5 perldelta - what is new for perl v5.26.0
9 This document describes the differences between the 5.24.0 release and the
14 This release includes three updates with widespread effects:
18 =item * C<.> no longer in C<@INC>
20 For security reasons, the current directory (C<.>) is no longer included
21 by default at the end of the module search path (C<@INC>). This may have
22 widespread implications for the building, testing and installing of
23 modules, and for the execution of scripts. See the section
24 L<< Removal of the current directory (C<.>) from C<@INC> >>
27 =item * C<do> may now warn
29 C<do> now gives a deprecation warning when it fails to load a file which
30 it would have loaded had C<.> been in C<@INC>.
32 =item * In regular expression patterns, a literal left brace C<"{">
35 See L</Unescaped literal C<"{"> characters in regular expression patterns are no longer permissible>.
39 =head1 Core Enhancements
41 =head2 New regular expression modifier C</xx>
43 Specifying two C<x> characters to modify a regular expression pattern
44 does everything that a single one does, but additionally TAB and SPACE
45 characters within a bracketed character class are generally ignored and
46 can be added to improve readability, like
47 S<C</[ ^ A-Z d-f p-x ]/xx>>. Details are at
48 L<perlre/E<sol>x and E<sol>xx>.
50 =head2 New Hash Function For 64-bit Builds
52 We have switched to a hybrid hash function to better balance
53 performance for short and long keys.
55 For short keys, 16 bytes and under, we use an optimised variant of
56 One At A Time Hard, and for longer keys we use Siphash 1-3. For very
57 long keys this is a big improvement in performance. For shorter keys
58 there is a modest improvement.
60 =head2 Indented Here-documents
62 This adds a new modifier '~' to here-docs that tells the parser
63 that it should look for /^\s*$DELIM\n/ as the closing delimiter.
65 These syntaxes are all supported:
76 The '~' modifier will strip, from each line in the here-doc, the
77 same whitespace that appears before the delimiter.
79 Newlines will be copied as-is, and lines that don't include the
80 proper beginning whitespace will cause perl to croak.
90 prints "Hello there\n" with no leading whitespace.
92 =head2 @{^CAPTURE}, %{^CAPTURE}, and %{^CAPTURE_ALL}
94 C<@{^CAPTURE}> exposes the capture buffers of the last match as an
95 array. So C<$1> is C<${^CAPTURE}[0]>. This is a more efficient equivalent
96 to code like C<substr($matched_string,$-[0],$+[0]-$-[0])>, and you don't
97 have to keep track of the C<$matched_string> either. This variable has no
98 single character equivalent. Note, like the other regex magic variables
99 the contents of this variable is dynamic, if you wish to store it beyond
100 the lifetime of the match you must copy it to another array.
102 C<%{^CAPTURE}> is the equivalent to C<%+> (I<i.e.>, named captures). Other than
103 being more self documenting there is no difference between the two forms.
105 C<%{^CAPTURE_ALL}> is the equivalent to C<%-> (I<i.e.>, all named captures).
106 Other than being more self documenting there is no difference between the
109 =head2 Unicode 9.0 is now supported
111 A list of changes is at L<http://www.unicode.org/versions/Unicode9.0.0/>.
112 Modules that are shipped with core Perl but not maintained by p5p do not
113 necessarily support Unicode 9.0. L<Unicode::Normalize> does work on 9.0.
115 =head2 Use of C<\p{I<script>}> uses the improved Script_Extensions property
117 Unicode 6.0 introduced an improved form of the Script (C<sc>) property, and
118 called it Script_Extensions (C<scx>). Perl now uses this improved
119 version when a property is specified as just C<\p{I<script>}>. This
120 should make programs be more accurate when determining if a character is
121 used in a given script, but there is a slight chance of breakage for
122 programs that very specifically needed the old behavior. The meaning of
123 compound forms, like C<\p{sc=I<script>}> are unchanged. See
124 L<perlunicode/Scripts>.
126 =head2 Declaring a reference to a variable
128 As an experimental feature, Perl now allows the referencing operator to come
129 after L<C<my()>|perlfunc/my>, L<C<state()>|perlfunc/state>,
130 L<C<our()>|perlfunc/our>, or L<C<local()>|perlfunc/local>. This syntax must
131 be enabled with C<use feature 'declared_refs'>. It is experimental, and will
132 warn by default unless C<no warnings 'experimental::refaliasing'> is in effect.
133 It is intended mainly for use in assignments to references. For example:
135 use experimental 'refaliasing', 'declared_refs';
138 See L<perlref/Assigning to References> for more details.
140 =head2 Perl can now do default collation in UTF-8 locales on platforms
143 Some platforms natively do a reasonable job of collating and sorting in
144 UTF-8 locales. Perl now works with those. For portability and full
145 control, L<Unicode::Collate> is still recommended, but now you may
146 not need to do anything special to get good-enough results, depending on
147 your application. See
148 L<perllocale/Category C<LC_COLLATE>: Collation: Text Comparisons and Sorting>.
150 =head2 Better locale collation of strings containing embedded C<NUL>
153 In locales that have multi-level character weights, C<NUL>s are now
154 ignored at the higher priority ones. There are still some gotchas in
155 some strings, though. See
156 L<perllocale/Collation of strings containing embedded C<NUL> characters>.
158 =head2 Lexical subroutines are no longer experimental
160 Using the C<lexical_subs> feature introduced in v5.18 no longer emits a warning. Existing
161 code that disables the C<experimental::lexical_subs> warning category
162 that the feature previously used will continue to work. The
163 C<lexical_subs> feature has no effect; all Perl code can use lexical
164 subroutines, regardless of what feature declarations are in scope.
166 =head2 C<CORE> subroutines for hash and array functions callable via
169 The hash and array functions in the C<CORE> namespace--C<keys>, C<each>,
170 C<values>, C<push>, C<pop>, C<shift>, C<unshift> and C<splice>--, can now
171 be called with ampersand syntax (C<&CORE::keys(\%hash>) and via reference
172 (C<< my $k = \&CORE::keys; $k-E<gt>(\%hash) >>). Previously they could only be
175 =head2 for XS code, create a safer utf8_hop() called utf8_hop_safe()
177 Unlike utf8_hop(), utf8_hop_safe() won't navigate before the beginning or after
178 the end of the supplied buffer.
182 =head2 Removal of the current directory (C<.>) from C<@INC>
184 The perl binary includes a default set of paths in C<@INC>. Historically
185 it has also included the current directory (C<.>) as the final entry,
186 unless run with taint mode enabled (C<perl -T>). While convenient, this has
187 security implications: for example, where a script attempts to load an
188 optional module when its current directory is untrusted (such as F</tmp>),
189 it could load and execute code from under that directory.
191 Starting with v5.26, C<.> is always removed by default, not just under
192 tainting. This has major implications for installing modules and executing
195 The following new features have been added to help ameliorate these
200 =item * C<Configure -Udefault_inc_excludes_dot>
202 There is a new C<Configure> option, C<default_inc_excludes_dot> (enabled
203 by default) which builds a perl executable without C<.>; unsetting this
204 option using C<-U> reverts perl to the old behaviour. This may fix your
205 path issues but will reintroduce all the security concerns, so don't
206 build a perl executable like this unless you're I<really> confident that
207 such issues are not a concern in your environment.
209 =item * C<PERL_USE_UNSAFE_INC>
211 There is a new environment variable recognised by the perl interpreter.
212 If this variable has the value C<1> when the perl interpreter starts up,
213 then C<.> will be automatically appended to C<@INC> (except under tainting).
215 This allows you restore the old perl interpreter behaviour on a
216 case-by-case basis. But note that this is intended to be a temporary crutch,
217 and this feature will likely be removed in some future perl version.
218 It is currently set by the C<cpan> utility and C<Test::Harness> to
219 ease installation of CPAN modules which have not been updated to handle the
220 lack of dot. Once again, don't use this unless you are sure that this
221 will not reintroduce any security concerns.
223 =item * A new mandatory warning issued by C<do>.
225 While it is well-known that C<use> and C<require> use C<@INC> to search
226 for the file to load, many people don't realise that C<do "file"> also
227 searches C<@INC> if the file is a relative path. With the removal of C<.>,
228 a simple C<do "file.pl"> will fail to read in and execute C<file.pl> from
229 the current directory. Since this is commonly expected behaviour, a new
230 mandatory warning is now issued whenever C<do> fails to load a file which
231 it otherwise would have found if dot had been in C<@INC>.
235 Here are some things script and module authors may need to do to make
236 their software work in the new regime.
240 =item * Script authors
242 If the issue is within your own code (rather than within included
243 modules), then you have two main options. Firstly, if you are confident
244 that your script will only be run within a trusted directory (under which
245 you expect to find trusted files and modules), then add C<.> back into the
249 my $dir = "/some/trusted/directory";
250 chdir $dir or die "Can't chdir to $dir: $!\n";
254 use "Foo::Bar"; # may load /some/trusted/directory/Foo/Bar.pm
255 do "config.pl"; # may load /some/trusted/directory/config.pl
257 On the other hand, if your script is intended to be run from within
258 untrusted directories (such as F</tmp>), then your script suddenly failing
259 to load files may be indicative of a security issue. You most likely want
260 to replace any relative paths with full paths; for example,
266 do "$ENV{HOME}/.foo_config.pl"
268 If you are absolutely certain that you want your script to load and
269 execute a file from the current directory, then use a C<./> prefix; for
272 do "./.foo_config.pl"
274 =item * Installing and using CPAN modules
276 If you install a CPAN module using an automatic tool like C<cpan>, then
277 this tool will itself set the C<PERL_USE_UNSAFE_INC> environment variable
278 while building and testing the module, which may be sufficient to install
279 a distribution which hasn't been updated to be dot-aware. If you want to
280 install such a module manually, then you'll need to replace the
281 traditional invocation:
283 perl Makefile.PL && make && make test && make install
287 (export PERL_USE_UNSAFE_INC=1; \
288 perl Makefile.PL && make && make test && make install)
290 Note that this only helps build and install an unfixed module. It's
291 possible for the tests to pass (since they were run under
292 C<PERL_USE_UNSAFE_INC=1>), but for the module itself to fail to perform
293 correctly in production. In this case you may have to temporarily modify
294 your script until such time as a fixed version of the module is released.
299 local @INC = (@INC, '.');
300 # assuming read_config() needs '.' in @INC
301 $config = Foo::Bar->read_config();
304 This is only rarely expected to be necessary. Again, if doing this,
305 assess the resultant risks first.
