your change works.
297 Even if it's clear to you today, it may not be clear to a porter
298 next month or next year.
299 A commit message isn't intended to take the place of comments in your
301 code. Commit messages should describe the change you made, while code
302 comments should describe the current state of the code.
303 If you've just implemented a new feature, complete with doc, tests and
305 well-commented code, a brief commit message will often suffice. If,
306 however, you've just changed a single character deep in the parser or
307 lexer, you might need to write a small novel to ensure that future
308 readers understand what you did and why you did it.
309 Comments, Comments, Comments
311 Be sure to adequately comment your code. While commenting every line
313 is unnecessary, anything that takes advantage of side effects of
314 operators, that creates changes that will be felt outside of the
315 function being patched, or that others may find confusing should be
316 documented. If you are going to err, it is better to err on the side
317 of adding too many comments than too few.
318 The best comments explain why the code does what it does, not what it
320 does.
321 Style
323 In general, please follow the particular style of the code you are
325 patching.
326 In particular, follow these general guidelines for patching Perl
328 sources:
329 • 4-wide indents for code, 2-wide indents for nested CPP "#define"s,
331 with 8-wide tabstops.
332 • Use spaces for indentation, not tab characters.
334 The codebase is a mixture of tabs and spaces for indentation, and
336 we are moving to spaces only. Converting lines you're patching
337 from 8-wide tabs to spaces will help this migration.
338 • Try not to exceed 79 columns
340 In general, we target 80 column lines. When sticking to 80 columns
342 would lead to torturous code or rework, it's fine to go longer.
343 Try to keep your excess past 80 to a minimum.
344 • ANSI C prototypes
346 • Uncuddled elses and "K&R" style for indenting control constructs
348 • No C++ style (//) comments
350 • Mark places that need to be revisited with XXX (and revisit often!)
352 • Opening brace lines up with "if" when conditional spans multiple
354 lines; should be at end-of-line otherwise
355 • In function definitions, name starts in column 0 (return value-type
357 is on previous line)
358 • Single space after keywords that are followed by parens, no space
360 between function name and following paren
361 • Avoid assignments in conditionals, but if they're unavoidable, use
363 extra paren, e.g. "if (a && (b = c)) ..."
364 • "return foo;" rather than "return(foo);"
366 • "if (!foo) ..." rather than "if (foo == FALSE) ..." etc.
368 • Do not declare variables using "register". It may be
370 counterproductive with modern compilers, and is deprecated in C++,
371 under which the Perl source is regularly compiled.
372 • In-line functions that are in headers that are accessible to XS
374 code need to be able to compile without warnings with commonly used
375 extra compilation flags, such as gcc's "-Wswitch-default" which
376 warns whenever a switch statement does not have a "default" case.
377 The use of these extra flags is to catch potential problems in
378 legal C code, and is often used by Perl aggregators, such as Linux
379 distributors.
380 Test suite
382 If your patch changes code (rather than just changing documentation),
384 you should also include one or more test cases which illustrate the bug
385 you're fixing or validate the new functionality you're adding. In
386 general, you should update an existing test file rather than create a
387 new one.
388 Your test suite additions should generally follow these guidelines
390 (courtesy of Gurusamy Sarathy <gsar@activestate.com>):
391 • Know what you're testing. Read the docs, and the source.
393 • Tend to fail, not succeed.
395 • Interpret results strictly.
397 • Use unrelated features (this will flush out bizarre interactions).
399 • Use non-standard idioms (otherwise you are not testing TIMTOWTDI).
401 • Avoid using hardcoded test numbers whenever possible (the
403 EXPECTED/GOT found in t/op/tie.t is much more maintainable, and
404 gives better failure reports).
405 • Give meaningful error messages when a test fails.
407 • Avoid using qx// and system() unless you are testing for them. If
409 you do use them, make sure that you cover _all_ perl platforms.
410 • Unlink any temporary files you create.
412 • Promote unforeseen warnings to errors with $SIG{__WARN__}.
414 • Be sure to use the libraries and modules shipped with the version
416 being tested, not those that were already installed.
417 • Add comments to the code explaining what you are testing for.
419 • Make updating the '1..42' string unnecessary. Or make sure that
421 you update it.
422 • Test _all_ behaviors of a given operator, library, or function.
424 Test all optional arguments.
426 Test return values in various contexts (boolean, scalar, list,
428 lvalue).
