1PERLPOLICY(1) Perl Programmers Reference Guide PERLPOLICY(1)
2
6 perlpolicy - Various and sundry policies and commitments related to the
7 Perl core
8
10 This document is the master document which records all written policies
11 about how the Perl 5 Porters collectively develop and maintain the Perl
12 core.
13
15 Perl 5 Porters
16 Subscribers to perl5-porters (the porters themselves) come in several
17 flavours. Some are quiet curious lurkers, who rarely pitch in and
18 instead watch the ongoing development to ensure they're forewarned of
19 new changes or features in Perl. Some are representatives of vendors,
20 who are there to make sure that Perl continues to compile and work on
21 their platforms. Some patch any reported bug that they know how to
22 fix, some are actively patching their pet area (threads, Win32, the
23 regexp -engine), while others seem to do nothing but complain. In
24 other words, it's your usual mix of technical people.
25 Among these people are the core Perl team. These are trusted
27 volunteers involved in the ongoing development of the Perl language and
28 interpreter. They are not required to be language developers or
29 committers.
30 Over this group of porters presides Larry Wall. He has the final word
32 in what does and does not change in any of the Perl programming
33 languages. These days, Larry spends most of his time on Raku, while
34 Perl 5 is shepherded by a steering council of porters responsible for
35 deciding what goes into each release and ensuring that releases happen
36 on a regular basis.
37 Larry sees Perl development along the lines of the US government:
39 there's the Legislature (the porters, represented by the core team),
40 the Executive branch (the steering council), and the Supreme Court
41 (Larry). The legislature can discuss and submit patches to the
42 executive branch all they like, but the executive branch is free to
43 veto them. Rarely, the Supreme Court will side with the executive
44 branch over the legislature, or the legislature over the executive
45 branch. Mostly, however, the legislature and the executive branch are
46 supposed to get along and work out their differences without
47 impeachment or court cases.
48 You might sometimes see reference to Rule 1 and Rule 2. Larry's power
50 as Supreme Court is expressed in The Rules:
51 1. Larry is always by definition right about how Perl should behave.
53 This means he has final veto power on the core functionality.
54 2. Larry is allowed to change his mind about any matter at a later
56 date, regardless of whether he previously invoked Rule 1.
57 Got that? Larry is always right, even when he was wrong. It's rare to
59 see either Rule exercised, but they are often alluded to.
60 For the specifics on how the members of the core team and steering
62 council are elected or rotated, consult perlgov, which spells it all
63 out in detail.
64
66 Perl 5 is developed by a community, not a corporate entity. Every
67 change contributed to the Perl core is the result of a donation.
68 Typically, these donations are contributions of code or time by
69 individual members of our community. On occasion, these donations come
70 in the form of corporate or organizational sponsorship of a particular
71 individual or project.
72 As a volunteer organization, the commitments we make are heavily
74 dependent on the goodwill and hard work of individuals who have no
75 obligation to contribute to Perl.
76 That being said, we value Perl's stability and security and have long
78 had an unwritten covenant with the broader Perl community to support
79 and maintain releases of Perl.
80 This document codifies the support and maintenance commitments that the
82 Perl community should expect from Perl's developers:
83 • We "officially" support the two most recent stable release series.
85 5.26.x and earlier are now out of support. As of the release of
86 5.32.0, we will "officially" end support for Perl 5.28.x, other
87 than providing security updates as described below.
88 • To the best of our ability, we will attempt to fix critical issues
90 in the two most recent stable 5.x release series. Fixes for the
91 current release series take precedence over fixes for the previous
92 release series.
93 • To the best of our ability, we will provide "critical" security
95 patches / releases for any major version of Perl whose 5.x.0
96 release was within the past three years. We can only commit to
97 providing these for the most recent .y release in any 5.x.y series.
98 • We will not provide security updates or bug fixes for development
100 releases of Perl.
101 • We encourage vendors to ship the most recent supported release of
103 Perl at the time of their code freeze.
104 • As a vendor, you may have a requirement to backport security fixes
106 beyond our 3 year support commitment. We can provide limited
107 support and advice to you as you do so and, where possible will try
108 to apply those patches to the relevant -maint branches in git,
109 though we may or may not choose to make numbered releases or
110 "official" patches available. See "SECURITY VULNERABILITY CONTACT
111 INFORMATION" in perlsec for details on how to begin that process.
112
114 Our community has a long-held belief that backward-compatibility is a
115 virtue, even when the functionality in question is a design flaw.