307 =item * Module Authors
309 If you maintain a CPAN distribution, it may need updating to run in
310 a dotless environment. Although C<cpan> and other such tools will
311 currently set the C<PERL_USE_UNSAFE_INC> during module build, this is a
312 temporary workaround for the set of modules which rely on C<.> being in
313 C<@INC> for installation and testing, and this may mask deeper issues. It
314 could result in a module which passes tests and installs, but which
317 During build, test and install, it will normally be the case that any perl
318 processes will be executing directly within the root directory of the
319 untarred distribution, or a known subdirectory of that, such as F<t/>. It
320 may well be that F<Makefile.PL> or F<t/foo.t> will attempt to include
321 local modules and configuration files using their direct relative
322 filenames, which will now fail.
324 However, as described above, automatic tools like F<cpan> will (for now)
325 set the C<PERL_USE_UNSAFE_INC> environment variable, which introduces
328 This makes it likely that your existing build and test code will work, but
329 this may mask issues with your code which only manifest when used after
330 install. It is prudent to try and run your build process with that
331 variable explicitly disabled:
333 (export PERL_USE_UNSAFE_INC=0; \
334 perl Makefile.PL && make && make test && make install)
336 This is more likely to show up any potential problems with your module's
337 build process, or even with the module itself. Fixing such issues will
338 ensure both that your module can again be installed manually, and that
339 it will still build once the C<PERL_USE_UNSAFE_INC> crutch goes away.
341 When fixing issues in tests due to the removal of dot from C<@INC>,
342 reinsertion of dot into C<@INC> should be performed with caution, for this
343 too may suppress real errors in your runtime code. You are encouraged
344 wherever possible to apply the aforementioned approaches with explicit
345 absolute/relative paths, or to relocate your needed files into a
346 subdirectory and insert that subdirectory into C<@INC> instead.
348 If your runtime code has problems under the dotless C<@INC>, then the comments
349 above on how to fix for script authors will mostly apply here too. Bear in
350 mind though that it is considered bad form for a module to globally add dot to
351 C<@INC>, since it introduces both a security risk and hides issues of
352 accidentally requiring dot in C<@INC>, as explained above.
356 =head2 "Escaped" colons and relative paths in PATH
358 On Unix systems, Perl treats any relative paths in the PATH environment
359 variable as tainted when starting a new process. Previously, it was
360 allowing a backslash to escape a colon (unlike the OS), consequently
361 allowing relative paths to be considered safe if the PATH was set to
362 something like C</\:.>. The check has been fixed to treat C<.> as tainted
365 =head2 C<-Di> switch is now required for PerlIO debugging output
367 Previously PerlIO debugging output would be sent to the file specified
368 by the C<PERLIO_DEBUG> environment variable if perl wasn't running
369 setuid and the C<-T> or C<-t> switches hadn't been parsed yet.
371 If perl performed output at a point where it hadn't yet parsed its
372 switches this could result in perl creating or overwriting the file
373 named by C<PERLIO_DEBUG> even when the C<-T> switch had been supplied.
375 Perl now requires the C<-Di> switch to produce PerlIO debugging
376 output. By default this is written to C<stderr>, but can optionally
377 be redirected to a file by setting the C<PERLIO_DEBUG> environment
380 If perl is running setuid or the C<-T> switch has supplied
381 C<PERLIO_DEBUG> is ignored and the debugging output is sent to
382 C<stderr> as for any other C<-D> switch.
384 =head1 Incompatible Changes
386 =head2 Unescaped literal C<"{"> characters in regular expression
387 patterns are no longer permissible
389 You have to now say something like C<"\{"> or C<"[{]"> to specify to
390 match a LEFT CURLY BRACKET; otherwise it is a fatal pattern compilation
391 error. This change will allow future extensions to the language.
393 These have been deprecated since v5.16, with a deprecation message
394 raised for some uses starting in v5.22. Unfortunately, the code added
395 to raise the message was buggy, and failed to warn in some cases where
396 it should have. Therefore, enforcement of this ban for these cases is
397 deferred until Perl 5.30, but the code has been fixed to raise a
398 default-on deprecation message for them in the meantime.
400 Some uses of literal C<"{"> occur in contexts where we do not foresee
401 the meaning ever being anything but the literal, such as the very first
402 character in the pattern, or after a C<"|"> meaning alternation. Thus
406 matches either of the strings C<{fee> or C<{fie>. To avoid forcing
407 unnecessary code changes, these uses do not need to be escaped, and no
408 warning is raised about them, and there are no current plans to change this.
410 But it is always correct to escape C<"{">, and the simple rule to
411 remember is to always do so.
413 =head2 C<scalar(%hash)> return signature changed
415 The value returned for C<scalar(%hash)> will no longer show information about
416 the buckets allocated in the hash. It will simply return the count of used
417 keys. It is thus equivalent to C<0+keys(%hash)>.
419 A form of backwards compatibility is provided via
420 L<C<Hash::Util::bucket_ratio()>|Hash::Util/bucket_ratio> which provides
422 C<scalar(%hash)> provided in Perl 5.24 and earlier.
424 =head2 C<keys> returned from an lvalue subroutine
426 C<keys> returned from an lvalue subroutine can no longer be assigned
429 sub foo : lvalue { keys(%INC) }
431 sub bar : lvalue { keys(@_) }
432 (bar) = 3; # also an error
434 This makes the lvalue sub case consistent with C<(keys %hash) = ...> and
435 C<(keys @_) = ...>, which are also errors. [perl #128187]
437 =head2 C<${^ENCODING}> has been removed
439 Consequently, the L<encoding> pragma's default mode is no longer supported. If
440 you still need to write your source code in encodings other than UTF-8, use a
441 source filter such as L<Filter::Encoding> on CPAN or L<encoding>'s C<Filter>
444 =head2 POSIX::tmpnam() has been removed
446 The fundamentally unsafe C<tmpnam()> interface was deprecated in
447 Perl 5.22 and has now been removed. In its place you can use,
448 for example, the L<File::Temp> interfaces.
450 =head2 require ::Foo::Bar is now illegal.
452 Formerly, C<require ::Foo::Bar> would try to read F</Foo/Bar.pm>. Now any
453 bareword require which starts with a double colon dies instead.
455 =head2 Literal control character variable names are no longer permissible
457 A variable name may no longer contain a literal control character under
458 any circumstances. These previously were allowed in single-character
459 names on ASCII platforms, but have been deprecated there since Perl
460 5.20. This affects things like C<$I<\cT>>, where I<\cT> is a literal
461 control (such as a C<NAK> or C<NEGATIVE ACKNOWLEDGE> character) in the
464 =head2 C<NBSP> is no longer permissible in C<\N{...}>
466 The name of a character may no longer contain non-breaking spaces. It
467 has been deprecated to do so since Perl 5.22.
471 =head2 String delimiters that aren't stand-alone graphemes are now deprecated
473 In order for Perl to eventually allow string delimiters to be Unicode
474 grapheme clusters (which look like a single character, but may be
475 a sequence of several ones), we have to stop allowing a single character
476 delimiter that isn't a grapheme by itself. These are unlikely to exist
477 in actual code, as they would typically display as attached to the
478 character in front of them.
480 =head1 Removed Deprecations
482 The C<\cI<X>> construct is intended to be a way to specify non-printable
483 characters. Previously it was deprecated to use it for a printable one,
484 which is better written as simply itself, perhaps preceded by a
485 backslash for non-word characters. Now this raises a warning, but not a
487 L<http://www.nntp.perl.org/group/perl.perl5.porters/2017/02/msg242944.html>.
489 =head1 Performance Enhancements
495 A hash in boolean context is now sometimes faster, I<e.g.>
499 This was already special-cased, but some cases were missed (such as
500 C<grep %$_, @AoH>, and even the ones which weren't have been improved.
502 =item * New Faster Hash Function on 64 bit builds
504 We use a different hash function for short and long keys. This should
505 improve performance and security, especially for long keys.
507 =item * readline is faster
509 Reading from a file line-by-line with C<readline()> or C<< E<lt>E<gt> >> should
510 now typically be faster due to a better implementation of the code that
511 searches for the next newline character.
515 Reduce cost of SvVALID().
519 C<$ref1 = $ref2> has been optimized.
523 Array and hash assignment are now faster, I<e.g.>
528 especially when the RHS is empty.
532 Reduce the number of odd special cases for the C<SvSCREAM> flag.
536 Avoid sv_catpvn() in do_vop() when unneeded.
540 Enhancements in Regex concat COW implementation.
544 Better optimise array and hash assignment: where an array or hash appears
545 in the LHS of a list assignment, such as C<(..., @a) = (...);>, it's
546 likely to be considerably faster, especially if it involves emptying the
547 array/hash. For example this code runs about 1/3 faster compared to
551 for my $i (1..10_000_000) {
559 Converting a single-digit string to a number is now substantially faster.
563 The internal op implementing the C<split> builtin has been simplified and
564 sped up. Firstly, it no longer requires a subsidiary internal C<pushre> op
565 to do its work. Secondly, code of the form C<my @x = split(...)> is now
566 optimised in the same way as C<@x = split(...)>, and is therefore a few
571 The rather slow implementation for the experimental subroutine signatures
572 feature has been made much faster; it is now comparable in speed with the
573 old-style C<my ($a, $b, @c) = @_>.
577 Bareword constant strings are now permitted to take part in constant
578 folding. They were originally exempted from constant folding in August 1999,
579 during the development of Perl 5.6, to ensure that C<use strict "subs">
580 would still apply to bareword constants. That has now been accomplished a
581 different way, so barewords, like other constants, now gain the performance
582 benefits of constant folding.
584 This also means that void-context warnings on constant expressions of
585 barewords now report the folded constant operand, rather than the operation;
586 this matches the behaviour for non-bareword constants.
590 =head1 Modules and Pragmata
592 =head2 Updated Modules and Pragmata
598 L<Archive::Tar> has been upgraded from version 2.04 to 2.24.
602 L<arybase> has been upgraded from version 0.11 to 0.12.
606 L<attributes> has been upgraded from version 0.27 to 0.29.
608 The deprecation message for the C<:unique> and C<:locked> attributes
609 now mention that they will disappear in Perl 5.28.