429 Use both global and lexical variables.
431 Don't forget the exceptional, pathological cases.
433 Patching a core module
435 This works just like patching anything else, with one extra
436 consideration.
437 Modules in the cpan/ directory of the source tree are maintained
439 outside of the Perl core. When the author updates the module, the
440 updates are simply copied into the core. See that module's
441 documentation or its listing on <https://metacpan.org/> for more
442 information on reporting bugs and submitting patches.
443 In most cases, patches to modules in cpan/ should be sent upstream and
445 should not be applied to the Perl core individually. If a patch to a
446 file in cpan/ absolutely cannot wait for the fix to be made upstream,
447 released to CPAN and copied to blead, you must add (or update) a
448 "CUSTOMIZED" entry in the Porting/Maintainers.pl file to flag that a
449 local modification has been made. See Porting/Maintainers.pl for more
450 details.
451 In contrast, modules in the dist/ directory are maintained in the core.
453 Updating perldelta
455 For changes significant enough to warrant a pod/perldelta.pod entry,
456 the porters will greatly appreciate it if you submit a delta entry
457 along with your actual change. Significant changes include, but are
458 not limited to:
459 • Adding, deprecating, or removing core features
461 • Adding, deprecating, removing, or upgrading core or dual-life
463 modules
464 • Adding new core tests
466 • Fixing security issues and user-visible bugs in the core
468 • Changes that might break existing code, either on the perl or C
470 level
471 • Significant performance improvements
473 • Adding, removing, or significantly changing documentation in the
475 pod/ directory
476 • Important platform-specific changes
478 Please make sure you add the perldelta entry to the right section
480 within pod/perldelta.pod. More information on how to write good
481 perldelta entries is available in the "Style" section of
482 Porting/how_to_write_a_perldelta.pod.
483 What makes for a good patch?
485 New features and extensions to the language can be contentious. There
486 is no specific set of criteria which determine what features get added,
487 but here are some questions to consider when developing a patch:
488 Does the concept match the general goals of Perl?
490 Our goals include, but are not limited to:
492 1. Keep it fast, simple, and useful.
494 2. Keep features/concepts as orthogonal as possible.
496 3. No arbitrary limits (platforms, data sizes, cultures).
498 4. Keep it open and exciting to use/patch/advocate Perl everywhere.
500 5. Either assimilate new technologies, or build bridges to them.
502 Where is the implementation?
504 All the talk in the world is useless without an implementation. In
506 almost every case, the person or people who argue for a new feature
507 will be expected to be the ones who implement it. Porters capable of
508 coding new features have their own agendas, and are not available to
509 implement your (possibly good) idea.
510 Backwards compatibility
512 It's a cardinal sin to break existing Perl programs. New warnings can
514 be contentious--some say that a program that emits warnings is not
515 broken, while others say it is. Adding keywords has the potential to
516 break programs, changing the meaning of existing token sequences or
517 functions might break programs.
518 The Perl 5 core includes mechanisms to help porters make backwards
520 incompatible changes more compatible such as the feature and deprecate
521 modules. Please use them when appropriate.
522 Could it be a module instead?
524 Perl 5 has extension mechanisms, modules and XS, specifically to avoid
526 the need to keep changing the Perl interpreter. You can write modules
527 that export functions, you can give those functions prototypes so they
528 can be called like built-in functions, you can even write XS code to
529 mess with the runtime data structures of the Perl interpreter if you
530 want to implement really complicated things.
531 Whenever possible, new features should be prototyped in a CPAN module
533 before they will be considered for the core.
534 Is the feature generic enough?
536 Is this something that only the submitter wants added to the language,
538 or is it broadly useful? Sometimes, instead of adding a feature with a
539 tight focus, the porters might decide to wait until someone implements
540 the more generalized feature.
541 Does it potentially introduce new bugs?
543 Radical rewrites of large chunks of the Perl interpreter have the
545 potential to introduce new bugs.
546 How big is it?
548 The smaller and more localized the change, the better. Similarly, a
550 series of small patches is greatly preferred over a single large patch.
551 Does it preclude other desirable features?
553 A patch is likely to be rejected if it closes off future avenues of
555 development. For instance, a patch that placed a true and final
556 interpretation on prototypes is likely to be rejected because there are
557 still options for the future of prototypes that haven't been addressed.
558 Is the implementation robust?
560 Good patches (tight code, complete, correct) stand more chance of going
562 in. Sloppy or incorrect patches might be placed on the back burner
563 until fixes can be made, or they might be discarded altogether without
564 further notice.
