This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
include some more data in new-perldelta output
[perl5.git] / pod / perldelta.pod
CommitLineData
44691e6f
AB
1=encoding utf8
2
3=head1 NAME
4
05c8f9ed 5perldelta - what is new for perl v5.16.0
2630d42b 6
05c8f9ed
RS
7=head1 DESCRIPTION
8
9This document describes differences between the 5.14.0 release and
10the 5.16.0 release.
11
12If you are upgrading from an earlier release such as 5.12.0, first read
13L<perl5140delta>, which describes differences between 5.12.0 and
145.14.0.
15
7af9a95b 16Some bug fixes in this release have been backported to later
c7957d51
FC
17releases of 5.14.x. Those are indicated with the 5.14.x version in
18parentheses.
19
05c8f9ed
RS
20=head1 Notice
21
8c551b2a
RS
22With the release of Perl 5.16.0, the 5.12.x series of releases is now out of
23its support period. There may be future 5.12.x releases, but only in the
7c54b938
RS
24event of a critical security issue. Users of Perl 5.12 or earlier should
25consider upgrading to a more recent release of Perl.
26
27This policy is described in greater detail in
28L<perlpolicy|perlpolicy/MAINTENANCE AND SUPPORT>.
05c8f9ed
RS
29
30=head1 Core Enhancements
31
32=head2 C<use I<VERSION>>
33
34As of this release, version declarations like C<use v5.16> now disable
35all features before enabling the new feature bundle. This means that
36the following holds true:
37
38 use 5.016;
39 # only 5.16 features enabled here
40 use 5.014;
41 # only 5.14 features enabled here (not 5.16)
42
43C<use v5.12> and higher continue to enable strict, but explicit C<use
44strict> and C<no strict> now override the version declaration, even
45when they come first:
46
47 no strict;
48 use 5.012;
49 # no strict here
50
51There is a new ":default" feature bundle that represents the set of
52features enabled before any version declaration or C<use feature> has
53been seen. Version declarations below 5.10 now enable the ":default"
1540de7c 54feature set. This does not actually change the behavior of C<use
05c8f9ed
RS
55v5.8>, because features added to the ":default" set are those that were
56traditionally enabled by default, before they could be turned off.
57
58C<< no feature >> now resets to the default feature set. To disable all
59features (which is likely to be a pretty special-purpose request, since
60it presumably won't match any named set of semantics) you can now
61write C<< no feature ':all' >>.
62
63C<$[> is now disabled under C<use v5.16>. It is part of the default
64feature set and can be turned on or off explicitly with C<use feature
65'array_base'>.
66
67=head2 C<__SUB__>
68
69The new C<__SUB__> token, available under the C<current_sub> feature
70(see L<feature>) or C<use v5.16>, returns a reference to the current
71subroutine, making it easier to write recursive closures.
72
73=head2 New and Improved Built-ins
74
75=head3 More consistent C<eval>
76
77The C<eval> operator sometimes treats a string argument as a sequence of
78characters and sometimes as a sequence of bytes, depending on the
79internal encoding. The internal encoding is not supposed to make any
80difference, but there is code that relies on this inconsistency.
81
82The new C<unicode_eval> and C<evalbytes> features (enabled under C<use
835.16.0>) resolve this. The C<unicode_eval> feature causes C<eval
84$string> to treat the string always as Unicode. The C<evalbytes>
85features provides a function, itself called C<evalbytes>, which
86evaluates its argument always as a string of bytes.
87
88These features also fix oddities with source filters leaking to outer
89dynamic scopes.
90
91See L<feature> for more detail.
92
93=head3 C<substr> lvalue revamp
94
7af9a95b 95=for comment Does this belong here, or under Incompatible Changes?
05c8f9ed
RS
96
97When C<substr> is called in lvalue or potential lvalue context with two
98or three arguments, a special lvalue scalar is returned that modifies
99the original string (the first argument) when assigned to.
100
101Previously, the offsets (the second and third arguments) passed to
102C<substr> would be converted immediately to match the string, negative
103offsets being translated to positive and offsets beyond the end of the
104string being truncated.
105
106Now, the offsets are recorded without modification in the special
107lvalue scalar that is returned, and the original string is not even
108looked at by C<substr> itself, but only when the returned lvalue is
109read or modified.
110
111These changes result in an incompatible change:
112
113If the original string changes length after the call to C<substr> but
114before assignment to its return value, negative offsets will remember
115their position from the end of the string, affecting code like this:
116
117 my $string = "string";
118 my $lvalue = \substr $string, -4, 2;
5bbddb31 119 print $$lvalue, "\n"; # prints "ri"
05c8f9ed 120 $string = "bailing twine";
5bbddb31 121 print $$lvalue, "\n"; # prints "wi"; used to print "il"
05c8f9ed
RS
122
123The same thing happens with an omitted third argument. The returned
124lvalue will always extend to the end of the string, even if the string
125becomes longer.
126
127Since this change also allowed many bugs to be fixed (see
1540de7c 128L</The C<substr> operator>), and since the behavior
014dc7eb 129of negative offsets has never been specified, the
05c8f9ed
RS
130change was deemed acceptable.
131
132=head3 Return value of C<tied>
133
134The value returned by C<tied> on a tied variable is now the actual
135scalar that holds the object to which the variable is tied. This
7af9a95b 136lets ties be weakened with C<Scalar::Util::weaken(tied
05c8f9ed
RS
137$tied_variable)>.
138
139=head2 Unicode Support
140
141=head3 Supports (I<almost>) Unicode 6.1
142
143Besides the addition of whole new scripts, and new characters in
144existing scripts, this new version of Unicode, as always, makes some
145changes to existing characters. One change that may trip up some
146applications is that the General Category of two characters in the
147Latin-1 range, PILCROW SIGN and SECTION SIGN, has been changed from
148Other_Symbol to Other_Punctuation. The same change has been made for
149a character in each of Tibetan, Ethiopic, and Aegean.
150The code points U+3248..U+324F (CIRCLED NUMBER TEN ON BLACK SQUARE
151through CIRCLED NUMBER EIGHTY ON BLACK SQUARE) have had their General
152Category changed from Other_Symbol to Other_Numeric. The Line Break
7af9a95b 153property has changes for Hebrew and Japanese; and because of
05c8f9ed
RS
154other changes in 6.1, the Perl regular expression construct C<\X> now
155works differently for some characters in Thai and Lao.
156
157New aliases (synonyms) have been defined for many property values;
158these, along with the previously existing ones, are all cross-indexed in
159L<perluniprops>.