613 L<B> has been upgraded from version 1.62 to 1.68.
617 L<B::Concise> has been upgraded from version 0.996 to 0.999.
619 Its output is now more descriptive for C<op_private> flags.
623 L<B::Debug> has been upgraded from version 1.23 to 1.24.
627 L<B::Deparse> has been upgraded from version 1.37 to 1.40.
631 L<B::Xref> has been upgraded from version 1.05 to 1.06.
633 It now uses 3-arg C<open()> instead of 2-arg C<open()>. [perl #130122]
637 L<base> has been upgraded from version 2.23 to 2.25.
641 L<bignum> has been upgraded from version 0.42 to 0.47.
645 L<Carp> has been upgraded from version 1.40 to 1.42.
649 L<charnames> has been upgraded from version 1.43 to 1.44.
653 L<Compress::Raw::Bzip2> has been upgraded from version 2.069 to 2.074.
657 L<Compress::Raw::Zlib> has been upgraded from version 2.069 to 2.074.
661 L<Config::Perl::V> has been upgraded from version 0.25 to 0.28.
665 L<CPAN> has been upgraded from version 2.11 to 2.18.
669 L<CPAN::Meta> has been upgraded from version 2.150005 to 2.150010.
673 L<Data::Dumper> has been upgraded from version 2.160 to 2.167.
675 The XS implementation now supports Deparse.
677 This fixes a stack management bug. [perl #130487].
681 L<DB_File> has been upgraded from version 1.835 to 1.840.
685 L<Devel::Peek> has been upgraded from version 1.23 to 1.26.
689 L<Devel::PPPort> has been upgraded from version 3.32 to 3.35.
693 L<Devel::SelfStubber> has been upgraded from version 1.05 to 1.06.
695 It now uses 3-arg C<open()> instead of 2-arg C<open()>. [perl #130122]
699 L<diagnostics> has been upgraded from version 1.34 to 1.36.
701 It now uses 3-arg C<open()> instead of 2-arg C<open()>. [perl #130122]
705 L<Digest> has been upgraded from version 1.17 to 1.17_01.
709 L<Digest::MD5> has been upgraded from version 2.54 to 2.55.
713 L<Digest::SHA> has been upgraded from version 5.95 to 5.96.
717 L<DynaLoader> has been upgraded from version 1.38 to 1.42.
721 L<Encode> has been upgraded from version 2.80 to 2.88.
725 L<encoding> has been upgraded from version 2.17 to 2.19.
727 This module's default mode is no longer supported. It now
728 dies when imported, unless the C<Filter> option is being used.
732 L<encoding::warnings> has been upgraded from version 0.12 to 0.13.
734 This module is no longer supported. It emits a warning to
735 that effect and then does nothing.
739 L<Errno> has been upgraded from version 1.25 to 1.28.
741 Document that using C<%!> loads Errno for you.
743 It now uses 3-arg C<open()> instead of 2-arg C<open()>. [perl #130122]
747 L<ExtUtils::Embed> has been upgraded from version 1.33 to 1.34.
749 It now uses 3-arg C<open()> instead of 2-arg C<open()>. [perl #130122]
753 L<ExtUtils::MakeMaker> has been upgraded from version 7.10_01 to 7.24.
757 L<ExtUtils::Miniperl> has been upgraded from version 1.05 to 1.06.
761 L<ExtUtils::ParseXS> has been upgraded from version 3.31 to 3.34.
765 L<ExtUtils::Typemaps> has been upgraded from version 3.31 to 3.34.
769 L<feature> has been upgraded from version 1.42 to 1.47.
771 Fixes the Unicode Bug in the range operator.
775 L<File::Copy> has been upgraded from version 2.31 to 2.32.
779 L<File::Fetch> has been upgraded from version 0.48 to 0.52.
783 L<File::Glob> has been upgraded from version 1.26 to 1.28.
785 Issue a deprecation message for C<File::Glob::glob()>.
789 L<File::Spec> has been upgraded from version 3.63 to 3.67.
793 L<FileHandle> has been upgraded from version 2.02 to 2.03.
797 L<Filter::Simple> has been upgraded from version 0.92 to 0.93.
799 It no longer treats C<no MyFilter> immediately following C<use MyFilter> as
800 end-of-file. [perl #107726]
804 L<Getopt::Long> has been upgraded from version 2.48 to 2.49.
808 L<Getopt::Std> has been upgraded from version 1.11 to 1.12.
812 L<Hash::Util> has been upgraded from version 0.19 to 0.22.
816 L<HTTP::Tiny> has been upgraded from version 0.056 to 0.070.
818 Internal 599-series errors now include the redirect history.
822 L<I18N::LangTags> has been upgraded from version 0.40 to 0.42.
824 It now uses 3-arg C<open()> instead of 2-arg C<open()>. [perl #130122]
828 L<IO> has been upgraded from version 1.36 to 1.38.
832 IO-Compress has been upgraded from version 2.069 to 2.074.
836 L<IO::Socket::IP> has been upgraded from version 0.37 to 0.38.
840 L<IPC::Cmd> has been upgraded from version 0.92 to 0.96.
844 L<IPC::SysV> has been upgraded from version 2.06_01 to 2.07.
848 L<JSON::PP> has been upgraded from version 2.27300 to 2.27400_02.
852 L<lib> has been upgraded from version 0.63 to 0.64.
854 It now uses 3-arg C<open()> instead of 2-arg C<open()>. [perl #130122]
858 L<List::Util> has been upgraded from version 1.42_02 to 1.46_02.
862 L<Locale::Codes> has been upgraded from version 3.37 to 3.42.
866 L<Locale::Maketext> has been upgraded from version 1.26 to 1.28.
870 L<Locale::Maketext::Simple> has been upgraded from version 0.21 to 0.21_01.
874 L<Math::BigInt> has been upgraded from version 1.999715 to 1.999806.
876 There have also been some core customizations.
880 L<Math::BigInt::FastCalc> has been upgraded from version 0.40 to 0.5005.
884 L<Math::BigRat> has been upgraded from version 0.260802 to 0.2611.
888 L<Math::Complex> has been upgraded from version 1.59 to 1.5901.
892 L<Memoize> has been upgraded from version 1.03 to 1.03_01.
896 L<Module::CoreList> has been upgraded from version 5.20170420 to 5.20170520.
900 L<Module::Load::Conditional> has been upgraded from version 0.64 to 0.68.
904 L<Module::Metadata> has been upgraded from version 1.000031 to 1.000033.
908 L<mro> has been upgraded from version 1.18 to 1.20.
912 L<Net::Ping> has been upgraded from version 2.43 to 2.55.
914 IPv6 addresses and C<AF_INET6> sockets are now supported, along with several
917 Remove sudo from 500_ping_icmp.t.
919 Avoid stderr noise in tests
921 Check for echo in new L<Net::Ping> tests.
925 L<NEXT> has been upgraded from version 0.65 to 0.67.
929 L<Opcode> has been upgraded from version 1.34 to 1.39.
933 L<open> has been upgraded from version 1.10 to 1.11.
937 L<OS2::Process> has been upgraded from version 1.11 to 1.12.
939 It now uses 3-arg C<open()> instead of 2-arg C<open()>. [perl #130122]
943 L<overload> has been upgraded from version 1.26 to 1.28.
945 Its compilation speed has been improved slightly.
949 L<parent> has been upgraded from version 0.234 to 0.236.
953 L<perl5db.pl> has been upgraded from version 1.50 to 1.51.
955 Ignore F</dev/tty> on non-Unix systems. [perl #113960]
959 L<Perl::OSType> has been upgraded from version 1.009 to 1.010.
963 L<perlfaq> has been upgraded from version 5.021010 to 5.021011.
967 L<PerlIO> has been upgraded from version 1.09 to 1.10.
971 L<PerlIO::encoding> has been upgraded from version 0.24 to 0.25.
975 L<PerlIO::scalar> has been upgraded from version 0.24 to 0.26.
979 L<Pod::Checker> has been upgraded from version 1.60 to 1.73.
983 L<Pod::Functions> has been upgraded from version 1.10 to 1.11.
987 L<Pod::Html> has been upgraded from version 1.22 to 1.2202.
991 L<Pod::Perldoc> has been upgraded from version 3.25_02 to 3.28.
995 L<Pod::Simple> has been upgraded from version 3.32 to 3.35.
999 L<Pod::Usage> has been upgraded from version 1.68 to 1.69.
1003 L<POSIX> has been upgraded from version 1.65 to 1.76. This remedies several
1004 defects in making its symbols exportable. [perl #127821]
1005 The C<POSIX::tmpnam()> interface has been removed,
1006 see L</"POSIX::tmpnam() has been removed">.
1007 Trying to import POSIX subs that have no real implementations
1008 (like C<POSIX::atend()>) now fails at import time, instead of
1009 waiting until runtime.
1013 L<re> has been upgraded from version 0.32 to 0.34
1015 This adds support for the new L<C<E<47>xx>|perlre/E<sol>x and E<sol>xx>
1016 regular expression pattern modifier, and a change to the L<S<C<use re
1017 'strict'>>|re/'strict' mode> experimental feature. When S<C<re
1018 'strict'>> is enabled, a warning now will be generated for all
1019 unescaped uses of the two characters C<}> and C<]> in regular
1020 expression patterns (outside bracketed character classes) that are taken
1021 literally. This brings them more in line with the C<)> character which
1022 is always a metacharacter unless escaped. Being a metacharacter only
1023 sometimes, depending on action at a distance, can lead to silently
1024 having the pattern mean something quite different than was intended,
1025 which the S<C<re 'strict'>> mode is intended to minimize.
1029 L<Safe> has been upgraded from version 2.39 to 2.40.
1033 L<Scalar::Util> has been upgraded from version 1.42_02 to 1.46_02.
1037 L<Storable> has been upgraded from version 2.56 to 2.62.
1039 Fixes [perl #130098].
1043 L<Symbol> has been upgraded from version 1.07 to 1.08.
1047 L<Sys::Syslog> has been upgraded from version 0.33 to 0.35.
1051 L<Term::ANSIColor> has been upgraded from version 4.04 to 4.06.
1055 L<Term::ReadLine> has been upgraded from version 1.15 to 1.16.
1057 It now uses 3-arg C<open()> instead of 2-arg C<open()>. [perl #130122]
1061 L<Test> has been upgraded from version 1.28 to 1.30.
1063 It now uses 3-arg C<open()> instead of 2-arg C<open()>. [perl #130122]
1067 L<Test::Harness> has been upgraded from version 3.36 to 3.38.
1071 L<Test::Simple> has been upgraded from version 1.001014 to 1.302073.
1075 L<Thread::Queue> has been upgraded from version 3.09 to 3.12.
1079 L<Thread::Semaphore> has been upgraded from 2.12 to 2.13.
1081 Added the C<down_timed> method.
1085 L<threads> has been upgraded from version 2.07 to 2.15.
1087 Compatibility with 5.8 has been restored.
1089 Fixes [perl #130469].
1093 L<threads::shared> has been upgraded from version 1.51 to 1.56.
1095 This fixes [cpan #119529], [perl #130457]
1099 L<Tie::Hash::NamedCapture> has been upgraded from version 0.09 to 0.10.
1103 L<Time::HiRes> has been upgraded from version 1.9733 to 1.9741.
1105 It now builds on systems with C++11 compilers (such as G++ 6 and Clang++
1108 Now uses C<clockid_t>.
1112 L<Time::Local> has been upgraded from version 1.2300 to 1.25.
1116 L<Unicode::Collate> has been upgraded from version 1.14 to 1.19.
1120 L<Unicode::UCD> has been upgraded from version 0.64 to 0.68.
1122 It now uses 3-arg C<open()> instead of 2-arg C<open()>. [perl #130122]
1126 L<version> has been upgraded from version 0.9916 to 0.9917.
1130 L<VMS::DCLsym> has been upgraded from version 1.06 to 1.08.
1132 It now uses 3-arg C<open()> instead of 2-arg C<open()>. [perl #130122]
1136 L<warnings> has been upgraded from version 1.36 to 1.37.
1140 L<XS::Typemap> has been upgraded from version 0.14 to 0.15.
1144 L<XSLoader> has been upgraded from version 0.21 to 0.27.
1146 Fixed a security hole in which binary files could be loaded from a path
1147 outside of L<C<@INC>|perlvar/@INC>.