116 We would all love to unmake some mistakes we've made over the past
118 decades. Living with every design error we've ever made can lead to
119 painful stagnation. Unwinding our mistakes is very, very difficult.
120 Doing so without actively harming our users is nearly impossible.
121 Lately, ignoring or actively opposing compatibility with earlier
123 versions of Perl has come into vogue. Sometimes, a change is proposed
124 which wants to usurp syntax which previously had another meaning.
125 Sometimes, a change wants to improve previously-crazy semantics.
126 Down this road lies madness.
128 Requiring end-user programmers to change just a few language
130 constructs, even language constructs which no well-educated developer
131 would ever intentionally use is tantamount to saying "you should not
132 upgrade to a new release of Perl unless you have 100% test coverage and
133 can do a full manual audit of your codebase." If we were to have tools
134 capable of reliably upgrading Perl source code from one version of Perl
135 to another, this concern could be significantly mitigated.
136 We want to ensure that Perl continues to grow and flourish in the
138 coming years and decades, but not at the expense of our user community.
139 Existing syntax and semantics should only be marked for destruction in
141 very limited circumstances. If they are believed to be very rarely
142 used, stand in the way of actual improvement to the Perl language or
143 perl interpreter, and if affected code can be easily updated to
144 continue working, they may be considered for removal. When in doubt,
145 caution dictates that we will favor backward compatibility. When a
146 feature is deprecated, a statement of reasoning describing the decision
147 process will be posted, and a link to it will be provided in the
148 relevant perldelta documents.
149 Using a lexical pragma to enable or disable legacy behavior should be
151 considered when appropriate, and in the absence of any pragma legacy
152 behavior should be enabled. Which backward-incompatible changes are
153 controlled implicitly by a 'use v5.x.y' is a decision which should be
154 made by the steering council in consultation with the community.
155 Historically, we've held ourselves to a far higher standard than
157 backward-compatibility -- bugward-compatibility. Any accident of
158 implementation or unintentional side-effect of running some bit of code
159 has been considered to be a feature of the language to be defended with
160 the same zeal as any other feature or functionality. No matter how
161 frustrating these unintentional features may be to us as we continue to
162 improve Perl, these unintentional features often deserve our
163 protection. It is very important that existing software written in
164 Perl continue to work correctly. If end-user developers have adopted a
165 bug as a feature, we need to treat it as such.
166 New syntax and semantics which don't break existing language constructs
168 and syntax have a much lower bar. They merely need to prove themselves
169 to be useful, elegant, well designed, and well tested. In most cases,
170 these additions will be marked as experimental for some time. See
171 below for more on that.
172 Terminology
174 To make sure we're talking about the same thing when we discuss the
175 removal of features or functionality from the Perl core, we have
176 specific definitions for a few words and phrases.
177 experimental
179 If something in the Perl core is marked as experimental, we may
180 change its behaviour, deprecate or remove it without notice. While
181 we'll always do our best to smooth the transition path for users of
182 experimental features, you should contact the perl5-porters
183 mailinglist if you find an experimental feature useful and want to
184 help shape its future.
185 Experimental features must be experimental in two stable releases
187 before being marked non-experimental. Experimental features will
188 only have their experimental status revoked when they no longer
189 have any design-changing bugs open against them and when they have
190 remained unchanged in behavior for the entire length of a
191 development cycle. In other words, a feature present in v5.20.0
192 may be marked no longer experimental in v5.22.0 if and only if its
193 behavior is unchanged throughout all of v5.21.
194 deprecated
196 If something in the Perl core is marked as deprecated, we may
197 remove it from the core in the future, though we might not.
198 Generally, backward incompatible changes will have deprecation
199 warnings for two release cycles before being removed, but may be
200 removed after just one cycle if the risk seems quite low or the
201 benefits quite high.
202 As of Perl 5.12, deprecated features and modules warn the user as
204 they're used. When a module is deprecated, it will also be made
205 available on CPAN. Installing it from CPAN will silence
206 deprecation warnings for that module.
207 If you use a deprecated feature or module and believe that its
209 removal from the Perl core would be a mistake, please contact the
210 perl5-porters mailinglist and plead your case. We don't deprecate
211 things without a good reason, but sometimes there's a
212 counterargument we haven't considered. Historically, we did not
213 distinguish between "deprecated" and "discouraged" features.
214 discouraged
216 From time to time, we may mark language constructs and features
217 which we consider to have been mistakes as discouraged.