565 Is the implementation generic enough to be portable?
567 The worst patches make use of system-specific features. It's highly
569 unlikely that non-portable additions to the Perl language will be
570 accepted.
571 Is the implementation tested?
573 Patches which change behaviour (fixing bugs or introducing new
575 features) must include regression tests to verify that everything works
576 as expected.
577 Without tests provided by the original author, how can anyone else
579 changing perl in the future be sure that they haven't unwittingly
580 broken the behaviour the patch implements? And without tests, how can
581 the patch's author be confident that his/her hard work put into the
582 patch won't be accidentally thrown away by someone in the future?
583 Is there enough documentation?
585 Patches without documentation are probably ill-thought out or
587 incomplete. No features can be added or changed without documentation,
588 so submitting a patch for the appropriate pod docs as well as the
589 source code is important.
590 Is there another way to do it?
592 Larry said "Although the Perl Slogan is There's More Than One Way to Do
594 It, I hesitate to make 10 ways to do something". This is a tricky
595 heuristic to navigate, though--one man's essential addition is another
596 man's pointless cruft.
597 Does it create too much work?
599 Work for the committers, work for Perl programmers, work for module
601 authors, ... Perl is supposed to be easy.
602 Patches speak louder than words
604 Working code is always preferred to pie-in-the-sky ideas. A patch to
606 add a feature stands a much higher chance of making it to the language
607 than does a random feature request, no matter how fervently argued the
608 request might be. This ties into "Will it be useful?", as the fact
609 that someone took the time to make the patch demonstrates a strong
610 desire for the feature.
611
613 The core uses the same testing style as the rest of Perl, a simple
614 "ok/not ok" run through Test::Harness, but there are a few special
615 considerations.
616 There are three ways to write a test in the core: Test::More, t/test.pl
618 and ad hoc "print $test ? "ok 42\n" : "not ok 42\n"". The decision of
619 which to use depends on what part of the test suite you're working on.
620 This is a measure to prevent a high-level failure (such as Config.pm
621 breaking) from causing basic functionality tests to fail.
622 The t/test.pl library provides some of the features of Test::More, but
624 avoids loading most modules and uses as few core features as possible.
625 If you write your own test, use the Test Anything Protocol
627 <https://testanything.org>.
628 • t/base, t/comp and t/opbasic
630 Since we don't know if "require" works, or even subroutines, use ad
632 hoc tests for these three. Step carefully to avoid using the
633 feature being tested. Tests in t/opbasic, for instance, have been
634 placed there rather than in t/op because they test functionality
635 which t/test.pl presumes has already been demonstrated to work.
636 • All other subdirectories of t/
638 Now that basic require() and subroutines are tested, you can use
640 the t/test.pl library.
641 You can also use certain libraries like Config conditionally, but
643 be sure to skip the test gracefully if it's not there.
644 • Test files not found under t/
646 This category includes .t files underneath directories such as
648 dist, ext and lib. Since the core of Perl has now been tested,
649 Test::More can and now should be used. You can also use the full
650 suite of core modules in the tests. (As noted in "Patching a core
651 module" above, changes to .t files found under cpan/ should be
652 submitted to the upstream maintainers of those modules.)
653 When you say "make test", Perl uses the t/TEST program to run the test
655 suite (except under Win32 where it uses t/harness instead). All tests
656 are run from the t/ directory, not the directory which contains the
657 test. This causes some problems with the tests in lib/, so here's some
658 opportunity for some patching.
659 You must be triply conscious of cross-platform concerns. This usually
661 boils down to using File::Spec, avoiding things like "fork()" and
662 "system()" unless absolutely necessary, and not assuming that a given
663 character has a particular ordinal value (code point) or that its UTF-8
664 representation is composed of particular bytes.