1149 It now uses 3-arg C<open()> instead of 2-arg C<open()>. [perl #130122]
1153 =head1 Documentation
1155 =head2 New Documentation
1157 =head3 L<perldeprecation>
1159 This file documents all upcoming deprecations, and some of the deprecations
1160 which already have been removed. The purpose of this documentation is
1161 two-fold: document what will disappear, and by which version, and serve
1162 as a guide for people dealing with code which has features that no longer
1163 work after an upgrade of their perl.
1165 =head2 Changes to Existing Documentation
1167 We have attempted to update the documentation to reflect the changes
1168 listed in this document. If you find any we have missed, send email to
1169 L<mailto:perlbug@perl.org>.
1171 Additionally all references to Usenet have been removed, and the
1172 following selected changes have been made:
1180 Removed obsolete text about L<C<defined()>|perlfunc/defined>
1181 on aggregates that should have been deleted earlier, when the feature
1186 Corrected documentation of L<C<eval()>|perlfunc/eval>,
1187 and L<C<evalbytes()>|perlfunc/evalbytes>.
1191 Clarified documentation of L<C<seek()>|perlfunc/seek>,
1192 L<C<tell()>|perlfunc/tell> and L<C<sysseek()>|perlfunc/sysseek>
1193 regarding that positions are in bytes vs. characters.
1194 L<[perl #128607]|https://rt.perl.org/Public/Bug/Display.html?id=128607>
1198 Clarified documentation of L<C<sort()>|perlfunc/sort LIST> concerning
1199 the variables C<$a> and C<$b>.
1203 In L<C<split()>|perlfunc/split> added a caution about its use in Perls
1204 before 5.11, and noted that certain pattern modifiers are legal in it.
1208 Removed obsolete documentation of L<C<study()>|perlfunc/study>, noting
1209 that it is now a no-op.
1213 Noted that L<C<vec()>|perlfunc/vec> doesn't work well when the string
1214 contains characters whose code points are above 255.
1225 L<formatted printing of operands of C<Size_t> and C<SSize_t>|perlguts/Formatted Printing of Size_t and SSize_t>
1235 Clariy indentation rules, and note that we are migrating away from using
1236 tabs to indent, replacing them with sequences of SPACE characters.
1240 =head3 L<perlhacktips>
1246 Give another reason to use C<cBOOL> to cast an expression to boolean.
1250 Note that there are macros C<TRUE> and C<FALSE> available to express
1255 =head3 L<perlinterp>
1261 L<perlinterp> has been expanded to give a more detailed example of how to
1262 hunt around in the parser for how a given operator is handled.
1266 =head3 L<perllocale>
1270 Some locales aren't compatible with Perl. Note that these can cause
1275 =head3 L<perlmodinstall>
1281 Various clarifications have been added.
1285 =head3 L<perlmodlib>
1291 Updated the site mirror list.
1301 Added a section on calling methods using their fully qualified names.
1305 Do not discourage manual @ISA.
1315 Mention C<Moo> more.
1325 Clarify behavior of single quote regexps.
1335 The first part was extensively rewritten to incorporate various basic
1336 points, that in earlier versions were mentioned in sort of an appendix
1337 on Version 8 regular expressions.
1339 Note that it is common to have the C</x> modifier and forget that this
1340 means that C<"#"> has to be escaped.
1350 Add introductory material
1354 Note that a metacharacter occurring in a context where it can't mean
1355 that, silently loses its meta-ness and matches literally.
1356 L<C<use re 'strict'>|re/'strict' mode> can catch some of these.
1360 =head3 L<perlunicode>
1366 Corrected the text about Unicode BYTE ORDER MARK handling.
1370 Updated the text to correspond with changes in Unicode UTS#18, concerning
1371 regular expressions, and Perl compatibility with what it says.
1381 Document C<@ISA>. Was documented other places, not not in L<perlvar>.
1387 =head2 New Diagnostics
1395 Since C<.> is now removed from C<@INC> by default, C<do> will now trigger
1396 a warning recommending to fix the C<do> statement:
1398 L<do "%s" failed, '.' is no longer in @INC|perldiag/do "%s" failed, '.' is no longer in @INC; did you mean do ".E<sol>%s"?>
1402 Using the empty pattern (which re-executes the last successfully-matched
1403 pattern) inside a code block in another regex, as in C</(?{ s!!new! })/>, has
1404 always previously yielded a segfault. It now produces an error:
1405 L<Infinite recursion in regex|perldiag/"Infinite recursion in regex">.
1409 L<The experimental declared_refs feature is not enabled|perldiag/"The experimental declared_refs feature is not enabled">
1411 (F) To declare references to variables, as in C<my \%x>, you must first enable
1414 no warnings "experimental::declared_refs";
1415 use feature "declared_refs";
1419 L<Version control conflict marker|perldiag/"Version control conflict marker">
1421 (F) The parser found a line starting with C<E<lt>E<lt>E<lt>E<lt>E<lt>E<lt>E<lt>>,
1422 C<E<gt>E<gt>E<gt>E<gt>E<gt>E<gt>E<gt>>, or C<=======>. These may be left by a
1423 version control system to mark conflicts after a failed merge operation.
1427 L<%s: command not found|perldiag/"%s: command not found">
1429 (A) You've accidentally run your script through B<bash> or another shell
1430 instead of Perl. Check the #! line, or manually feed your script into
1431 Perl yourself. The #! line at the top of your file could look like:
1437 L<%s: command not found: %s|perldiag/"%s: command not found: %s">
1439 (A) You've accidentally run your script through B<zsh> or another shell
1440 instead of Perl. Check the #! line, or manually feed your script into
1441 Perl yourself. The #! line at the top of your file could look like:
1447 L<Unescaped left brace in regex is illegal here|perldiag/Unescaped left brace in regex is illegal here in regex; marked by S<E<lt>-- HERE> in m/%s/>
1449 Unescaped left braces are now illegal in some contexts in regular expression
1450 patterns. In other contexts, they are still just deprecated; they will
1451 be illegal in Perl 5.30.
1455 L<Bareword in require contains "%s"|perldiag/"Bareword in require contains "%s"">
1459 L<Bareword in require maps to empty filename|perldiag/"Bareword in require maps to empty filename">
1463 L<Bareword in require maps to disallowed filename "%s"|perldiag/"Bareword in require maps to disallowed filename "%s"">
1467 L<Bareword in require must not start with a double-colon: "%s"|perldiag/"Bareword in require must not start with a double-colon: "%s"">
1477 L<Use of unassigned code point or non-standalone grapheme for a delimiter will be a fatal error starting in Perl 5.30|perldiag/"Use of unassigned code point or non-standalone grapheme for a delimiter will be a fatal error starting in Perl 5.30">
1479 See L</Deprecations>
1483 L<Declaring references is experimental|perldiag/"Declaring references is experimental">
1485 (S experimental::declared_refs) This warning is emitted if you use a reference
1486 constructor on the right-hand side of C<my()>, C<state()>, C<our()>, or
1487 C<local()>. Simply suppress the warning if you want to use the feature, but
1488 know that in doing so you are taking the risk of using an experimental feature
1489 which may change or be removed in a future Perl version:
1491 no warnings "experimental::declared_refs";
1492 use feature "declared_refs";
1497 L<C<${^ENCODING}> is no longer supported. Its use will be fatal in Perl 5.28|perldiag/"${^ENCODING} is no longer supported. Its use will be fatal in Perl 5.28">
1499 The special variable C<${^ENCODING}>, formerly used to implement
1500 the C<encoding> pragma, is no longer supported as of Perl 5.26.
1504 Since C<.> is now removed from C<@INC> by default, C<do> will now trigger
1505 a warning recommending to fix the C<do> statement:
1507 L<do "%s" failed, '.' is no longer in @INC|perldiag/do "%s" failed, '.' is no longer in @INC; did you mean do ".E<sol>%s"?>
1511 =head2 Changes to Existing Diagnostics
1517 When a C<require> fails, we now do not provide C<@INC> when the C<require>
1518 is for a file instead of a module.
1522 When C<@INC> is not scanned for a C<require> call, we no longer display
1523 C<@INC> to avoid confusion.
1527 Attribute "locked" is deprecated, and will disappear in Perl 5.28
1531 Attribute "unique" is deprecated, and will disappear in Perl 5.28
1535 Constants from lexical variables potentially modified elsewhere are
1536 deprecated. This will not be allowed in Perl 5.32
1540 Deprecated use of my() in false conditional. This will be a fatal error
1545 dump() better written as CORE::dump(). dump() will no longer be available
1550 ${^ENCODING} is no longer supported. Its use will be fatal in Perl 5.28
1554 File::Glob::glob() will disappear in perl 5.30. Use File::Glob::bsd_glob()
1559 %s() is deprecated on :utf8 handles. This will be a fatal error in Perl 5.30
1563 $* is no longer supported. Its use will be fatal in Perl 5.30
1567 $* is no longer supported. Its use will be fatal in Perl 5.30
1571 Opening dirhandle %s also as a file. This will be a fatal error in Perl 5.28
1575 Opening filehandle %s also as a directory. This will be a fatal
1580 Setting $/ to a reference to %s as a form of slurp is deprecated,
1581 treating as undef. This will be fatal in Perl 5.28
1585 Unescaped left brace in regex is deprecated here (and will be fatal
1586 in Perl 5.30), passed through in regex; marked by S<< E<lt>-- HERE >> in m/%s/
1590 Unknown charname '' is deprecated. Its use will be fatal in Perl 5.28
1594 Use of bare E<lt>E<lt> to mean E<lt>E<lt>"" is deprecated. Its use will be fatal in Perl 5.28
1598 Use of code point 0x%s is deprecated; the permissible max is 0x%s.
1599 This will be fatal in Perl 5.28
1603 Use of comma-less variable list is deprecated. Its use will be fatal
1608 Use of inherited AUTOLOAD for non-method %s() is deprecated. This
1609 will be fatal in Perl 5.28
1613 Use of strings with code points over 0xFF as arguments to %s operator
1614 is deprecated. This will be a fatal error in Perl 5.28
1618 Improve error for missing tie() package/method. This brings the error messages
1619 in line with the ones used for normal method calls, despite not using
1624 Make the sysread()/syswrite/() etc :utf8 handle warnings default. These
1625 warnings were under 'deprecated' previously.