218 Discouraged features aren't currently candidates for removal, but
219 we may later deprecate them if they're found to stand in the way of
220 a significant improvement to the Perl core.
221 removed
223 Once a feature, construct or module has been marked as deprecated,
224 we may remove it from the Perl core. Unsurprisingly, we say we've
225 removed these things. When a module is removed, it will no longer
226 ship with Perl, but will continue to be available on CPAN.
227
229 New releases of maintenance branches should only contain changes that
230 fall into one of the "acceptable" categories set out below, but must
231 not contain any changes that fall into one of the "unacceptable"
232 categories. (For example, a fix for a crashing bug must not be
233 included if it breaks binary compatibility.)
234 It is not necessary to include every change meeting these criteria, and
236 in general the focus should be on addressing security issues, crashing
237 bugs, regressions and serious installation issues. The temptation to
238 include a plethora of minor changes that don't affect the installation
239 or execution of perl (e.g. spelling corrections in documentation)
240 should be resisted in order to reduce the overall risk of overlooking
241 something. The intention is to create maintenance releases which are
242 both worthwhile and which users can have full confidence in the
243 stability of. (A secondary concern is to avoid burning out the maint-
244 release manager or overwhelming other committers voting on changes to
245 be included (see "Getting changes into a maint branch" below).)
246 The following types of change may be considered acceptable, as long as
248 they do not also fall into any of the "unacceptable" categories set out
249 below:
250 • Patches that fix CVEs or security issues. These changes should be
252 passed using the security reporting mechanism rather than applied
253 directly; see "SECURITY VULNERABILITY CONTACT INFORMATION" in
254 perlsec.
255 • Patches that fix crashing bugs, assertion failures and memory
257 corruption but which do not otherwise change perl's functionality
258 or negatively impact performance.
259 • Patches that fix regressions in perl's behavior relative to
261 previous releases, no matter how old the regression, since some
262 people may upgrade from very old versions of perl to the latest
263 version.
264 • Patches that fix bugs in features that were new in the
266 corresponding 5.x.0 stable release.
267 • Patches that fix anything which prevents or seriously impacts the
269 build or installation of perl.
270 • Portability fixes, such as changes to Configure and the files in
272 the hints/ folder.
273 • Minimal patches that fix platform-specific test failures.
275 • Documentation updates that correct factual errors, explain
277 significant bugs or deficiencies in the current implementation, or
278 fix broken markup.
279 • Updates to dual-life modules should consist of minimal patches to
281 fix crashing bugs or security issues (as above). Any changes made
282 to dual-life modules for which CPAN is canonical should be
283 coordinated with the upstream author.
284 The following types of change are NOT acceptable:
286 • Patches that break binary compatibility. (Please talk to the
288 steering council.)
289 • Patches that add or remove features.
291 • Patches that add new warnings or errors or deprecate features.
293 • Ports of Perl to a new platform, architecture or OS release that
295 involve changes to the implementation.
296 • New versions of dual-life modules should NOT be imported into
298 maint. Those belong in the next stable series.
299 If there is any question about whether a given patch might merit
301 inclusion in a maint release, then it almost certainly should not be
302 included.
303 Getting changes into a maint branch
305 Historically, only the single-person project manager cherry-picked
306 changes from bleadperl into maintperl. This has scaling problems. At
307 the same time, maintenance branches of stable versions of Perl need to
308 be treated with great care. To that end, as of Perl 5.12, we have a
309 new process for maint branches.
310 Any committer may cherry-pick any commit from blead to a maint branch
312 by first adding an entry to the relevant voting file in the maint-votes
313 branch announcing the commit as a candidate for back-porting, and then
314 waiting for at least two other committers to add their votes in support
315 of this (i.e. a total of at least three votes is required before a
316 commit may be back-ported).
317 Most of the work involved in both rounding up a suitable set of
319 candidate commits and cherry-picking those for which three votes have
320 been cast will be done by the maint branch release manager, but anyone
321 else is free to add other proposals if they're keen to ensure certain
322 fixes don't get overlooked or fear they already have been.
323 Other voting mechanisms may also be used instead (e.g. sending mail to
325 perl5-porters and at least two other committers responding to the list
326 giving their assent), as long as the same number of votes is gathered
327 in a transparent manner. Specifically, proposals of which changes to
328 cherry-pick must be visible to everyone on perl5-porters so that the
329 views of everyone interested may be heard.