160
161The return value of C<charnames::viacode()> is affected by other
162changes:
163
164 Code point Old Name New Name
165 U+000A LINE FEED (LF) LINE FEED
166 U+000C FORM FEED (FF) FORM FEED
167 U+000D CARRIAGE RETURN (CR) CARRIAGE RETURN
168 U+0085 NEXT LINE (NEL) NEXT LINE
169 U+008E SINGLE-SHIFT 2 SINGLE-SHIFT-2
170 U+008F SINGLE-SHIFT 3 SINGLE-SHIFT-3
171 U+0091 PRIVATE USE 1 PRIVATE USE-1
172 U+0092 PRIVATE USE 2 PRIVATE USE-2
173 U+2118 SCRIPT CAPITAL P WEIERSTRASS ELLIPTIC FUNCTION
174
175Perl will accept any of these names as input, but
176C<charnames::viacode()> now returns the new name of each pair. The
177change for U+2118 is considered by Unicode to be a correction, that is
178the original name was a mistake (but again, it will remain forever valid
179to use it to refer to U+2118). But most of these changes are the
180fallout of the mistake Unicode 6.0 made in naming a character used in
181Japanese cell phones to be "BELL", which conflicts with the longstanding
182industry use of (and Unicode's recommendation to use) that name
7af9a95b
TC
183to mean the ASCII control character at U+0007. Therefore, that name
184has been deprecated in Perl since v5.14, and any use of it will raise a
05c8f9ed 185warning message (unless turned off). The name "ALERT" is now the
7af9a95b 186preferred name for this code point, with "BEL" an acceptable short
05c8f9ed 187form. The name for the new cell phone character, at code point U+1F514,
7af9a95b
TC
188remains undefined in this version of Perl (hence we don't
189implement quite all of Unicode 6.1), but starting in v5.18, BELL will mean
05c8f9ed
RS
190this character, and not U+0007.
191
192Unicode has taken steps to make sure that this sort of mistake does not
7af9a95b 193happen again. The Standard now includes all generally accepted
05c8f9ed
RS
194names and abbreviations for control characters, whereas previously it
195didn't (though there were recommended names for most of them, which Perl
196used). This means that most of those recommended names are now
197officially in the Standard. Unicode did not recommend names for the
198four code points listed above between U+008E and U+008F, and in
199standardizing them Unicode subtly changed the names that Perl had
200previously given them, by replacing the final blank in each name by a
201hyphen. Unicode also officially accepts names that Perl had deprecated,
202such as FILE SEPARATOR. Now the only deprecated name is BELL.
203Finally, Perl now uses the new official names instead of the old
204(now considered obsolete) names for the first four code points in the
205list above (the ones which have the parentheses in them).
206
207Now that the names have been placed in the Unicode standard, these kinds
208of changes should not happen again, though corrections, such as to
209U+2118, are still possible.
210
211Unicode also added some name abbreviations, which Perl now accepts:
212SP for SPACE;
213TAB for CHARACTER TABULATION;
214NEW LINE, END OF LINE, NL, and EOL for LINE FEED;
215LOCKING-SHIFT ONE for SHIFT OUT;
216LOCKING-SHIFT ZERO for SHIFT IN;
217and ZWNBSP for ZERO WIDTH NO-BREAK SPACE.
218
219More details on this version of Unicode are provided in
220L<http://www.unicode.org/versions/Unicode6.1.0/>.
221
222=head3 C<use charnames> is no longer needed for C<\N{I<name>}>
223
224When C<\N{I<name>}> is encountered, the C<charnames> module is now
225automatically loaded when needed as if the C<:full> and C<:short>
226options had been specified. See L<charnames> for more information.
227
228=head3 C<\N{...}> can now have Unicode loose name matching
229
230This is described in the C<charnames> item in
231L</Updated Modules and Pragmata> below.
232
233=head3 Unicode Symbol Names
234
235Perl now has proper support for Unicode in symbol names. It used to be
236that C<*{$foo}> would ignore the internal UTF8 flag and use the bytes of
237the underlying representation to look up the symbol. That meant that
238C<*{"\x{100}"}> and C<*{"\xc4\x80"}> would return the same thing. All
239these parts of Perl have been fixed to account for Unicode:
240
241=over
242
243=item *
244
245Method names (including those passed to C<use overload>)
246
247=item *
248
7af9a95b 249Typeglob names (including names of variables, subroutines, and filehandles)
05c8f9ed
RS
250
251=item *
252
253Package names
254
255=item *
256
257C<goto>
258
259=item *
260
261Symbolic dereferencing
262
263=item *
264
265Second argument to C<bless()> and C<tie()>
266
267=item *
268
269Return value of C<ref()>
270
271=item *
272
273Subroutine prototypes
274
275=item *
276
277Attributes
278
279=item *
280
281Various warnings and error messages that mention variable names or values,
282methods, etc.
283
284=back
285
286In addition, a parsing bug has been fixed that prevented C<*{é}> from
287implicitly quoting the name, but instead interpreted it as C<*{+é}>, which
288would cause a strict violation.
289
290C<*{"*a::b"}> automatically strips off the * if it is followed by an ASCII
291letter. That has been extended to all Unicode identifier characters.
292
293One-character non-ASCII non-punctuation variables (like C<$é>) are now
294subject to "Used only once" warnings. They used to be exempt, as they
e019578d 295were treated as punctuation variables.
05c8f9ed
RS
296
297Also, single-character Unicode punctuation variables (like C<$‰>) are now
298supported [perl #69032].
299
300=head3 Improved ability to mix locales and Unicode, including UTF-8 locales
301
302An optional parameter has been added to C<use locale>
303
304 use locale ':not_characters';
305
306which tells Perl to use all but the C<LC_CTYPE> and C<LC_COLLATE>
307portions of the current locale. Instead, the character set is assumed
7af9a95b 308to be Unicode. This lets locales and Unicode be seamlessly mixed,
05c8f9ed
RS
309including the increasingly frequent UTF-8 locales. When using this
310hybrid form of locales, the C<:locale> layer to the L<open> pragma can
311be used to interface with the file system, and there are CPAN modules
312available for ARGV and environment variable conversions.
313
314Full details are in L<perllocale>.
315
316=head3 New function C<fc> and corresponding escape sequence C<\F> for Unicode foldcase
317
318Unicode foldcase is an extension to lowercase that gives better results
319when comparing two strings case-insensitively. It has long been used
320internally in regular expression C</i> matching. Now it is available
321explicitly through the new C<fc> function call (enabled by
322S<C<"use feature 'fc'">>, or C<use v5.16>, or explicitly callable via
323C<CORE::fc>) or through the new C<\F> sequence in double-quotish
324strings.
325
326Full details are in L<perlfunc/fc>.
327
328=head3 The Unicode C<Script_Extensions> property is now supported.
329
330New in Unicode 6.0, this is an improved C<Script> property. Details
331are in L<perlunicode/Scripts>.
332
333=head2 XS Changes
334
335=head3 Improved typemaps for Some Builtin Types
336
7af9a95b 337Most XS authors will know there is a longstanding bug in the
05c8f9ed
RS
338OUTPUT typemap for T_AVREF (C<AV*>), T_HVREF (C<HV*>), T_CVREF (C<CV*>),
339and T_SVREF (C<SVREF> or C<\$foo>) that requires manually decrementing
340the reference count of the return value instead of the typemap taking
341care of this. For backwards-compatibility, this cannot be changed in the
342default typemaps. But we now provide additional typemaps
343C<T_AVREF_REFCOUNT_FIXED>, etc. that do not exhibit this bug. Using
344them in your extension is as simple as having one line in your
345C<TYPEMAP> section:
346
347 HV* T_HVREF_REFCOUNT_FIXED
348
349=head3 C<is_utf8_char()>
350
351The XS-callable function C<is_utf8_char()>, when presented with
352malformed UTF-8 input, can read up to 12 bytes beyond the end of the
88132a12
KW
353string. This cannot be fixed without changing its API, and so its
354use is now deprecated. Use C<is_utf8_char_buf()> (described just below)
355instead.