665 There are several functions available to specify characters and code
667 points portably in tests. The always-preloaded functions
668 "utf8::unicode_to_native()" and its inverse "utf8::native_to_unicode()"
669 take code points and translate appropriately. The file
670 t/charset_tools.pl has several functions that can be useful. It has
671 versions of the previous two functions that take strings as inputs --
672 not single numeric code points: "uni_to_native()" and
673 "native_to_uni()". If you must look at the individual bytes comprising
674 a UTF-8 encoded string, "byte_utf8a_to_utf8n()" takes as input a string
675 of those bytes encoded for an ASCII platform, and returns the
676 equivalent string in the native platform. For example,
677 "byte_utf8a_to_utf8n("\xC2\xA0")" returns the byte sequence on the
678 current platform that form the UTF-8 for "U+00A0", since "\xC2\xA0" are
679 the UTF-8 bytes on an ASCII platform for that code point. This
680 function returns "\xC2\xA0" on an ASCII platform, and "\x80\x41" on an
681 EBCDIC 1047 one.
682 But easiest is, if the character is specifiable as a literal, like "A"
684 or "%", to use that; if not so specificable, you can use "\N{}" , if
685 the side effects aren't troublesome. Simply specify all your
686 characters in hex, using "\N{U+ZZ}" instead of "\xZZ". "\N{}" is the
687 Unicode name, and so it always gives you the Unicode character.
688 "\N{U+41}" is the character whose Unicode code point is 0x41, hence is
689 'A' on all platforms. The side effects are:
690 • These select Unicode rules. That means that in double-quotish
692 strings, the string is always converted to UTF-8 to force a Unicode
693 interpretation (you can "utf8::downgrade()" afterwards to convert
694 back to non-UTF8, if possible). In regular expression patterns,
695 the conversion isn't done, but if the character set modifier would
696 otherwise be "/d", it is changed to "/u".
697 • If you use the form "\N{character name}", the charnames module gets
699 automatically loaded. This may not be suitable for the test level
700 you are doing.
701 If you are testing locales (see perllocale), there are helper functions
703 in t/loc_tools.pl to enable you to see what locales there are on the
704 current platform.
705 Special "make test" targets
707 There are various special make targets that can be used to test Perl
708 slightly differently than the standard "test" target. Not all them are
709 expected to give a 100% success rate. Many of them have several
710 aliases, and many of them are not available on certain operating
711 systems.
712 • test_porting
714 This runs some basic sanity tests on the source tree and helps
716 catch basic errors before you submit a patch.
717 • minitest
719 Run miniperl on t/base, t/comp, t/cmd, t/run, t/io, t/op, t/uni and
721 t/mro tests.
722 miniperl is a minimalistic perl built to bootstrap building
724 extensions, utilties, documentation etc. It doesn't support
725 dynamic loading and depending on the point in the build process
726 will only have access to a limited set of core modules. miniperl
727 is not intended for day to day use.
728 • test.valgrind check.valgrind
730 (Only in Linux) Run all the tests using the memory leak + naughty
732 memory access tool "valgrind". The log files will be named
733 testname.valgrind.
734 • test_harness
736 Run the test suite with the t/harness controlling program, instead
738 of t/TEST. t/harness is more sophisticated, and uses the
739 Test::Harness module, thus using this test target supposes that
740 perl mostly works. The main advantage for our purposes is that it
741 prints a detailed summary of failed tests at the end. Also, unlike
742 t/TEST, it doesn't redirect stderr to stdout.
743 Note that under Win32 t/harness is always used instead of t/TEST,
745 so there is no special "test_harness" target.
746 Under Win32's "test" target you may use the TEST_SWITCHES and
748 TEST_FILES environment variables to control the behaviour of
749 t/harness. This means you can say
750 nmake test TEST_FILES="op/*.t"
752 nmake test TEST_SWITCHES="-torture" TEST_FILES="op/*.t"
753 • test-notty test_notty
755 Sets PERL_SKIP_TTY_TEST to true before running normal test.
757 Parallel tests
759 The core distribution can now run its regression tests in parallel on
760 Unix-like and Windows platforms. On Unix, instead of running "make
761 test", set "TEST_JOBS" in your environment to the number of tests to
762 run in parallel, and run "make test_harness". On a Bourne-like shell,
763 this can be done as
764 TEST_JOBS=3 make test_harness # Run 3 tests in parallel
766 An environment variable is used, rather than parallel make itself,
768 because TAP::Harness needs to be able to schedule individual non-
769 conflicting test scripts itself, and there is no standard interface to
770 "make" utilities to interact with their job schedulers.
771 Tests are normally run in a logical order, with the sanity tests first,
773 then the main tests of the Perl core functionality, then the tests for
774 the non-core modules. On many-core systems, this may not use the
775 hardware as effectively as possible. By also specifying
776 TEST_JOBS=19 PERL_TEST_HARNESS_ASAP=1 make -j19 test_harness
778 you signal that you want the tests to finish in wall-clock time as
780 short as possible. After the sanity tests are completed, this causes
781 the remaining ones to be packed into the available cores as tightly as
782 we know how. This has its greatest effect on slower, many-core
783 systems. Throughput was sped up by 20% on an outmoded 24-core system;
784 less on more recent faster ones with fewer cores.