1629 'do' errors now refer to 'do' (not 'require').
1633 Details as to the exact problem have been added to the diagnostics that
1634 occur when malformed UTF-8 is encountered when trying to convert to a
1639 Executing C<undef $x> where C<$x> is tied or magical no longer incorrectly
1640 blames the variable for an uninitialized-value warning encountered by the
1645 Code like C<$x = $x . "a"> was incorrectly failing to yield a
1646 L<use of uninitialized value|perldiag/"Use of uninitialized value%s">
1647 warning when C<$x> was a lexical variable with an undefined value. That has
1648 now been fixed. [perl #127877]
1652 When the error "Experimental push on scalar is now forbidden" is raised for
1653 the hash functions C<keys>, C<each>, and C<values>, it is now followed by
1654 the more helpful message, "Type of arg 1 to whatever must be hash or
1655 array". [perl #127976]
1659 C<undef *_; shift> or C<undef *_; pop> inside a subroutine, with no
1660 argument to C<shift> or C<pop>, began crashing in Perl 5.14, but has now
1665 C<< "string$scalar-E<gt>$*" >> now correctly prefers concat overloading to
1666 string overloading if C<< $scalar-E<gt>$* >> returns an overloaded object,
1667 bringing it into consistency with C<$$scalar>.
1671 C<< /@0{0*-E<gt>@*/*0 >> and similar contortions used to crash, but no longer
1672 do, but merely produce a syntax error. [perl #128171]
1676 C<do> or C<require> with a reference or typeglob which, when stringified,
1677 contains a null character started crashing in Perl 5.20, but has now been
1678 fixed. [perl #128182]
1682 =head1 Utility Changes
1684 =head2 F<c2ph> and F<pstruct>
1690 These old utilities have long since superceded by L<h2xs>, and are
1691 now gone from the distribution.
1695 =head2 F<Porting/pod_lib.pl>
1701 Removed spurious executable bit.
1705 Account for possibility of DOS file endings.
1709 =head2 F<Porting/sync-with-cpan>
1719 =head2 F<perf/benchmarks>
1725 Tidy file, rename some symbols.
1729 =head2 F<Porting/checkAUTHORS.pl>
1735 Replace obscure character range with C<\w>.
1739 =head2 F<t/porting/regen.t>
1745 try to be more helpful when tests fail.
1749 =head2 F<utils/h2xs.PL>
1755 Avoid infinite loop for enums.
1765 Long lines in the message body are now wrapped at 900 characters, to stay
1766 well within the 1000-character limit imposed by SMTP mail transfer agents.
1767 This is particularly likely to be important for the list of arguments to
1768 C<Configure>, which can readily exceed the limit if, for example, it names
1769 several non-default installation paths. This change also adds the first unit
1770 tests for perlbug. [perl #128020]
1774 =head1 Configuration and Compilation
1780 C<-Ddefault_inc_excludes_dot> has been turned on by default.
1784 The C<dtrace> build process has further changes:
1790 If the C<-xnolibs> is available, use that so a F<dtrace> perl can be
1791 built within a FreeBSD jail.
1795 On systems that build a dtrace object file (FreeBSD, Solaris and
1796 SystemTap's dtrace emulation), copy the input objects to a separate
1797 directory and process them there, and use those objects in the link,
1798 since C<dtrace -G> also modifies these objects.
1802 Add libelf to the build on FreeBSD 10.x, since dtrace adds references
1807 Generate a dummy dtrace_main.o if C<dtrace -G> fails to build it. A
1808 default build on Solaris generates probes from the unused inline
1809 functions, while they don't on FreeBSD, which causes C<dtrace -G> to
1818 You can now disable perl's use of the PERL_HASH_SEED and
1819 PERL_PERTURB_KEYS environment variables by configuring perl with
1820 C<-Accflags=NO_PERL_HASH_ENV>.
1824 You can now disable perl's use of the PERL_HASH_SEED_DEBUG environment
1825 variable by configuring perl with
1826 C<-Accflags=-DNO_PERL_HASH_SEED_DEBUG>.
1830 Zero out the alignment bytes when calculating the bytes for 80-bit C<NaN>
1831 and C<Inf> to make builds more reproducible. [perl #130133]
1835 Since 5.18 for testing purposes we have included support for
1836 building perl with a variety of non-standard, and non-recommended
1837 hash functions. Since we do not recommend the use of these functions
1838 we have removed them and their corresponding build options. Specifically
1839 this includes the following build options:
1843 PERL_HASH_FUNC_SUPERFAST
1844 PERL_HASH_FUNC_MURMUR3
1845 PERL_HASH_FUNC_ONE_AT_A_TIME
1846 PERL_HASH_FUNC_ONE_AT_A_TIME_OLD
1847 PERL_HASH_FUNC_MURMUR_HASH_64A
1848 PERL_HASH_FUNC_MURMUR_HASH_64B
1852 Remove "Warning: perl appears in your path"
1854 This install warning is more or less obsolete, since most platforms already
1855 *will* have a /usr/bin/perl or similar provided by the OS.
1859 Reduce verbosity of "make install.man"
1861 Previously, two progress messages were emitted for each manpage: one by
1862 installman itself, and one by the function in install_lib.pl that it calls to
1863 actually install the file. Disabling the second of those in each case saves
1864 over 750 lines of unhelpful output.
1868 Cleanup for clang -Weverything support. [perl 129961]
1872 Configure: signbit scan was assuming too much, stop assuming negative 0.
1876 Various compiler warnings have been silenced.
1880 Several smaller changes have been made to remove impediments to compiling under
1885 Builds using C<USE_PAD_RESET> now work again; this configuration had
1890 A probe for C<gai_strerror> was added to F<Configure> that checks if the
1891 the gai_strerror() routine is available and can be used to
1892 translate error codes returned by getaddrinfo() into human
1897 F<Configure> now aborts if both "-Duselongdouble" and "-Dusequadmath" are
1899 L<[perl #126203]|https://rt.perl.org/Public/Bug/Display.html?id=126203>
1903 Fixed a bug in which F<Configure> could append "-quadmath" to the archname even
1904 if it was already present.
1905 L<[perl #128538]|https://rt.perl.org/Public/Bug/Display.html?id=128538>
1909 Clang builds with "-DPERL_GLOBAL_STRUCT" or "-DPERL_GLOBAL_STRUCT_PRIVATE" have
1910 been fixed (by disabling Thread Safety Analysis for these configurations).
1914 F<make_ext.pl> no longer updates a module's F<pm_to_blib> file when no
1915 files require updates. This could cause dependencies, F<perlmain.c>
1916 in particular, to be rebuilt unnecessarily. [perl #126710]
1920 The output of C<perl -V> has been reformatted so that each configuration
1921 and compile-time option is now listed one per line, to improve
1926 C<Configure> now builds C<miniperl> and C<generate_uudmap> if you
1927 invoke it with C<-Dusecrosscompiler> but not C<-Dtargethost=somehost>.
1928 This means you can supply your target platform C<config.sh>, generate
1929 the headers and proceed to build your cross-target perl. [perl #127234]
1933 Builds with C<-Accflags=-DPERL_TRACE_OPS> now only dump the operator
1934 counts when the environment variable C<PERL_TRACE_OPS> to be set to a
1935 non-zero integer. This allows C<make test> to pass on such a build.
1939 When building with GCC 6 and link-time optimization (the C<-flto> option to
1940 C<gcc>), C<Configure> was treating all probed symbols as present on the
1941 system, regardless of whether they actually exist. This has been fixed.
1946 The F<t/test.pl> library is used for internal testing of Perl itself, and
1947 also copied by several CPAN modules. Some of those modules must work on
1948 older versions of Perl, so F<t/test.pl> must in turn avoid newer Perl
1949 features. Compatibility with Perl 5.8 was inadvertently removed some time
1950 ago; it has now been restored. [perl #128052]
1954 The build process no longer emits an extra blank line before building each
1955 "simple" extension (those with only F<*.pm> and F<*.pod> files).
1965 F<XS-APItest/t/utf8.t>: Several small fixes and enhancements.
1969 Tests for locales were erroneously using locales incompatible with Perl.
1973 Some parts of the test suite that try to exhaustively test edge cases in the
1974 regex implementation have been restricted to running for a maximum of five
1975 minutes. On slow systems they could otherwise take several hours, without
1976 significantly improving our understanding of the correctness of the code
1979 In addition, some of those test cases have been split into more files, to
1980 allow them to be run in parallel on suitable systems.
1984 A new internal facility allows analysing the time taken by the individual
1985 tests in Perl's own test suite; see F<Porting/harness-timer-report.pl>.
1989 F<t/re/regexp_nonull.t> has been added to test that the regular expression
1990 engine can handle scalars that do not have a null byte just past the end of
1995 A new test script, F<t/op/decl-refs.t>, has been added to test the new feature,
1996 "Declaring a reference to a variable".
2000 A new test script, F<t/re/anyof.t>, has been added to test that the ANYOF nodes
2001 generated by bracketed character classes are as expected.
2005 F<t/harness> now tries really hard not to run tests outside of the Perl
2006 source tree. [perl #124050]
2010 =head1 Platform Support
2012 =head2 New Platforms
2018 Perl now compiles under NetBSD on VAX machines. However, it's not
2019 possible for that platform to implement floating-point infinities and
2020 NaNs compatibly with most modern systems, which implement the IEEE-754
2021 floating point standard. The hexadecimal floating point (C<0x...p[+-]n>
2022 literals, C<printf %a>) is not implemented, either.
2023 The C<make test> passes 98% of tests.
2029 Test fixes and minor updates.
2033 Account for lack of C<inf>, C<nan>, and C<-0.0> support.
2039 =head2 Platform-Specific Notes
2045 don't treat -Dprefix=/usr as special, instead require an extra option
2046 -Ddarwin_distribution to produce the same results.
2050 Finish removing POSIX deprecated functions.
2054 OS X El Capitan doesn't implement the clock_gettime() or clock_getres() APIs;
2055 emulate them as necessary.
2059 Deprecated syscall(2) on macOS 10.12.
2063 Several tests have been updated to work (or be skipped) on EBCDIC platforms.
2067 L<Net::Ping> UDP test is skipped on HP-UX.
2071 The hints for Hurd have been improved enabling malloc wrap and reporting the
2072 GNU libc used (previously it was an empty string when reported).
2076 VAX floating point formats are now supported on NetBSD.
2084 The path separator for the C<PERL5LIB> and C<PERLLIB> environment entries is
2085 now a colon (C<:>) when running under a Unix shell. There is no change when
2086 running under DCL (it's still C<|>).