05c8f9ed
RS
356
357=head3 Added C<is_utf8_char_buf()>
358
359This function is designed to replace the deprecated L</is_utf8_char()>
360function. It includes an extra parameter to make sure it doesn't read
361past the end of the input buffer.
362
363=head3 Other C<is_utf8_foo()> functions, as well as C<utf8_to_foo()>, etc.
364
7af9a95b
TC
365Most other XS-callable functions that take UTF-8 encoded input
366implicitly assume that the UTF-8 is valid (not malformed) with respect to
05c8f9ed
RS
367buffer length. Do not do things such as change a character's case or
368see if it is alphanumeric without first being sure that it is valid
369UTF-8. This can be safely done for a whole string by using one of the
370functions C<is_utf8_string()>, C<is_utf8_string_loc()>, and
371C<is_utf8_string_loclen()>.
372
373=head3 New Pad API
374
375Many new functions have been added to the API for manipulating lexical
376pads. See L<perlapi/Pad Data Structures> for more information.
377
378=head2 Changes to Special Variables
379
380=head3 C<$$> can be assigned to
381
382C<$$> was made read-only in Perl 5.8.0. But only sometimes: C<local $$>
383would make it writable again. Some CPAN modules were using C<local $$> or
384XS code to bypass the read-only check, so there is no reason to keep C<$$>
385read-only. (This change also allowed a bug to be fixed while maintaining
386backward compatibility.)
387
388=head3 C<$^X> converted to an absolute path on FreeBSD, OS X and Solaris
389
390C<$^X> is now converted to an absolute path on OS X, FreeBSD (without
391needing F</proc> mounted) and Solaris 10 and 11. This augments the
7af9a95b 392previous approach of using F</proc> on Linux, FreeBSD, and NetBSD
05c8f9ed
RS
393(in all cases, where mounted).
394
395This makes relocatable perl installations more useful on these platforms.
396(See "Relocatable @INC" in F<INSTALL>)
397
398=head2 Debugger Changes
399
400=head3 Features inside the debugger
401
402The current Perl's L<feature> bundle is now enabled for commands entered
403in the interactive debugger.
404
405=head3 New option for the debugger's B<t> command
406
407The B<t> command in the debugger, which toggles tracing mode, now
408accepts a numeric argument that determines how many levels of subroutine
409calls to trace.
410
411=head3 C<enable> and C<disable>
412
413The debugger now has C<disable> and C<enable> commands for disabling
414existing breakpoints and re-enabling them. See L<perldebug>.
415
416=head3 Breakpoints with file names
417
7af9a95b
TC
418The debugger's "b" command for setting breakpoints now lets a line
419number be prefixed with a file name. See
05c8f9ed
RS
420L<perldebug/"b [file]:[line] [condition]">.
421
422=head2 The C<CORE> Namespace
423
424=head3 The C<CORE::> prefix
425
426The C<CORE::> prefix can now be used on keywords enabled by
427L<feature.pm|feature>, even outside the scope of C<use feature>.
428
429=head3 Subroutines in the C<CORE> namespace
430
431Many Perl keywords are now available as subroutines in the CORE namespace.
7af9a95b 432This lets them be aliased:
05c8f9ed
RS
433
434 BEGIN { *entangle = \&CORE::tie }
435 entangle $variable, $package, @args;
436
437And for prototypes to be bypassed:
438
439 sub mytie(\[%$*@]$@) {
440 my ($ref, $pack, @args) = @_;
441 ... do something ...
442 goto &CORE::tie;
443 }
444
445Some of these cannot be called through references or via C<&foo> syntax,
446but must be called as barewords.
447
448See L<CORE> for details.
449
450=head2 Other Changes
451
452=head3 Anonymous handles
453
454Automatically generated file handles are now named __ANONIO__ when the
455variable name cannot be determined, rather than $__ANONIO__.
456
457=head3 Autoloaded sort Subroutines
458
459Custom sort subroutines can now be autoloaded [perl #30661]:
460
461 sub AUTOLOAD { ... }
462 @sorted = sort foo @list; # uses AUTOLOAD
463
464=head3 C<continue> no longer requires the "switch" feature
465
466The C<continue> keyword has two meanings. It can introduce a C<continue>
7af9a95b
TC
467block after a loop, or it can exit the current C<when> block. Up to now,
468the latter meaning was valid only with the "switch" feature enabled, and
05c8f9ed
RS
469was a syntax error otherwise. Since the main purpose of feature.pm is to
470avoid conflicts with user-defined subroutines, there is no reason for
471C<continue> to depend on it.
472
473=head3 DTrace probes for interpreter phase change
474
475The C<phase-change> probes will fire when the interpreter's phase
476changes, which tracks the C<${^GLOBAL_PHASE}> variable. C<arg0> is
7af9a95b 477the new phase name; C<arg1> is the old one. This is useful
05c8f9ed 478for limiting your instrumentation to one or more of: compile time,
7af9a95b 479run time, or destruct time.
05c8f9ed
RS
480
481=head3 C<__FILE__()> Syntax
482
483The C<__FILE__>, C<__LINE__> and C<__PACKAGE__> tokens can now be written
484with an empty pair of parentheses after them. This makes them parse the
485same way as C<time>, C<fork> and other built-in functions.
486
487=head3 The C<\$> prototype accepts any scalar lvalue
488
489The C<\$> and C<\[$]> subroutine prototypes now accept any scalar lvalue
7af9a95b 490argument. Previously they accepted only scalars beginning with C<$> and
05c8f9ed
RS
491hash and array elements. This change makes them consistent with the way
492the built-in C<read> and C<recv> functions (among others) parse their
493arguments. This means that one can override the built-in functions with
494custom subroutines that parse their arguments the same way.
495
496=head3 C<_> in subroutine prototypes
497
498The C<_> character in subroutine prototypes is now allowed before C<@> or
499C<%>.
500
501=head1 Security
502
503=head2 Use C<is_utf8_char_buf()> and not C<is_utf8_char()>
504
505The latter function is now deprecated because its API is insufficient to
506guarantee that it doesn't read (up to 12 bytes in the worst case) beyond
507the end of its input string. See
508L<is_utf8_char_buf()|/Added is_utf8_char_buf()>.
509
510=head2 Malformed UTF-8 input could cause attempts to read beyond the end of the buffer
511
512Two new XS-accessible functions, C<utf8_to_uvchr_buf()> and
513C<utf8_to_uvuni_buf()> are now available to prevent this, and the Perl
514core has been converted to use them.
515See L</Internal Changes>.
516
517=head2 C<File::Glob::bsd_glob()> memory error with GLOB_ALTDIRFUNC (CVE-2011-2728).