785 Note that the command line above added a "-j" parameter to make, so as
787 to cause parallel compilation. This may or may not work on your
788 platform.
789 Running tests by hand
791 You can run part of the test suite by hand by using one of the
792 following commands from the t/ directory:
793 ./perl -I../lib TEST list-of-.t-files
795 or
797 ./perl -I../lib harness list-of-.t-files
799 (If you don't specify test scripts, the whole test suite will be run.)
801 Using t/harness for testing
803 If you use "harness" for testing, you have several command line options
804 available to you. The arguments are as follows, and are in the order
805 that they must appear if used together.
806 harness -v -torture -re=pattern LIST OF FILES TO TEST
808 harness -v -torture -re LIST OF PATTERNS TO MATCH
809 If "LIST OF FILES TO TEST" is omitted, the file list is obtained from
811 the manifest. The file list may include shell wildcards which will be
812 expanded out.
813 • -v
815 Run the tests under verbose mode so you can see what tests were
817 run, and debug output.
818 • -torture
820 Run the torture tests as well as the normal set.
822 • -re=PATTERN
824 Filter the file list so that all the test files run match PATTERN.
826 Note that this form is distinct from the -re LIST OF PATTERNS form
827 below in that it allows the file list to be provided as well.
828 • -re LIST OF PATTERNS
830 Filter the file list so that all the test files run match
832 /(LIST|OF|PATTERNS)/. Note that with this form the patterns are
833 joined by '|' and you cannot supply a list of files, instead the
834 test files are obtained from the MANIFEST.
835 You can run an individual test by a command similar to
837 ./perl -I../lib path/to/foo.t
839 except that the harnesses set up some environment variables that may
841 affect the execution of the test:
842 • PERL_CORE=1
844 indicates that we're running this test as part of the perl core
846 test suite. This is useful for modules that have a dual life on
847 CPAN.
848 • PERL_DESTRUCT_LEVEL=2
850 is set to 2 if it isn't set already (see "PERL_DESTRUCT_LEVEL" in
852 perlhacktips).
853 • PERL
855 (used only by t/TEST) if set, overrides the path to the perl
857 executable that should be used to run the tests (the default being
858 ./perl).
859 • PERL_SKIP_TTY_TEST
861 if set, tells to skip the tests that need a terminal. It's
863 actually set automatically by the Makefile, but can also be forced
864 artificially by running 'make test_notty'.
865 Other environment variables that may influence tests
867 • PERL_TEST_Net_Ping
869 Setting this variable runs all the Net::Ping modules tests,
871 otherwise some tests that interact with the outside world are
872 skipped. See perl58delta.
873 • PERL_TEST_NOVREXX
875 Setting this variable skips the vrexx.t tests for OS2::REXX.
877 • PERL_TEST_NUMCONVERTS
879 This sets a variable in op/numconvert.t.
881 • PERL_TEST_MEMORY
883 Setting this variable includes the tests in t/bigmem/. This should
885 be set to the number of gigabytes of memory available for testing,
886 eg. "PERL_TEST_MEMORY=4" indicates that tests that require 4GiB of
887 available memory can be run safely.
888 See also the documentation for the Test and Test::Harness modules, for
890 more environment variables that affect testing.
891 Performance testing
893 The file t/perf/benchmarks contains snippets of perl code which are
894 intended to be benchmarked across a range of perls by the
895 Porting/bench.pl tool. If you fix or enhance a performance issue, you
896 may want to add a representative code sample to the file, then run
897 bench.pl against the previous and current perls to see what difference
898 it has made, and whether anything else has slowed down as a
899 consequence.
900 The file t/perf/opcount.t is designed to test whether a particular code
902 snippet has been compiled into an optree containing specified numbers
903 of particular op types. This is good for testing whether optimisations
904 which alter ops, such as converting an "aelem" op into an "aelemfast"
905 op, are really doing that.
906 The files t/perf/speed.t and t/re/speed.t are designed to test things
908 that run thousands of times slower if a particular optimisation is
909 broken (for example, the utf8 length cache on long utf8 strings). Add
910 a test that will take a fraction of a second normally, and minutes
911 otherwise, causing the test file to time out on failure.