2090 C<configure.com> now recognizes the VSI-branded C compiler and no longer
2091 recognizes the "DEC"-branded C compiler (as there hasn't been such a thing for
2102 Support for compiling perl on Windows using Microsoft Visual Studio 2015
2103 (containing Visual C++ 14.0) has been added.
2105 This version of VC++ includes a completely rewritten C run-time library, some
2106 of the changes in which mean that work done to resolve a socket close() bug in
2107 perl #120091 and perl #118059 is not workable in its current state with this
2108 version of VC++. Therefore, we have effectively reverted that bug fix for
2109 VS2015 onwards on the basis that being able to build with VS2015 onwards is
2110 more important than keeping the bug fix. We may revisit this in the future to
2111 attempt to fix the bug again in a way that is compatible with VS2015.
2113 These changes do not affect compilation with GCC or with Visual Studio versions
2114 up to and including VS2013, I<i.e.>, the bug fix is retained (unchanged) for those
2117 Note that you may experience compatibility problems if you mix a perl built
2118 with GCC or VS E<lt>= VS2013 with XS modules built with VS2015, or if you mix a
2119 perl built with VS2015 with XS modules built with GCC or VS E<lt>= VS2013.
2120 Some incompatibility may arise because of the bug fix that has been reverted
2121 for VS2015 builds of perl, but there may well be incompatibility anyway because
2122 of the rewritten CRT in VS2015 (I<e.g.>, see discussion at
2123 http://stackoverflow.com/questions/30412951).
2133 Tweaks for Win32 VC vs GCC detection makefile code. This fixes issue that CCHOME
2134 depends on CCTYPE, which in auto detect mode is set after CCHOME, so CCHOME uses
2135 the uninit CCTYPE var. Also fix else vs .ELSE in makefile.mk
2139 fp definitions have been updated.
2145 Drop support for Linux a.out Linux has used ELF for over twenty years.
2149 OpenBSD 6 still does not support returning pid, gid or uid with SA_SIGINFO.
2150 Make sure this is accounted for.
2154 t/uni/overload.t: Skip hanging test on FreeBSD.
2158 =head1 Internal Changes
2164 The C<op_class()> API function has been added. This is like the existing
2165 C<OP_CLASS()> macro, but can more accurately determine what struct an op
2166 has been allocated as. For example C<OP_CLASS()> might return
2167 C<OA_BASEOP_OR_UNOP> indicating that ops of this type are usually
2168 allocated as an C<OP> or C<UNOP>; while C<op_class()> will return
2169 C<OPclass_BASEOP> or C<OPclass_UNOP> as appropriate.
2173 The output format of the C<op_dump()> function (as used by C<perl -Dx>)
2174 has changed: it now displays an "ASCII-art" tree structure, and shows more
2175 low-level details about each op, such as its address and class.
2179 New versions of macros like C<isALPHA_utf8> and C<toLOWER_utf8> have
2180 been added, each with the
2181 suffix C<_safe>, like C<isSPACE_utf8_safe>. These take an extra
2182 parameter, giving an upper limit of how far into the string it is safe
2183 to read. Using the old versions could cause attempts to read beyond the
2184 end of the input buffer if the UTF-8 is not well-formed, and their use
2185 now raises a deprecation warning. Details are at
2186 L<perlapi/Character classification>.
2190 Calling macros like C<isALPHA_utf8> on malformed UTF-8 have issued a
2191 deprecation warning since Perl 5.18. They now die.
2192 Similarly, macros like C<toLOWER_utf8> on malformed UTF-8 now die.
2196 Calling the functions C<utf8n_to_uvchr> and its derivatives, while
2197 passing a string length of 0 is now asserted against in DEBUGGING
2198 builds, and otherwise returns the Unicode REPLACEMENT CHARACTER. If
2199 you have nothing to decode, you shouldn't call the decode function.
2203 The functions C<utf8n_to_uvchr> and its derivatives now return the
2204 Unicode REPLACEMENT CHARACTER if called with UTF-8 that has the overlong
2205 malformation, and that malformation is allowed by the input parameters.
2206 This malformation is where the UTF-8 looks valid syntactically, but
2207 there is a shorter sequence that yields the same code point. This has
2208 been forbidden since Unicode version 3.1.
2212 The functions C<utf8n_to_uvchr> and its derivatives now accept an input
2213 flag to allow the overflow malformation. This malformation is when the
2214 UTF-8 may be syntactically valid, but the code point it represents is
2215 not capable of being represented in the word length on the platform.
2216 What "allowed" means in this case is that the function doesn't return an
2217 error, and advances the parse pointer to beyond the UTF-8 in question,
2218 but it returns the Unicode REPLACEMENT CHARACTER as the value of the
2219 code point (since the real value is not representable).
2223 The C<PADOFFSET> type has changed from being unsigned to signed, and
2224 several pad-related variables such as C<PL_padix> have changed from being
2225 of type C<I32> to type C<PADOFFSET>.
2229 The function C<L<perlapi/utf8n_to_uvchr>> has been changed to not
2230 abandon searching for other malformations when the first one is
2231 encountered. A call to it thus can generate multiple diagnostics,
2232 instead of just one.
2236 A new function, C<L<perlapi/utf8n_to_uvchr_error>>, has been added for
2237 use by modules that need to know the details of UTF-8 malformations
2238 beyond pass/fail. Previously, the only ways to know why a sequence was
2239 ill-formed was to capture and parse the generated diagnostics, or to do
2244 Several new functions for handling Unicode have been added to the API:
2245 C<L<perlapi/is_strict_utf8_string>>,
2246 C<L<perlapi/is_c9strict_utf8_string>>,
2247 C<L<perlapi/is_utf8_string_flags>>,
2248 C<L<perlapi/is_strict_utf8_string_loc>>,
2249 C<L<perlapi/is_strict_utf8_string_loclen>>,
2250 C<L<perlapi/is_c9strict_utf8_string_loc>>,
2251 C<L<perlapi/is_c9strict_utf8_string_loclen>>,
2252 C<L<perlapi/is_utf8_string_loc_flags>>,
2253 C<L<perlapi/is_utf8_string_loclen_flags>>,
2254 C<L<perlapi/is_utf8_fixed_width_buf_flags>>,
2255 C<L<perlapi/is_utf8_fixed_width_buf_loc_flags>>,
2256 C<L<perlapi/is_utf8_fixed_width_buf_loclen_flags>>.
2258 These functions are all extensions of the C<is_utf8_string_*()> functions,
2259 that apply various restrictions to the UTF-8 recognized as valid.
2263 A new API function C<sv_setvpv_bufsize()> allows simultaneously setting the
2264 length and allocated size of the buffer in an C<SV>, growing the buffer if
2269 A new API macro C<SvPVCLEAR()> sets its C<SV> argument to an empty string,
2270 like Perl-space C<$x = ''>, but with several optimisations.
2274 All parts of the internals now agree that the C<sassign> op is a C<BINOP>;
2275 previously it was listed as a C<BASEOP> in F<regen/opcodes>, which meant
2276 that several parts of the internals had to be special-cased to accommodate
2277 it. This oddity's original motivation was to handle code like C<$x ||= 1>;
2278 that is now handled in a simpler way.
2282 Several new internal C macros have been added that take a string literal as
2283 arguments, alongside existing routines that take the equivalent value as two
2284 arguments, a character pointer and a length. The advantage of this is that
2285 the length of the string is calculated automatically, rather than having to
2286 be done manually. These routines are now used where appropriate across the
2291 The code in F<gv.c> that determines whether a variable has a special meaning
2292 to Perl has been simplified.
2296 The C<DEBUGGING>-mode output for regex compilation and execution has been
2301 Several macros and functions have been added to the public API for
2302 dealing with Unicode and UTF-8-encoded strings. See
2303 L<perlapi/Unicode Support>.
2307 Use C<my_strlcat()> in C<locale.c>. While C<strcat()> is safe in this context,
2308 some compilers were optimizing this to C<strcpy()> causing a porting test to
2309 fail that looks for unsafe code. Rather than fighting this, we just use
2310 C<my_strlcat()> instead.
2314 Three new ops, C<OP_ARGELEM>, C<OP_ARGDEFELEM> and C<OP_ARGCHECK> have
2315 been added. These are intended principally to implement the individual
2316 elements of a subroutine signature, plus any overall checking required.
2320 Perl no longer panics when switching into some locales on machines with
2321 buggy C<strxfrm()> implementations in their libc. [perl #121734]
2325 Perl is now built with the C<PERL_OP_PARENT> compiler define enabled by
2326 default. To disable it, use the C<PERL_NO_OP_PARENT> compiler define.
2327 This flag alters how the C<op_sibling> field is used in C<OP> structures,
2328 and has been available optionally since perl 5.22.
2330 See L<perl5220delta/"Internal Changes"> for more details of what this
2335 The meanings of some internal SV flags have been changed
2337 OPpRUNTIME, SVpbm_VALID, SVpbm_TAIL, SvTAIL_on, SvTAIL_off, SVrepl_EVAL,
2342 Change C<hv_fetch(…, "…", …, …)> to C<hv_fetchs(…, "…", …)>
2344 The dual-life dists all use Devel::PPPort, so they can use this function even
2345 though it was only added in 5.10.
2349 =head1 Selected Bug Fixes
2355 C< $-{$name} > would leak an C<AV> on each access if the regular
2356 expression had no named captures. The same applies to access to any
2357 hash tied with L<Tie::Hash::NamedCapture> and C<< all =E<gt> 1 >>. [perl
2362 Attempting to use the deprecated variable C<$#> as the object in an
2363 indirect object method call could cause a heap use after free or
2364 buffer overflow. [perl #129274]
2368 When checking for an indirect object method call in some rare cases
2369 the parser could reallocate the line buffer but then continue to use
2370 pointers to the old buffer. [perl #129190]
2374 Supplying a glob as the format argument to L<perlfunc/formline> would
2375 cause an assertion failure. [perl #130722]
2379 Code like C< $value1 =~ qr/.../ ~~ $value2 > would have the match
2380 converted into a qr// operator, leaving extra elements on the stack to
2381 confuse any surrounding expression. [perl #130705]
2385 Since 5.24 in some obscure cases, a regex which included code blocks
2386 from multiple sources (I<e.g.>, via embedded via qr// objects) could end up
2387 with the wrong current pad and crash or give weird results. [perl #129881]
2391 Occasionally C<local()>s in a code block within a patterns weren't being
2392 undone when the pattern matching backtracked over the code block.
2397 Using C<substr()> to modify a magic variable could access freed memory
2398 in some cases. [perl #129340]
2402 Under C<use utf8>, the entire Perl program is now checked that the UTF-8
2403 is wellformed. This resolves [perl #126310].
2407 The range operator C<..> on strings now handles its arguments correctly when in
2408 the scope of the L<< C<unicode_strings>|feature/"The 'unicode_strings' feature" >>
2409 feature. The previous behaviour was sufficiently unexpected that we believe no
2410 correct program could have made use of it.