518
519Calling C<File::Glob::bsd_glob> with the unsupported flag
520GLOB_ALTDIRFUNC would cause an access violation / segfault. A Perl
521program that accepts a flags value from an external source could expose
522itself to denial of service or arbitrary code execution attacks. There
523are no known exploits in the wild. The problem has been corrected by
524explicitly disabling all unsupported flags and setting unused function
006b5090 525pointers to null. Bug reported by Clément Lecigne. (5.14.2)
05c8f9ed
RS
526
527=head2 Privileges are now set correctly when assigning to C<$(>
528
7af9a95b 529A hypothetical bug (probably unexploitable in practice) because the
05c8f9ed 530incorrect setting of the effective group ID while setting C<$(> has been
7af9a95b 531fixed. The bug would have affected only systems that have C<setresgid()>
e019578d 532but not C<setregid()>, but no such systems are known to exist.
05c8f9ed
RS
533
534=head1 Deprecations
535
536=head2 Don't read the Unicode data base files in F<lib/unicore>
537
538It is now deprecated to directly read the Unicode data base files.
539These are stored in the F<lib/unicore> directory. Instead, you should
540use the new functions in L<Unicode::UCD>. These provide a stable API,
541and give complete information.
542
7af9a95b 543Perl may at some point in the future change or remove these files. The
e019578d 544file which applications were most likely to have used is
05c8f9ed
RS
545F<lib/unicore/ToDigit.pl>. L<Unicode::UCD/prop_invmap()> can be used to
546get at its data instead.
547
548=head2 XS functions C<is_utf8_char()>, C<utf8_to_uvchr()> and
549C<utf8_to_uvuni()>
550
551This function is deprecated because it could read beyond the end of the
552input string. Use the new L<is_utf8_char_buf()|/Added is_utf8_char_buf()>,
553C<utf8_to_uvchr_buf()> and C<utf8_to_uvuni_buf()> instead.
554
555=head1 Future Deprecations
556
557This section serves as a notice of features that are I<likely> to be
558removed or L<deprecated|perlpolicy/deprecated> in the next release of
559perl (5.18.0). If your code depends on these features, you should
560contact the Perl 5 Porters via the L<mailing
561list|http://lists.perl.org/list/perl5-porters.html> or L<perlbug> to
562explain your use case and inform the deprecation process.
563
564=head2 Core Modules
565
566These modules may be marked as deprecated I<from the core>. This only
567means that they will no longer be installed by default with the core
568distribution, but will remain available on the CPAN.
569
570=over
571
572=item *
573
574CPANPLUS
575
576=item *
577
578Filter::Simple
579
580=item *
581
582PerlIO::mmap
583
584=item *
585
53b25539
RS
586Pod::LaTeX
587
588=item *
589
590Pod::Parser
05c8f9ed
RS
591
592=item *
593
594SelfLoader
595
596=item *
597
598Text::Soundex
599
600=item *
601
602Thread.pm
603
604=back
605
606=head2 Platforms with no supporting programmers:
607
608These platforms will probably have their
609special build support removed during the
6105.17.0 development series.
611
612=over
613
614=item *
615
616BeOS
617
618=item *
619
620djgpp
621
622=item *
623
624dgux
625
626=item *
627
628EPOC
629
630=item *
631
632MPE/iX
633
634=item *
635
636Rhapsody
637
638=item *
639
640UTS
641
642=item *
643
644VM/ESA
645
646=back
647
648=head2 Other Future Deprecations
649
650=over
651
652=item *
653
654Swapping of $< and $>
655
656For more information about this future deprecation, see L<the relevant RT
657ticket|https://rt.perl.org/rt3/Ticket/Display.html?id=96212>.
658
659=item *
660
661sfio, stdio
662
e57f4500
LT
663Perl supports being built without PerlIO proper, using a stdio or sfio
664wrapper instead. A perl build like this will not support IO layers and
665thus Unicode IO, making it rather handicapped.
666
667PerlIO supports a C<stdio> layer if stdio use is desired, and similarly a
668sfio layer could be produced.
669
05c8f9ed
RS
670=item *
671
672Unescaped literal C<< "{" >> in regular expressions.
673
7af9a95b
TC
674Starting with v5.20, it is planned to require a literal C<"{"> to be
675escaped, for example by preceding it with a backslash. In v5.18, a
676deprecated warning message will be emitted for all such uses.
677This affects only patterns that are to match a literal C<"{">. Other
678uses of this character, such as part of a quantifier or sequence as in
679those below, are completely unaffected:
05c8f9ed
RS
680
681 /foo{3,5}/
682 /\p{Alphabetic}/
683 /\N{DIGIT ZERO}
684
7af9a95b
TC
685Removing this will permit extensions to Perl's pattern syntax and better
686error checking for existing syntax. See L<perlre/Quantifiers> for an
05c8f9ed
RS
687example.
688
081e557a
KW
689=item *
690
7af9a95b 691Revamping C<< "\Q" >> semantics in double-quotish strings when combined with other escapes.
081e557a 692
7af9a95b 693There are several bugs and inconsistencies involving combinations
43b0ea4c 694of C<\Q> and escapes like C<\x>, C<\L>, etc., within a C<\Q...\E> pair.
081e557a
KW
695These need to be fixed, and doing so will necessarily change current
696behavior. The changes have not yet been settled.
697
05c8f9ed
RS
698=back
699
700=head1 Incompatible Changes
701
702=head2 Special blocks called in void context
703
704Special blocks (C<BEGIN>, C<CHECK>, C<INIT>, C<UNITCHECK>, C<END>) are now
705called in void context. This avoids wasteful copying of the result of the
706last statement [perl #108794].
707
708=head2 The C<overloading> pragma and regexp objects
709
710With C<no overloading>, regular expression objects returned by C<qr//> are
711now stringified as "Regexp=REGEXP(0xbe600d)" instead of the regular
712expression itself [perl #108780].
713
714=head2 Two XS typemap Entries removed
715
716Two presumably unused XS typemap entries have been removed from the
717core typemap: T_DATAUNIT and T_CALLBACK. If you are, against all odds,
e019578d 718a user of these, please see the instructions on how to restore them
05c8f9ed
RS
719in L<perlxstypemap>.
720
721=head2 Unicode 6.1 has incompatibilities with Unicode 6.0
722
723These are detailed in L</Supports (almost) Unicode 6.1> above.
724You can compile this version of Perl to use Unicode 6.0. See
725L<perlunicode/Hacking Perl to work on earlier Unicode versions (for very serious hackers only)>.
726
727=head2 Borland compiler
728
729All support for the Borland compiler has been dropped. The code had not
730worked for a long time anyway.
731
732=head2 Certain deprecated Unicode properties are no longer supported by default
733
734Perl should never have exposed certain Unicode properties that are used
735by Unicode internally and not meant to be publicly available. Use of
736these has generated deprecated warning messages since Perl 5.12. The
737removed properties are Other_Alphabetic,
738Other_Default_Ignorable_Code_Point, Other_Grapheme_Extend,
739Other_ID_Continue, Other_ID_Start, Other_Lowercase, Other_Math, and
740Other_Uppercase.