912 Building perl at older commits
914 In the course of hacking on the Perl core distribution, you may have
915 occasion to configure, build and test perl at an old commit. Sometimes
916 "make" will fail during this process. If that happens, you may be able
917 to salvage the situation by using the Devel::PatchPerl library from
918 CPAN (not included in the core) to bring the source code at that commit
919 to a buildable state.
920 Here's a real world example, taken from work done to resolve perl
922 #10118 <https://github.com/Perl/perl5/issues/10118>. Use of
923 Porting/bisect.pl had identified commit
924 "ba77e4cc9d1ceebf472c9c5c18b2377ee47062e6" as the commit in which a bug
925 was corrected. To confirm, a P5P developer wanted to configure and
926 build perl at commit "ba77e4c^" (presumably "bad") and then at
927 "ba77e4c" (presumably "good"). Normal configuration and build was
928 attempted:
929 $ sh ./Configure -des -Dusedevel
931 $ make test_prep
932 "make", however, failed with output (excerpted) like this:
934 cc -fstack-protector -L/usr/local/lib -o miniperl \
936 gv.o toke.o perly.o pad.o regcomp.o dump.o util.o \
937 mg.o reentr.o mro.o hv.o av.o run.o pp_hot.o sv.o \
938 pp.o scope.o pp_ctl.o pp_sys.o doop.o doio.o regexec.o \
939 utf8.o taint.o deb.o universal.o globals.o perlio.o \
940 numeric.o mathoms.o locale.o pp_pack.o pp_sort.o \
941 miniperlmain.o opmini.o perlmini.o
942 pp.o: In function `Perl_pp_pow':
943 pp.c:(.text+0x2db9): undefined reference to `pow'
944 ...
945 collect2: error: ld returned 1 exit status
946 makefile:348: recipe for target 'miniperl' failed
947 make: *** [miniperl] Error 1
948 Another P5P contributor recommended installation and use of
950 Devel::PatchPerl for this situation, first to determine the version of
951 perl at the commit in question, then to patch the source code at that
952 point to facilitate a build.
953 $ perl -MDevel::PatchPerl -e \
955 'print Devel::PatchPerl->determine_version("/path/to/sourcecode"),
956 "\n";'
957 5.11.1
958 $ perl -MDevel::PatchPerl -e \
959 'Devel::PatchPerl->patch_source("5.11.1", "/path/to/sourcecode");'
960 Once the source was patched, "./Configure" and "make test_prep" were
962 called and completed successfully, enabling confirmation of the
963 findings in RT #72414.
964
966 To hack on the Perl guts, you'll need to read the following things:
967 • perlsource
969 An overview of the Perl source tree. This will help you find the
971 files you're looking for.
972 • perlinterp
974 An overview of the Perl interpreter source code and some details on
976 how Perl does what it does.
977 • perlhacktut
979 This document walks through the creation of a small patch to Perl's
981 C code. If you're just getting started with Perl core hacking,
982 this will help you understand how it works.
983 • perlhacktips
985 More details on hacking the Perl core. This document focuses on
987 lower level details such as how to write tests, compilation issues,
988 portability, debugging, etc.
989 If you plan on doing serious C hacking, make sure to read this.
991 • perlguts
993 This is of paramount importance, since it's the documentation of
995 what goes where in the Perl source. Read it over a couple of times
996 and it might start to make sense - don't worry if it doesn't yet,
997 because the best way to study it is to read it in conjunction with
998 poking at Perl source, and we'll do that later on.
999 Gisle Aas's "illustrated perlguts", also known as illguts, has very
1001 helpful pictures:
1002 <https://metacpan.org/release/RURBAN/illguts-0.49>
1004 • perlxstut and perlxs
1006 A working knowledge of XSUB programming is incredibly useful for
1008 core hacking; XSUBs use techniques drawn from the PP code, the
1009 portion of the guts that actually executes a Perl program. It's a
1010 lot gentler to learn those techniques from simple examples and
1011 explanation than from the core itself.
1012 • perlapi
1014 The documentation for the Perl API explains what some of the
1016 internal functions do, as well as the many macros used in the
1017 source.
1018 • Porting/pumpkin.pod
1020 This is a collection of words of wisdom for a Perl porter; some of
1022 it is only useful to the pumpkin holders, but most of it applies to
1023 anyone wanting to go about Perl development.