2414 The S<split> operator did not ensure enough space was allocated for
2415 its return value in scalar context. It could then write a single
2416 pointer immediately beyond the end of the memory block allocated for
2417 the stack. [perl #130262]
2421 Using a large code point with the C<W> pack template character with
2422 the current output position aligned at just the right point could
2423 cause a write a single zero byte immediately beyond the end of an
2424 allocated buffer. [perl #129149]
2428 Supplying the form picture argument as part of the form argument list
2429 where the picture specifies modifying the argument could cause an
2430 access to the new freed compiled form. [perl #129125]
2434 Fix a problem with sort's build-in compare, where it would not sort
2435 correctly with 64-bit integers, and non-long doubles. [perl #130335]
2439 Fix issues with /(?{ ... E<lt>E<lt>EOF })/ that broke Method-Signatures. [perl #130398]
2443 Fix a macro which caused syntax error on an EBCDIC build.
2447 Prevent tests from getting hung up on 'NonStop' option. [perl #130445]
2451 Fixed an assertion failure with C<chop> and C<chomp>, which
2452 could be triggered by C<chop(@x =~ tr/1/1/)>. [perl #130198].
2456 Fixed a comment skipping error under C</x>; it could stop skipping a
2457 byte early, which could be in the middle of a UTF-8 character.
2462 F<perldb> now ignores F</dev/tty> on non-Unix systems. [perl #113960];
2466 Fix assertion failure for C<{}-E<gt>$x> when C<$x> isn't defined. [perl #130496].
2470 DragonFly BSD now has support for setproctitle(). [perl #130068].
2474 Fix an assertion error which could be triggered when lookahead string
2475 in patterns exceeded a minimum length. [perl #130522].
2479 Only warn once per literal about a misplaced C<_>. [perl #70878].
2483 Ensure range-start is set after error in C<tr///>. [perl #129342].
2487 Don't read past start of string for unmatched backref; otherwise,
2488 we may have heap buffer overflow. [perl #129377].
2492 Properly recognize mathematical digit ranges starting at U+1D7E.
2493 C<use re 'strict'> is supposed to warn if you use a range whose start
2494 and end digit aren't from the same group of 10. It didn't do that
2495 for five groups of mathematical digits starting at U+1D7E.
2499 A sub containing a "forward" declaration with the same name (I<e.g.>,
2500 C<sub c { sub c; }>) could sometimes crash or loop infinitely. [perl
2505 A crash in executing a regex with a floating UTF-8 substring against a
2506 target string that also used UTF-8 has been fixed. [perl #129350]
2510 Previously, a shebang line like C<#!perl -i u> could be erroneously
2511 interpreted as requesting the C<-u> option. This has been fixed. [perl
2516 The regex engine was previously producing incorrect results in some rare
2517 situations when backtracking past a trie that matches only one thing; this
2518 showed up as capture buffers (C<$1>, C<$2>, etc) erroneously containing data
2519 from regex execution paths that weren't actually executed for the final
2520 match. [perl #129897]
2524 Certain regexes making use of the experimental C<regex_sets> feature could
2525 trigger an assertion failure. This has been fixed. [perl #129322]
2529 Invalid assignments to a reference constructor (I<e.g.>, C<\eval=time>) could
2530 sometimes crash in addition to giving a syntax error. [perl #125679]
2534 The parser could sometimes crash if a bareword came after C<evalbytes>.
2539 Autoloading via a method call would warn erroneously ("Use of inherited
2540 AUTOLOAD for non-method") if there was a stub present in the package into
2541 which the invocant had been blessed. The warning is no longer emitted in
2542 such circumstances. [perl #47047]
2546 A sub containing with a "forward" declaration with the same name
2548 C<sub c { sub c; }>) could sometimes crash or loop infinitely. [perl
2553 The use of C<splice> on arrays with nonexistent elements could cause other
2554 operators to crash. [perl #129164]
2558 Fixed case where C<re_untuit_start> will overshoot the length of a utf8
2559 string. [perl #129012]
2563 Handle C<CXt_SUBST> better in C<Perl_deb_stack_all>, previously it wasn't
2564 checking that the I<current> C<cx> is the right type, and instead was always
2565 checking the base C<cx> (effectively a noop). [perl #129029]
2569 Fixed two possible use-after-free bugs in C<Perl_yylex>. C<Perl_yylex>
2570 maintains up to two pointers into the parser buffer, one of which can
2571 become stale under the right conditions. [perl #129069]
2575 Fixed a crash with C<s///l> where it thought it was dealing with UTF-8
2576 when it wasn't. [perl #129038]
2580 Fixed place where regex was not setting the syntax error correctly.
2585 The C<&.> operator (and the C<&> operator, when it treats its arguments as
2586 strings) were failing to append a trailing null byte if at least one string
2587 was marked as utf8 internally. Many code paths (system calls, regexp
2588 compilation) still expect there to be a null byte in the string buffer
2589 just past the end of the logical string. An assertion failure was the
2590 result. [perl #129287]
2594 Check C<pack_sockaddr_un()>'s return value because C<pack_sockaddr_un()>
2595 silently truncates the supplied path if it won't fit into the C<sun_path>
2596 member of C<sockaddr_un>. This may change in the future, but for now
2597 check the path in theC<sockaddr> matches the desired path, and skip if
2598 it doesn't. [perl #128095]
2602 Make sure C<PL_oldoldbufptr> is preserved in C<scan_heredoc()>. In some
2603 cases this is used in building error messages. [perl #128988]
2607 Check for null PL_curcop in IN_LC() [perl #129106]
2611 Fixed the parser error handling for an 'C<:attr(foo>' that does not have
2616 Fix C<Perl_delimcpy()> to handle a backslash as last char, this
2617 actually fixed two bugs, [perl #129064] and [perl #129176].
2621 [perl #129267] rework gv_fetchmethod_pvn_flags separator parsing to
2622 prevent possible string overrun with invalid len in gv.c
2626 Problems with in-place array sorts: code like C<@a = sort { ... } @a>,
2627 where the source and destination of the sort are the same plain array, are
2628 optimised to do less copying around. Two side-effects of this optimisation
2629 were that the contents of C<@a> as visible to to sort routine were
2630 partially sorted, and under some circumstances accessing C<@a> during the
2631 sort could crash the interpreter. Both these issues have been fixed, and
2632 Sort functions see the original value of C<@a>.
2636 Non-ASCII string delimiters are now reported correctly in error messages
2637 for unterminated strings. [perl #128701]
2641 C<pack("p", ...)> used to emit its warning ("Attempt to pack pointer to
2642 temporary value") erroneously in some cases, but has been fixed.
2646 C<@DB::args> is now exempt from "used once" warnings. The warnings only
2647 occurred under B<-w>, because F<warnings.pm> itself uses C<@DB::args>
2652 The use of built-in arrays or hash slices in a double-quoted string no
2653 longer issues a warning ("Possible unintended interpolation...") if the
2654 variable has not been mentioned before. This affected code like
2655 C<qq|@DB::args|> and C<qq|@SIG{'CHLD', 'HUP'}|>. (The special variables
2656 C<@-> and C<@+> were already exempt from the warning.)
2660 C<gethostent> and similar functions now perform a null check internally, to
2661 avoid crashing with torsocks. This was a regression from 5.22. [perl
2666 C<defined *{'!'}>, C<defined *{'['}>, and C<defined *{'-'}> no longer leak
2667 memory if the typeglob in question has never been accessed before.
2671 Mentioning the same constant twice in a row (which is a syntax error) no
2672 longer fails an assertion under debugging builds. This was a regression
2673 from 5.20. [perl #126482]
2677 Many issues relating to C<printf "%a"> of hexadecimal floating point
2678 were fixed. In addition, the "subnormals" (formerly known as "denormals")
2679 floating point anumbers are now supported both with the plain IEEE 754
2680 floating point numbers (64-bit or 128-bit) and the x86 80-bit
2681 "extended precision". Note that subnormal hexadecimal floating
2682 point literals will give a warning about "exponent underflow".
2683 [perl #128843, #128889, #128890, #128893, #128909, #128919]
2687 A regression in 5.24 with C<tr/\N{U+...}/foo/> when the code point was between
2688 128 and 255 has been fixed. [perl #128734].
2692 A regression from the previous development release, 5.23.3, where
2693 compiling a regular expression could crash the interpreter has been
2694 fixed. [perl #128686].
2698 Use of a string delimiter whose code point is above 2**31 now works
2699 correctly on platforms that allow this. Previously, certain characters,
2700 due to truncation, would be confused with other delimiter characters
2701 with special meaning (such as C<?> in C<m?...?>), resulting
2702 in inconsistent behaviour. Note that this is non-portable,
2703 and is based on Perl's extension to UTF-8, and is probably not
2704 displayable nor enterable by any editor. [perl #128738]
2708 C<@{x> followed by a newline where C<x> represents a control or non-ASCII
2709 character no longer produces a garbled syntax error message or a crash.
2714 An assertion failure with C<%: = 0> has been fixed.
2715 L<[perl #128238]|https://rt.perl.org/Public/Bug/Display.html?id=128238>
2719 In Perl 5.18, the parsing of C<"$foo::$bar"> was accidentally changed, such
2720 that it would be treated as C<$foo."::".$bar>. The previous behavior, which
2721 was to parse it as C<$foo:: . $bar>, has been restored.
2722 L<[perl #128478]|https://rt.perl.org/Public/Bug/Display.html?id=128478>
2726 Since Perl 5.20, line numbers have been off by one when perl is invoked with
2727 the B<-x> switch. This has been fixed.
2728 L<[perl #128508]|https://rt.perl.org/Public/Bug/Display.html?id=128508>
2732 Vivifying a subroutine stub in a deleted stash (I<e.g.>, C<delete $My::{"Foo::"};
2733 \&My::Foo::foo>) no longer crashes. It had begun crashing in Perl 5.18.
2734 L<[perl #128532]|https://rt.perl.org/Public/Bug/Display.html?id=128532>
2738 Some obscure cases of subroutines and file handles being freed at the same time
2739 could result in crashes, but have been fixed. The crash was introduced in Perl
2741 L<[perl #128597]|https://rt.perl.org/Public/Bug/Display.html?id=128597>
2745 Code that looks for a variable name associated with an uninitialized value
2746 could cause an assertion in cases where magic is involved, such as
2747 C<$ISA[0][0]>. This has now been fixed.
2748 L<[perl #128253]|https://rt.perl.org/Public/Bug/Display.html?id=128253>
2752 A crash caused by code generating the warning "Subroutine STASH::NAME
2753 redefined" in cases such as C<sub P::f{} undef *P::; *P::f =sub{};> has been
2754 fixed. In these cases, where the STASH is missing, the warning will now appear
2755 as "Subroutine NAME redefined".