741
742Perl may be recompiled to include any or all of them; instructions are
743given in
744L<perluniprops/Unicode character properties that are NOT accepted by Perl>.
745
746=head2 Dereferencing IO thingies as typeglobs
747
748The C<*{...}> operator, when passed a reference to an IO thingy (as in
749C<*{*STDIN{IO}}>), creates a new typeglob containing just that IO object.
750Previously, it would stringify as an empty string, but some operators would
751treat it as undefined, producing an "uninitialized" warning.
752Now it stringifies as __ANONIO__ [perl #96326].
753
754=head2 User-defined case-changing operations
755
756This feature was deprecated in Perl 5.14, and has now been removed.
757The CPAN module L<Unicode::Casing> provides better functionality without
758the drawbacks that this feature had, as are detailed in the 5.14
759documentation:
760L<http://perldoc.perl.org/5.14.0/perlunicode.html#User-Defined-Case-Mappings-%28for-serious-hackers-only%29>
761
762=head2 XSUBs are now 'static'
763
764XSUB C functions are now 'static', that is, they are not visible from
765outside the compilation unit. Users can use the new C<XS_EXTERNAL(name)>
1540de7c 766and C<XS_INTERNAL(name)> macros to pick the desired linking behavior.
05c8f9ed
RS
767The ordinary C<XS(name)> declaration for XSUBs will continue to declare
768non-'static' XSUBs for compatibility, but the XS compiler,
9d83d6bd 769L<ExtUtils::ParseXS> (C<xsubpp>) will emit 'static' XSUBs by default.
1540de7c 770L<ExtUtils::ParseXS>'s behavior can be reconfigured from XS using the
05c8f9ed
RS
771C<EXPORT_XSUB_SYMBOLS> keyword. See L<perlxs> for details.
772
773=head2 Weakening read-only references
774
775Weakening read-only references is no longer permitted. It should never
7af9a95b 776have worked anyway, and could sometimes result in crashes.
05c8f9ed
RS
777
778=head2 Tying scalars that hold typeglobs
779
780Attempting to tie a scalar after a typeglob was assigned to it would
781instead tie the handle in the typeglob's IO slot. This meant that it was
782impossible to tie the scalar itself. Similar problems affected C<tied> and
783C<untie>: C<tied $scalar> would return false on a tied scalar if the last
784thing returned was a typeglob, and C<untie $scalar> on such a tied scalar
785would do nothing.
786
787We fixed this problem before Perl 5.14.0, but it caused problems with some
788CPAN modules, so we put in a deprecation cycle instead.
789
790Now the deprecation has been removed and this bug has been fixed. So
791C<tie $scalar> will always tie the scalar, not the handle it holds. To tie
792the handle, use C<tie *$scalar> (with an explicit asterisk). The same
793applies to C<tied *$scalar> and C<untie *$scalar>.
794
795=head2 IPC::Open3 no longer provides C<xfork()>, C<xclose_on_exec()>
796and C<xpipe_anon()>
797
7af9a95b 798All three functions were private, undocumented, and unexported. They do
05c8f9ed
RS
799not appear to be used by any code on CPAN. Two have been inlined and one
800deleted entirely.
801
802=head2 C<$$> no longer caches PID
803
804Previously, if one called fork(3) from C, Perl's
805notion of C<$$> could go out of sync with what getpid() returns. By always
806fetching the value of C<$$> via getpid(), this potential bug is eliminated.
807Code that depends on the caching behavior will break. As described in
808L<Core Enhancements|/C<$$> can be assigned to>,
809C<$$> is now writable, but it will be reset during a
810fork.
811
812=head2 C<$$> and C<getppid()> no longer emulate POSIX semantics under LinuxThreads
813
814The POSIX emulation of C<$$> and C<getppid()> under the obsolete
815LinuxThreads implementation has been removed.
816This only impacts users of Linux 2.4 and
817users of Debian GNU/kFreeBSD up to and including 6.0, not the vast
818majority of Linux installations that use NPTL threads.
819
820This means that C<getppid()>, like C<$$>, is now always guaranteed to
821return the OS's idea of the current state of the process, not perl's
822cached version of it.
823
824See the documentation for L<$$|perlvar/$$> for details.
825
826=head2 C<< $< >>, C<< $> >>, C<$(> and C<$)> are no longer cached
827
828Similarly to the changes to C<$$> and C<getppid()>, the internal
829caching of C<< $< >>, C<< $> >>, C<$(> and C<$)> has been removed.
830
831When we cached these values our idea of what they were would drift out
832of sync with reality if someone (e.g., someone embedding perl) called
833C<sete?[ug]id()> without updating C<PL_e?[ug]id>. Having to deal with
834this complexity wasn't worth it given how cheap the C<gete?[ug]id()>
835system call is.
836
837This change will break a handful of CPAN modules that use the XS-level
838C<PL_uid>, C<PL_gid>, C<PL_euid> or C<PL_egid> variables.
839
840The fix for those breakages is to use C<PerlProc_gete?[ug]id()> to
7af9a95b 841retrieve them (e.g., C<PerlProc_getuid()>), and not to assign to
05c8f9ed
RS
842C<PL_e?[ug]id> if you change the UID/GID/EUID/EGID. There is no longer
843any need to do so since perl will always retrieve the up-to-date
844version of those values from the OS.
845
846=head2 Which Non-ASCII characters get quoted by C<quotemeta> and C<\Q> has changed
847
848This is unlikely to result in a real problem, as Perl does not attach
849special meaning to any non-ASCII character, so it is currently
850irrelevant which are quoted or not. This change fixes bug [perl #77654] and
5a2baa23 851brings Perl's behavior more into line with Unicode's recommendations.
05c8f9ed
RS
852See L<perlfunc/quotemeta>.
853
854=head1 Performance Enhancements
855
856=over
857
858=item *
859
860Improved performance for Unicode properties in regular expressions
861
862=for comment Can this be compacted some? -- rjbs, 2012-02-20
863
864Matching a code point against a Unicode property is now done via a
865binary search instead of linear. This means for example that the worst
866case for a 1000 item property is 10 probes instead of 1000. This
867inefficiency has been compensated for in the past by permanently storing
868in a hash the results of a given probe plus the results for the adjacent
86964 code points, under the theory that near-by code points are likely to
870be searched for. A separate hash was used for each mention of a Unicode
871property in each regular expression. Thus, C<qr/\p{foo}abc\p{foo}/>
872would generate two hashes. Any probes in one instance would be unknown
873to the other, and the hashes could expand separately to be quite large
874if the regular expression were used on many different widely-separated
fd85cc12 875code points.
05c8f9ed
RS
876Now, however, there is just one hash shared by all instances of a given
877property. This means that if C<\p{foo}> is matched against "A" in one
878regular expression in a thread, the result will be known immediately to
879all regular expressions, and the relentless march of using up memory is
880slowed considerably.
881
882=item *
883
884Version declarations with the C<use> keyword (e.g., C<use 5.012>) are now
885faster, as they enable features without loading F<feature.pm>.