330 It is not necessary for voting to be held on cherry-picking perldelta
332 entries associated with changes that have already been cherry-picked,
333 nor for the maint-release manager to obtain votes on changes required
334 by the Porting/release_managers_guide.pod where such changes can be
335 applied by the means of cherry-picking from blead.
336
338 A Social Contract about Artistic Control
339 What follows is a statement about artistic control, defined as the
340 ability of authors of packages to guide the future of their code and
341 maintain control over their work. It is a recognition that authors
342 should have control over their work, and that it is a responsibility of
343 the rest of the Perl community to ensure that they retain this control.
344 It is an attempt to document the standards to which we, as Perl
345 developers, intend to hold ourselves. It is an attempt to write down
346 rough guidelines about the respect we owe each other as Perl
347 developers.
348 This statement is not a legal contract. This statement is not a legal
350 document in any way, shape, or form. Perl is distributed under the GNU
351 Public License and under the Artistic License; those are the precise
352 legal terms. This statement isn't about the law or licenses. It's
353 about community, mutual respect, trust, and good-faith cooperation.
354 We recognize that the Perl core, defined as the software distributed
356 with the heart of Perl itself, is a joint project on the part of all of
357 us. From time to time, a script, module, or set of modules (hereafter
358 referred to simply as a "module") will prove so widely useful and/or so
359 integral to the correct functioning of Perl itself that it should be
360 distributed with the Perl core. This should never be done without the
361 author's explicit consent, and a clear recognition on all parts that
362 this means the module is being distributed under the same terms as Perl
363 itself. A module author should realize that inclusion of a module into
364 the Perl core will necessarily mean some loss of control over it, since
365 changes may occasionally have to be made on short notice or for
366 consistency with the rest of Perl.
367 Once a module has been included in the Perl core, however, everyone
369 involved in maintaining Perl should be aware that the module is still
370 the property of the original author unless the original author
371 explicitly gives up their ownership of it. In particular:
372 • The version of the module in the Perl core should still be
374 considered the work of the original author. All patches, bug
375 reports, and so forth should be fed back to them. Their
376 development directions should be respected whenever possible.
377 • Patches may be applied by the steering council without the explicit
379 cooperation of the module author if and only if they are very
380 minor, time-critical in some fashion (such as urgent security
381 fixes), or if the module author cannot be reached. Those patches
382 must still be given back to the author when possible, and if the
383 author decides on an alternate fix in their version, that fix
384 should be strongly preferred unless there is a serious problem with
385 it. Any changes not endorsed by the author should be marked as
386 such, and the contributor of the change acknowledged.
387 • The version of the module distributed with Perl should, whenever
389 possible, be the latest version of the module as distributed by the
390 author (the latest non-beta version in the case of public Perl
391 releases), although the steering council may hold off on upgrading
392 the version of the module distributed with Perl to the latest
393 version until the latest version has had sufficient testing.
394 In other words, the author of a module should be considered to have
396 final say on modifications to their module whenever possible (bearing
397 in mind that it's expected that everyone involved will work together
398 and arrive at reasonable compromises when there are disagreements).
399 As a last resort, however:
401 If the author's vision of the future of their module is sufficiently
403 different from the vision of the steering council and perl5-porters as
404 a whole so as to cause serious problems for Perl, the steering council
405 may choose to formally fork the version of the module in the Perl core
406 from the one maintained by the author. This should not be done lightly
407 and should always if at all possible be done only after direct input
408 from Larry. If this is done, it must then be made explicit in the
409 module as distributed with the Perl core that it is a forked version
410 and that while it is based on the original author's work, it is no
411 longer maintained by them. This must be noted in both the
412 documentation and in the comments in the source of the module.
413 Again, this should be a last resort only. Ideally, this should never
415 happen, and every possible effort at cooperation and compromise should
416 be made before doing this. If it does prove necessary to fork a module
417 for the overall health of Perl, proper credit must be given to the
418 original author in perpetuity and the decision should be constantly re-
419 evaluated to see if a remerging of the two branches is possible down
420 the road.
421 In all dealings with contributed modules, everyone maintaining Perl
423 should keep in mind that the code belongs to the original author, that
424 they may not be on perl5-porters at any given time, and that a patch is
425 not official unless it has been integrated into the author's copy of
426 the module. To aid with this, and with points #1, #2, and #3 above,
427 contact information for the authors of all contributed modules should
428 be kept with the Perl distribution.