2756 L<[perl #128257]|https://rt.perl.org/Public/Bug/Display.html?id=128257>
2760 Fixed an assertion triggered by some code that handles deprecated behavior in
2761 formats, I<e.g.>, in cases like this:
2767 L<[perl #128255]|https://rt.perl.org/Public/Bug/Display.html?id=128255>
2771 A possible divide by zero in string transformation code on Windows has been
2772 avoided, fixing a crash when collating an empty string.
2773 L<[perl #128618]|https://rt.perl.org/Public/Bug/Display.html?id=128618>
2777 Some regular expression parsing glitches could lead to assertion failures with
2778 regular expressions such as C</(?E<lt>=/> and C</(?E<lt>!/>. This has now been fixed.
2779 L<[perl #128170]|https://rt.perl.org/Public/Bug/Display.html?id=128170>
2783 C< until ($x = 1) { ... } > and C< ... until $x = 1 > now properly
2784 warn when syntax warnings are enabled. [perl #127333]
2788 socket() now leaves the error code returned by the system in C<$!> on
2789 failure. [perl #128316]
2793 Assignment variants of any bitwise ops under the C<bitwise> feature would
2794 crash if the left-hand side was an array or hash. [perl #128204]
2798 C<require> followed by a single colon (as in C<foo() ? require : ...> is
2799 now parsed correctly as C<require> with implicit $_, rather than
2800 C<require "">. [perl #128307]
2804 Scalar C<keys %hash> can now be assigned to consistently in all scalar
2805 lvalue contexts. Previously it worked for some contexts but not others.
2809 List assignment to C<vec> or C<substr> with an array or hash for its first
2810 argument used to result in crashes or "Can't coerce" error messages at run
2811 time, unlike scalar assignment, which would give an error at compile time.
2812 List assignment now gives a compile-time error, too. [perl #128260]
2816 Expressions containing an C<&&> or C<||> operator (or their synonyms C<and>
2817 and C<or>) were being compiled incorrectly in some cases. If the left-hand
2818 side consisted of either a negated bareword constant or a negated C<do {}>
2819 block containing a constant expression, and the right-hand side consisted of
2820 a negated non-foldable expression, one of the negations was effectively
2821 ignored. The same was true of C<if> and C<unless> statement modifiers,
2822 though with the left-hand and right-hand sides swapped. This long-standing
2823 bug has now been fixed. [perl #127952]
2827 C<reset> with an argument no longer crashes when encountering stash entries
2828 other than globs. [perl #128106]
2832 Assignment of hashes to, and deletion of, typeglobs named C<*::::::> no
2833 longer causes crashes. [perl #128086]
2837 Handle SvIMMORTALs in LHS of list assign. [perl #129991]
2841 [perl #130010] a5540cf breaks texinfo
2843 This involved user-defined Unicode properties.
2847 Fix error message for unclosed C<\N{> in regcomp.
2849 An unclosed C<\N{> could give the wrong error message
2850 C<"\N{NAME} must be resolved by the lexer">.
2854 List assignment in list context where the LHS contained aggregates and
2855 where there were not enough RHS elements, used to skip scalar lvalues.
2856 Previously, C<(($a,$b,@c,$d) = (1))> in list context returned C<($a)>; now
2857 it returns C<($a,$b,$d)>. C<(($a,$b,$c) = (1))> is unchanged: it still
2858 returns C<($a,$b,$c)>. This can be seen in the following:
2860 sub inc { $_++ for @_ }
2861 inc(($a,$b,@c,$d) = (10))
2863 Formerly, the values of C<($a,$b,$d)> would be left as C<(11,undef,undef)>;
2864 now they are C<(11,1,1)>.
2870 The basic problem is that code like this: /(?{ s!!! })/ can trigger infinite
2871 recursion on the C stack (not the normal perl stack) when the last successful
2872 pattern in scope is itself. Since the C stack overflows this manifests as an
2873 untrappable error/segfault, which then kills perl.
2875 We avoid the segfault by simply forbidding the use of the empty pattern when it
2876 would resolve to the currently executing pattern.
2880 [perl 128997] Avoid reading beyond the end of the line buffer when there's a
2881 short UTF-8 character at the end.
2885 [perl 129950] fix firstchar bitmap under utf8 with prefix optimisation.
2889 [perl 129954] Carp/t/arg_string.t: be liberal in f/p formats.
2893 [perl 129928] make do "a\0b" fail silently instead of throwing.
2897 [perl 129130] make chdir allocate the stack it needs.
2901 =head1 Known Problems
2907 Some modules have been broken by the L<context stack rework|/Internal Changes>.
2908 These modules were relying on non-guaranteed implementation details in perl.
2909 Their maintainers have been informed, and should contact perl5-porters for
2910 advice if needed. Below is a subset of these modules:
2914 =item * L<Algorithm::Permute>
2918 L<Coro> and perl 5.22 were already incompatible due to a change in the perl,
2919 and the reworking on the perl context stack creates a further incompatibility.
2920 perl5-porters has L<discussed the issue on the mailing
2921 list|http://www.nntp.perl.org/group/perl.perl5.porters/2016/05/msg236174.html>.
2923 =item * L<Data::Alias>
2927 =item * L<Scope::Upper>
2935 The module L<lexical::underscore> no longer works on Perl 5.24, because perl
2936 no longer has a lexical C<$_>!
2940 C<mod_perl> has been patched for compatibility for v5.22 and later but no
2941 release has been made. The relevant patch (and other changes) can be found in
2942 their source code repository, L<mirrored at
2943 GitHub|https://github.com/apache/mod_perl/commit/82827132efd3c2e25cc413c85af61bb63375da6e>.
2947 =head1 Errata From Previous Releases
2953 Parsing bad POSIX charclasses no longer leaks memory.
2954 L<[perl #128313]|https://rt.perl.org/Public/Bug/Display.html?id=128313>
2958 Fixed issues with recursive regexes. The behavior was fixed in Perl 5.24.
2959 L<[perl #126182]|https://rt.perl.org/Public/Bug/Display.html?id=126182>
2965 Jon Portnoy (AVENJ), a prolific Perl author and admired Gentoo community
2966 member, has passed away on August 10, 2016. He will be remembered and
2967 missed by all those who he came in contact with, and enriched with his
2968 intellect, wit, and spirit.
2970 It is with great sadness that we also note Kip Hampton's passing. Probably
2971 best known as the author of the Perl & XML column on XML.com, he was a
2972 core contributor to AxKit, an XML server platform that became an Apache
2973 Foundation project. He was a frequent speaker in the early days at
2974 OSCON, and most recently at YAPC::NA in Madison. He was frequently on
2975 irc.perl.org as `ubu`, generally in the #axkit-dahut community, the
2976 group responsible for YAPC::NA Asheville in 2011.
2978 Kip and his constant contributions to the community will be greatly
2981 =head1 Acknowledgements
2983 Perl 5.26.0 represents approximately 12 months of development since Perl 5.24.0
2984 and contains approximately 370,000 lines of changes across 2,600 files from 86
2987 Excluding auto-generated files, documentation and release tools, there were
2988 approximately 230,000 lines of changes to 1,800 .pm, .t, .c and .h files.
2990 Perl continues to flourish into its third decade thanks to a vibrant community
2991 of users and developers. The following people are known to have contributed the
2992 improvements that became Perl 5.26.0:
2994 Aaron Crane, Abigail, Ævar Arnfjörð Bjarmason, Alex Vandiver, Andreas
2995 König, Andreas Voegele, Andrew Fresh, Andy Lester, Aristotle Pagaltzis, Chad
2996 Granum, Chase Whitener, Chris 'BinGOs' Williams, Chris Lamb, Christian Hansen,
2997 Christian Millour, Colin Newell, Craig A. Berry, Dagfinn Ilmari Mannsåker, Dan
2998 Collins, Daniel Dragan, Dave Cross, Dave Rolsky, David Golden, David H.
2999 Gutteridge, David Mitchell, Dominic Hargreaves, Doug Bell, E. Choroba, Ed Avis,
3000 Father Chrysostomos, François Perrad, Hauke D, H.Merijn Brand, Hugo van der
3001 Sanden, Ivan Pozdeev, James E Keenan, James Raspass, Jarkko Hietaniemi, Jerry
3002 D. Hedden, Jim Cromie, J. Nick Koston, John Lightsey, Karen Etheridge, Karl
3003 Williamson, Leon Timmermans, Lukas Mai, Matthew Horsfall, Maxwell Carey, Misty
3004 De Meo, Neil Bowers, Nicholas Clark, Nicolas R., Niko Tyni, Pali, Paul
3005 Marquess, Peter Avalos, Petr Písař, Pino Toscano, Rafael Garcia-Suarez, Reini
3006 Urban, Renee Baecker, Ricardo Signes, Richard Levitte, Rick Delaney, Salvador
3007 Fandiño, Samuel Thibault, Sawyer X, Sébastien Aperghis-Tramoni, Sergey
3008 Aleynikov, Shlomi Fish, Smylers, Stefan Seifert, Steffen Müller, Stevan
3009 Little, Steve Hay, Steven Humphrey, Sullivan Beck, Theo Buehler, Thomas Sibley,
3010 Todd Rinaldo, Tomasz Konojacki, Tony Cook, Unicode Consortium, Yaroslav Kuzmin,
3013 The list above is almost certainly incomplete as it is automatically generated
3014 from version control history. In particular, it does not include the names of
3015 the (very much appreciated) contributors who reported issues to the Perl bug
3018 Many of the changes included in this version originated in the CPAN modules
3019 included in Perl's core. We're grateful to the entire CPAN community for
3020 helping Perl to flourish.
3022 For a more complete list of all of Perl's historical contributors, please see
3023 the F<AUTHORS> file in the Perl source distribution.
3025 =head1 Reporting Bugs
3027 If you find what you think is a bug, you might check the articles recently
3028 posted to the comp.lang.perl.misc newsgroup and the perl bug database at
3029 L<https://rt.perl.org/> . There may also be information at
3030 L<http://www.perl.org/> , the Perl Home Page.
3032 If you believe you have an unreported bug, please run the L<perlbug> program
3033 included with your release. Be sure to trim your bug down to a tiny but
3034 sufficient test case. Your bug report, along with the output of C<perl -V>,
3035 will be sent off to C<perlbug@perl.org> to be analysed by the Perl porting team.
3037 If the bug you are reporting has security implications which make it
3038 inappropriate to send to a publicly archived mailing list, then see
3039 L<perlsec/SECURITY VULNERABILITY CONTACT INFORMATION>
3040 for details of how to report the issue.
3044 The F<Changes> file for an explanation of how to view exhaustive details on
3047 The F<INSTALL> file for how to build Perl.
3049 The F<README> file for general stuff.
3051 The F<Artistic> and F<Copying> files for copyright information.