886
887=item *
888
889C<local $_> is faster now, as it no longer iterates through magic that it
890is not going to copy anyway.
891
892=item *
893
894Perl 5.12.0 sped up the destruction of objects whose classes define
895empty C<DESTROY> methods (to prevent autoloading), by simply not
1540de7c 896calling such empty methods. This release takes this optimization a
05c8f9ed
RS
897step further, by not calling any C<DESTROY> method that begins with a
898C<return> statement. This can be useful for destructors that are only
899used for debugging:
900
901 use constant DEBUG => 1;
902 sub DESTROY { return unless DEBUG; ... }
903
904Constant-folding will reduce the first statement to C<return;> if DEBUG
1540de7c 905is set to 0, triggering this optimization.
05c8f9ed
RS
906
907=item *
908
909Assigning to a variable that holds a typeglob or copy-on-write scalar
910is now much faster. Previously the typeglob would be stringified or
911the copy-on-write scalar would be copied before being clobbered.
912
913=item *
914
915Assignment to C<substr> in void context is now more than twice its
916previous speed. Instead of creating and returning a special lvalue
917scalar that is then assigned to, C<substr> modifies the original string
918itself.
919
920=item *
921
922C<substr> no longer calculates a value to return when called in void
923context.
924
925=item *
926
927Due to changes in L<File::Glob>, Perl's C<glob> function and its C<<
928<...> >> equivalent are now much faster. The splitting of the pattern
7af9a95b 929into words has been rewritten in C, resulting in speed-ups of 20% for
05c8f9ed
RS
930some cases.
931
932This does not affect C<glob> on VMS, as it does not use File::Glob.
933
934=item *
935
936The short-circuiting operators C<&&>, C<||>, and C<//>, when chained
937(such as C<$a || $b || $c>), are now considerably faster to short-circuit,
938due to reduced optree traversal.
939
940=item *
941
942The implementation of C<s///r> makes one fewer copy of the scalar's value.
943
944=item *
945
05c8f9ed
RS
946Recursive calls to lvalue subroutines in lvalue scalar context use less
947memory.
948
949=back
950
951=head1 Modules and Pragmata
952
953=head2 Deprecated Modules
954
955=over
956
957=item L<Version::Requirements>
958
959Version::Requirements is now DEPRECATED, use L<CPAN::Meta::Requirements>,
960which is a drop-in replacement. It will be deleted from perl.git blead
961in v5.17.0.
962
963=back
964
965=head2 New Modules and Pragmata
966
967=over 4
968
969=item *
970
971L<arybase> -- this new module implements the C<$[> variable.
972
973=item *
974
9d83d6bd 975L<PerlIO::mmap> 0.010 has been added to the Perl core.
05c8f9ed
RS
976
977The C<mmap> PerlIO layer is no longer implemented by perl itself, but has
978been moved out into the new L<PerlIO::mmap> module.
979
980=back
981
982=head2 Updated Modules and Pragmata
983
b7f87eaf
RS
984This is only an overview of selected module updates. For a complete list of
985updates, run:
986
987 $ corelist --diff 5.14.0 5.16.0
988
989You can substitute your favorite version in place of 5.14.0, too.
990
05c8f9ed
RS
991=over 4
992
993=item *
994
91acaa3b 995L<Archive::Extract> has been upgraded from version 0.48 to 0.58.
b46815fd
RS
996
997Includes a fix for FreeBSD to only use C<unzip> if it is located in
998C</usr/local/bin>, as FreeBSD 9.0 will ship with a limited C<unzip> in
999C</usr/bin>.
1000
1001=item *
1002
91acaa3b 1003L<Archive::Tar> has been upgraded from version 1.76 to 1.82.
b46815fd
RS
1004
1005Adjustments to handle files >8gb (>0777777777777 octal) and a feature
1006to return the MD5SUM of files in the archive.
1007
1008=item *
1009
91acaa3b 1010L<base> has been upgraded from version 2.16 to 2.18.
b46815fd
RS
1011
1012C<base> no longer sets a module's C<$VERSION> to "-1" when a module it
1013loads does not define a C<$VERSION>. This change has been made because
1014"-1" is not a valid version number under the new "lax" criteria used
1015internally by C<UNIVERSAL::VERSION>. (See L<version> for more on "lax"
1016version criteria.)
1017
1018C<base> no longer internally skips loading modules it has already loaded
1019and instead relies on C<require> to inspect C<%INC>. This fixes a bug
1020when C<base> is used with code that clear C<%INC> to force a module to
1021be reloaded.
1022
1023=item *
1024
91acaa3b 1025L<Carp> has been upgraded from version 1.20 to 1.26.
b46815fd
RS
1026
1027It now includes last read filehandle info and puts a dot after the file
1028and line number, just like errors from C<die> [perl #106538].
1029
1030=item *
1031
91acaa3b 1032L<charnames> has been updated from version 1.18 to 1.30.
b46815fd 1033
959e5662 1034C<charnames> can now be invoked with a new option, C<:loose>,
b46815fd
RS
1035which is like the existing C<:full> option, but enables Unicode loose
1036name matching. Details are in L<charnames/LOOSE MATCHES>.
1037
1038=item *
1039
1040L<B::Deparse> has been upgraded from version 1.03 to 1.14. This fixes
1041numerous deparsing bugs.
1042
1043=item *
1044
91acaa3b 1045L<CGI> has been upgraded from version 3.52 to 3.59.
b46815fd 1046
e019578d 1047It uses the public and documented FCGI.pm API in CGI::Fast. CGI::Fast was
b46815fd
RS
1048using an FCGI API that was deprecated and removed from documentation
1049more than ten years ago. Usage of this deprecated API with FCGI E<gt>=
10500.70 or FCGI E<lt>= 0.73 introduces a security issue.
1051L<https://rt.cpan.org/Public/Bug/Display.html?id=68380>
1052L<http://web.nvd.nist.gov/view/vuln/detail?vulnId=CVE-2011-2766>
1053
1054Things that may break your code:
1055
1056C<url()> was fixed to return C<PATH_INFO> when it is explicitly requested
1057with either the C<path=E<gt>1> or C<path_info=E<gt>1> flag.
1058
1059If your code is running under mod_rewrite (or compatible) and you are
1060calling C<self_url()> or you are calling C<url()> and passing
5f08bd67 1061C<path_info=E<gt>1>, these methods will actually be returning
04521166 1062C<PATH_INFO> now, as you have explicitly requested or C<self_url()>
b46815fd
RS
1063has requested on your behalf.
1064
1065The C<PATH_INFO> has been omitted in such URLs since the issue was
1066introduced in the 3.12 release in December, 2005.
1067
1068This bug is so old your application may have come to depend on it or
1069workaround it. Check for application before upgrading to this release.
1070
1071Examples of affected method calls:
1072
6a9cb335
FC
1073 $q->url(-absolute => 1, -query => 1, -path_info => 1);
1074 $q->url(-path=>1);
1075 $q->url(-full=>1,-path=>1);
1076 $q->url(-rewrite=>1,-path=>1);
b46815fd
RS
1077 $q->self_url();
1078
1079We no longer read from STDIN when the Content-Length is not set,
7af9a95b 1080preventing requests with no Content-Length from sometimes freezing.
b46815fd 1081This is consistent with the CGI RFC 3875, and is also consistent with
71d9f308 1082CGI::Simple. However, the old behavior may have been expected by some
b46815fd
RS
1083command-line uses of CGI.pm.