429 Finally, the Perl community as a whole recognizes that respect for
431 ownership of code, respect for artistic control, proper credit, and
432 active effort to prevent unintentional code skew or communication gaps
433 is vital to the health of the community and Perl itself. Members of a
434 community should not normally have to resort to rules and laws to deal
435 with each other, and this document, although it contains rules so as to
436 be clear, is about an attitude and general approach. The first step in
437 any dispute should be open communication, respect for opposing views,
438 and an attempt at a compromise. In nearly every circumstance nothing
439 more will be necessary, and certainly no more drastic measure should be
440 used until every avenue of communication and discussion has failed.
441
443 Perl's documentation is an important resource for our users. It's
444 incredibly important for Perl's documentation to be reasonably coherent
445 and to accurately reflect the current implementation.
446 Just as P5P collectively maintains the codebase, we collectively
448 maintain the documentation. Writing a particular bit of documentation
449 doesn't give an author control of the future of that documentation. At
450 the same time, just as source code changes should match the style of
451 their surrounding blocks, so should documentation changes.
452 Examples in documentation should be illustrative of the concept they're
454 explaining. Sometimes, the best way to show how a language feature
455 works is with a small program the reader can run without modification.
456 More often, examples will consist of a snippet of code containing only
457 the "important" bits. The definition of "important" varies from
458 snippet to snippet. Sometimes it's important to declare "use strict"
459 and "use warnings", initialize all variables and fully catch every
460 error condition. More often than not, though, those things obscure the
461 lesson the example was intended to teach.
462 As Perl is developed by a global team of volunteers, our documentation
464 often contains spellings which look funny to somebody. Choice of
465 American/British/Other spellings is left as an exercise for the author
466 of each bit of documentation. When patching documentation, try to
467 emulate the documentation around you, rather than changing the existing
468 prose.
469 In general, documentation should describe what Perl does "now" rather
471 than what it used to do. It's perfectly reasonable to include notes in
472 documentation about how behaviour has changed from previous releases,
473 but, with very few exceptions, documentation isn't "dual-life" -- it
474 doesn't need to fully describe how all old versions used to work.
475
477 The official forum for the development of perl is the perl5-porters
478 mailing list, mentioned above, and its bugtracker at GitHub. Posting
479 to the list and the bugtracker is not a right: all participants in
480 discussion are expected to adhere to a standard of conduct.
481 • Always be civil.
483 • Heed the moderators.
485 Civility is simple: stick to the facts while avoiding demeaning
487 remarks, belittling other individuals, sarcasm, or a presumption of bad
488 faith. It is not enough to be factual. You must also be civil.
489 Responding in kind to incivility is not acceptable. If you relay
490 otherwise-unposted comments to the list from a third party, you take
491 responsibility for the content of those comments, and you must
492 therefore ensure that they are civil.
493 While civility is required, kindness is encouraged; if you have any
495 doubt about whether you are being civil, simply ask yourself, "Am I
496 being kind?" and aspire to that.
497 If the list moderators tell you that you are not being civil, carefully
499 consider how your words have appeared before responding in any way.
500 Were they kind? You may protest, but repeated protest in the face of a
501 repeatedly reaffirmed decision is not acceptable. Repeatedly
502 protesting about the moderators' decisions regarding a third party is
503 also unacceptable, as is continuing to initiate off-list contact with
504 the moderators about their decisions.
505 Unacceptable behavior will result in a public and clearly identified
507 warning. A second instance of unacceptable behavior from the same
508 individual will result in removal from the mailing list and GitHub
509 issue tracker, for a period of one calendar month. The rationale for
510 this is to provide an opportunity for the person to change the way they
511 act.
512 After the time-limited ban has been lifted, a third instance of
514 unacceptable behavior will result in a further public warning. A
515 fourth or subsequent instance will result in an indefinite ban. The
516 rationale is that, in the face of an apparent refusal to change
517 behavior, we must protect other community members from future
518 unacceptable actions. The moderators may choose to lift an indefinite
519 ban if the person in question affirms they will not transgress again.
520 Removals, like warnings, are public.
522 The list of moderators will be public knowledge. At present, it is:
524 Karen Etheridge, Neil Bowers, Nicholas Clark, Ricardo Signes, Todd
525 Rinaldo.
526
528 "Social Contract about Contributed Modules" originally by Russ Allbery
529 <rra@stanford.edu> and the perl5-porters.
530perl v5.34.1 2022-03-15 PERLPOLICY(1)