1024
1026 The CPAN testers ( <http://cpantesters.org/> ) are a group of
1027 volunteers who test CPAN modules on a variety of platforms.
1028 Perl Smokers ( <https://www.nntp.perl.org/group/perl.daily-build/> and
1030 <https://www.nntp.perl.org/group/perl.daily-build.reports/> )
1031 automatically test Perl source releases on platforms with various
1032 configurations.
1033 Both efforts welcome volunteers. In order to get involved in smoke
1035 testing of the perl itself visit
1036 <https://metacpan.org/release/Test-Smoke>. In order to start smoke
1037 testing CPAN modules visit
1038 <https://metacpan.org/release/CPANPLUS-YACSmoke> or
1039 <https://metacpan.org/release/minismokebox> or
1040 <https://metacpan.org/release/CPAN-Reporter>.
1041
1043 If you've read all the documentation in the document and the ones
1044 listed above, you're more than ready to hack on Perl.
1045 Here's some more recommendations
1047 • Subscribe to perl5-porters, follow the patches and try and
1049 understand them; don't be afraid to ask if there's a portion you're
1050 not clear on - who knows, you may unearth a bug in the patch...
1051 • Do read the README associated with your operating system, e.g.
1053 README.aix on the IBM AIX OS. Don't hesitate to supply patches to
1054 that README if you find anything missing or changed over a new OS
1055 release.
1056 • Find an area of Perl that seems interesting to you, and see if you
1058 can work out how it works. Scan through the source, and step over
1059 it in the debugger. Play, poke, investigate, fiddle! You'll
1060 probably get to understand not just your chosen area but a much
1061 wider range of perl's activity as well, and probably sooner than
1062 you'd think.
1063 "The Road goes ever on and on, down from the door where it began."
1065 If you can do these things, you've started on the long road to Perl
1066 porting. Thanks for wanting to help make Perl better - and happy
1067 hacking!
1068 Metaphoric Quotations
1070 If you recognized the quote about the Road above, you're in luck.
1071 Most software projects begin each file with a literal description of
1073 each file's purpose. Perl instead begins each with a literary allusion
1074 to that file's purpose.
1075 Like chapters in many books, all top-level Perl source files (along
1077 with a few others here and there) begin with an epigrammatic
1078 inscription that alludes, indirectly and metaphorically, to the
1079 material you're about to read.
1080 Quotations are taken from writings of J.R.R. Tolkien pertaining to his
1082 Legendarium, almost always from The Lord of the Rings. Chapters and
1083 page numbers are given using the following editions:
1084 • The Hobbit, by J.R.R. Tolkien. The hardcover, 70th-anniversary
1086 edition of 2007 was used, published in the UK by Harper Collins
1087 Publishers and in the US by the Houghton Mifflin Company.
1088 • The Lord of the Rings, by J.R.R. Tolkien. The hardcover,
1090 50th-anniversary edition of 2004 was used, published in the UK by
1091 Harper Collins Publishers and in the US by the Houghton Mifflin
1092 Company.
1093 • The Lays of Beleriand, by J.R.R. Tolkien and published posthumously
1095 by his son and literary executor, C.J.R. Tolkien, being the 3rd of
1096 the 12 volumes in Christopher's mammoth History of Middle Earth.
1097 Page numbers derive from the hardcover edition, first published in
1098 1983 by George Allen & Unwin; no page numbers changed for the
1099 special 3-volume omnibus edition of 2002 or the various trade-paper
1100 editions, all again now by Harper Collins or Houghton Mifflin.
1101 Other JRRT books fair game for quotes would thus include The Adventures
1103 of Tom Bombadil, The Silmarillion, Unfinished Tales, and The Tale of
1104 the Children of Hurin, all but the first posthumously assembled by
1105 CJRT. But The Lord of the Rings itself is perfectly fine and probably
1106 best to quote from, provided you can find a suitable quote there.
1107 So if you were to supply a new, complete, top-level source file to add
1109 to Perl, you should conform to this peculiar practice by yourself
1110 selecting an appropriate quotation from Tolkien, retaining the original
1111 spelling and punctuation and using the same format the rest of the
1112 quotes are in. Indirect and oblique is just fine; remember, it's a
1113 metaphor, so being meta is, after all, what it's for.
1114
1116 This document was originally written by Nathan Torkington, and is
1117 maintained by the perl5-porters mailing list.
1118perl v5.36.3 2023-11-30 PERLHACK(1)