1084
b8206f26
KW
1085In addition, the DELETE HTTP verb is now supported.
1086
b46815fd
RS
1087=item *
1088
91acaa3b 1089L<Compress::Zlib> has been upgraded from version 2.035 to 2.048.
b46815fd
RS
1090
1091IO::Compress::Zip and IO::Uncompress::Unzip now have support for LZMA
1092(method 14). There is a fix for a CRC issue in IO::Compress::Unzip and
71d9f308 1093it supports Streamed Stored context now. And fixed a Zip64 issue in
b46815fd
RS
1094IO::Compress::Zip when the content size was exactly 0xFFFFFFFF.
1095
1096=item *
1097
91acaa3b 1098L<Digest::SHA> has been upgraded from version 5.61 to 5.71.
b46815fd
RS
1099
1100Added BITS mode to the addfile method and shasum. This makes
7af9a95b 1101partial-byte inputs possible via files/STDIN and lets shasum check
b46815fd
RS
1102all 8074 NIST Msg vectors, where previously special programming was
1103required to do this.
1104
1105=item *
1106
91acaa3b 1107L<Encode> has been upgraded from version 2.42 to 2.44.
b46815fd
RS
1108
1109Missing aliases added, a deep recursion error fixed and various
1110documentation updates.
1111
1112Addressed 'decode_xs n-byte heap-overflow' security bug in Unicode.xs
006b5090 1113(CVE-2011-2939). (5.14.2)
b46815fd
RS
1114
1115=item *
1116
1117L<ExtUtils::CBuilder> updated from version 0.280203 to 0.280206.
1118
1119The new version appends CFLAGS and LDFLAGS to their Config.pm
1120counterparts.
1121
1122=item *
1123
1124L<ExtUtils::ParseXS> has been upgraded from version 2.2210 to 3.16.
1125
1126Much of L<ExtUtils::ParseXS>, the module behind the XS compiler C<xsubpp>,
71d9f308 1127was rewritten and cleaned up. It has been made somewhat more extensible
b46815fd
RS
1128and now finally uses strictures.
1129
1130The typemap logic has been moved into a separate module,
1131L<ExtUtils::Typemaps>. See L</New Modules and Pragmata>, above.
1132
1133For a complete set of changes, please see the ExtUtils::ParseXS
1134changelog, available on the CPAN.
1135
1136=item *
1137
91acaa3b 1138L<File::Glob> has been upgraded from version 1.12 to 1.17.
b46815fd
RS
1139
1140On Windows, tilde (~) expansion now checks the C<USERPROFILE> environment
1141variable, after checking C<HOME>.
1142
1143It has a new C<:bsd_glob> export tag, intended to replace C<:glob>. Like
1144C<:glob> it overrides C<glob> with a function that does not split the glob
1145pattern into words, but, unlike C<:glob>, it iterates properly in scalar
1146context, instead of returning the last file.
1147
1148There are other changes affecting Perl's own C<glob> operator (which uses
1149File::Glob internally, except on VMS). See L</Performance Enhancements>
1150and L</Selected Bug Fixes>.
1151
1152=item *
1153
1154L<FindBin> updated from version 1.50 to 1.51.
1155
1156It no longer returns a wrong result if a script of the same name as the
1157current one exists in the path and is executable.
1158
1159=item *
1160
91acaa3b 1161L<HTTP::Tiny> has been upgraded from version 0.012 to 0.017.
b46815fd
RS
1162
1163Added support for using C<$ENV{http_proxy}> to set the default proxy host.
1164
1165Adds additional shorthand methods for all common HTTP verbs,
1166a C<post_form()> method for POST-ing x-www-form-urlencoded data and
1167a C<www_form_urlencode()> utility method.
1168
1169=item *
1170
1171L<IO> has been upgraded from version 1.25_04 to 1.25_06, and L<IO::Handle>
1172from version 1.31 to 1.33.
1173
1174Together, these upgrades fix a problem with IO::Handle's C<getline> and
1175C<getlines> methods. When these methods are called on the special ARGV
1176handle, the next file is automatically opened, as happens with the built-in
1177C<E<lt>E<gt>> and C<readline> functions. But, unlike the built-ins, these
1178methods were not respecting the caller's use of the L<open> pragma and
7af9a95b 1179applying the appropriate I/O layers to the newly-opened file
b46815fd
RS
1180[rt.cpan.org #66474].
1181
1182=item *
1183
91acaa3b 1184L<IPC::Cmd> has been upgraded from version 0.70 to 0.76.
b46815fd
RS
1185
1186Capturing of command output (both C<STDOUT> and C<STDERR>) is now supported
1187using L<IPC::Open3> on MSWin32 without requiring L<IPC::Run>.
1188
1189=item *
1190
91acaa3b 1191L<IPC::Open3> has been upgraded from version 1.09 to 1.12.
b46815fd 1192
b46815fd
RS
1193Fixes a bug which prevented use of C<open3> on Windows when C<*STDIN>,
1194C<*STDOUT> or C<*STDERR> had been localized.
1195
b46815fd
RS
1196Fixes a bug which prevented duplicating numeric file descriptors on Windows.
1197
b46815fd
RS
1198C<open3> with "-" for the program name works once more. This was broken in
1199version 1.06 (and hence in Perl 5.14.0) [perl #95748].
1200
b46815fd
RS
1201=item *
1202
91acaa3b 1203L<Locale::Codes> has been upgraded from version 3.16 to 3.21.
b46815fd
RS
1204
1205Added Language Extension codes (langext) and Language Variation codes (langvar)
1206as defined in the IANA language registry.
1207
1208Added language codes from ISO 639-5
1209
1210Added language/script codes from the IANA language subtag registry
1211
f0a7e474 1212Fixed an uninitialized value warning [rt.cpan.org #67438].
b46815fd 1213
f0a7e474
FC
1214Fixed the return value for the all_XXX_codes and all_XXX_names functions
1215[rt.cpan.org #69100].
b46815fd
RS
1216
1217Reorganized modules to move Locale::MODULE to Locale::Codes::MODULE to allow
1218for cleaner future additions. The original four modules (Locale::Language,
1219Locale::Currency, Locale::Country, Locale::Script) will continue to work, but
1220all new sets of codes will be added in the Locale::Codes namespace.
1221
1222The code2XXX, XXX2code, all_XXX_codes, and all_XXX_names functions now
1223support retired codes. All codesets may be specified by a constant or
1224by their name now. Previously, they were specified only by a constant.
1225
1226The alias_code function exists for backward compatibility. It has been
1227replaced by rename_country_code. The alias_code function will be
1228removed some time after September, 2013.
1229
1230All work is now done in the central module (Locale::Codes). Previously,
1231some was still done in the wrapper modules (Locale::Codes::*). Added
1232Language Family codes (langfam) as defined in ISO 639-5.
1233
1234=item *
1235
1236L<Math::BigFloat> has been upgraded from version 1.993 to 1.997.
1237
1540de7c 1238The C<numify> method has been corrected to return a normalized Perl number
b46815fd
RS
1239(the result of C<0 + $thing>), instead of a string [rt.cpan.org #66732].
1240
1241=item *
1242
1243L<Math::BigInt> has been upgraded from version 1.994 to 1.998.
1244
1245It provides a new C<bsgn> method that complements the C<babs> method.
1246
1247It fixes the internal C<objectify> function's handling of "foreign objects"
1248so they are converted to the appropriate class (Math::BigInt or
1249Math::BigFloat).
1250
1251=item *
1252
91acaa3b 1253L<Math::BigRat> has been upgraded from version 0.2602 to 0.2603.
b46815fd
RS
1254
1255C<int()> on a Math::BigRat object containing -1/2 now creates a
1256Math::BigInt containing 0, rather than -0. L<Math::BigInt> does not even
1257support negative zero, so the resulting object was actually malformed
1258[perl #95530].
1259
1260=item *
1261
91acaa3b
FC
1262L<Math::Complex> has been upgraded from version 1.56 to 1.59
1263and L<Math::Trig> from version 1.2 to 1.22.
b46815fd
RS
1264
1265Fixes include: correct copy constructor usage; fix polarwise formatting with
1266numeric format specifier; and more stable C<great_circle_direction> algorithm.
1267
1268=item *
1269
1270L<Module::CoreList> has been upgraded from version 2.51 to 2.66.
1271
1272The C<corelist> utility now understands the C<-r> option for displaying
1273Perl release dates and the C<--diff> option to print the set of modlib
1274changes between two perl distributions.
1275
1276=item *
1277
9d83d6bd 1278L<Module::Metadata> has been upgraded from version 1.000004 to 1.000009.
b46815fd
RS
1279
1280Adds C<provides> method to generate a CPAN META provides data structure
1281correctly; use of C<package_versions_from_directory> is discouraged.
1282
1283=item *
1284
91acaa3b 1285L<ODBM_File> has been upgraded from version 1.10 to 1.12.
b46815fd
RS
1286
1287The XS code is now compiled with C<PERL_NO_GET_CONTEXT>, which will aid
1288performance under ithreads.
1289
1290=item *
1291
1292L<open> has been upgraded from version 1.08 to 1.10.
1293
62f11196 1294It no longer turns off layers on standard handles when invoked without the
b46815fd
RS
1295":std" directive. Similarly, when invoked I<with> the ":std" directive, it
1296now clears layers on STDERR before applying the new ones, and not just on
1297STDIN and STDOUT [perl #92728].
1298
1299=item *
1300
91acaa3b 1301L<overload> has been upgraded from version 1.13 to 1.18.
b46815fd
RS
1302
1303C<overload::Overloaded> no longer calls C<can> on the class, but uses
1304another means to determine whether the object has overloading. It was
1305never correct for it to call C<can>, as overloading does not respect
1306AUTOLOAD. So classes that autoload methods and implement C<can> no longer
1307have to account for overloading [perl #40333].
1308
1309A warning is now produced for invalid arguments. See L</New Diagnostics>.
1310
1311=item *
1312
1313L<PerlIO::scalar> has been upgraded from version 0.11 to 0.14.
1314
1315(This is the module that implements C<< open $fh, '>', \$scalar >>.)
1316
1317It fixes a problem with C<< open my $fh, ">", \$scalar >> not working if
006b5090 1318C<$scalar> is a copy-on-write scalar. (5.14.2)
b46815fd
RS
1319
1320It also fixes a hang that occurs with C<readline> or C<< <$fh> >> if a
1321typeglob has been assigned to $scalar [perl #92258].
1322
1323It no longer assumes during C<seek> that $scalar is a string internally.
1324If it didn't crash, it was close to doing so [perl #92706]. Also, the
1325internal print routine no longer assumes that the position set by C<seek>
1326is valid, but extends the string to that position, filling the intervening
1327bytes (between the old length and the seek position) with nulls
1328[perl #78980].
1329
1330Printing to an in-memory handle now works if the $scalar holds a reference,
1331stringifying the reference before modifying it. References used to be
1332treated as empty strings.
1333
1334Printing to an in-memory handle no longer crashes if the $scalar happens to
1335hold a number internally, but no string buffer.
1336
1337Printing to an in-memory handle no longer creates scalars that confuse
1338the regular expression engine [perl #108398].
1339
1340=item *
1341
9d83d6bd 1342L<Pod::Functions> has been upgraded from version 1.04 to 1.05.
b46815fd
RS
1343
1344F<Functions.pm> is now generated at perl build time from annotations in
71d9f308 1345F<perlfunc.pod>. This will ensure that L<Pod::Functions> and L<perlfunc>
b46815fd
RS
1346remain in synchronisation.
1347
1348=item *
1349
1350L<Pod::Html> has been upgraded from version 1.11 to 1.1502.
1351
1352This is an extensive rewrite of Pod::Html to use L<Pod::Simple> under
1353the hood. The output has changed significantly.
1354
1355=item *
1356
1357L<Pod::Perldoc> has been upgraded from version 3.15_03 to 3.17.
1358
1359It corrects the search paths on VMS [perl #90640]. (5.14.1)
1360
1361The B<-v> option now fetches the right section for C<$0>.
1362
1363This upgrade has numerous significant fixes. Consult its changelog on
1364the CPAN for more information.
1365
1366=item *
1367
91acaa3b 1368L<POSIX> has been upgraded from version 1.24 to 1.30.
b46815fd 1369
71d9f308 1370L<POSIX> no longer uses L<AutoLoader>. Any code which was relying on this
7af9a95b 1371implementation detail was buggy, and may fail because of this change.
b46815fd 1372The module's Perl code has been considerably simplified, roughly halving
71d9f308 1373the number of lines, with no change in functionality. The XS code has
b46815fd 1374been refactored to reduce the size of the shared object by about 12%,
71d9f308 1375with no change in functionality. More POSIX functions now have tests.
b46815fd 1376
26be0cae 1377C<sigsuspend> and C<pause> now run signal handlers before returning, as the
b46815fd
RS
1378whole point of these two functions is to wait until a signal has
1379arrived, and then return I<after> it has been triggered. Delayed, or
1380"safe", signals were preventing that from happening, possibly resulting in
1381race conditions [perl #107216].
1382
1383C<POSIX::sleep> is now a direct call into the underlying OS C<sleep>
71d9f308
FC
1384function, instead of being a Perl wrapper on C<CORE::sleep>.
1385C<POSIX::dup2> now returns the correct value on Win32 (I<i.e.>, the file
1386descriptor). C<POSIX::SigSet> C<sigsuspend> and C<sigpending> and
1387C<POSIX::pause> now dispatch safe signals immediately before returning to
1388their caller.
b46815fd
RS
1389
1390C<POSIX::Termios::setattr> now defaults the third argument to C<TCSANOW>,
7af9a95b 1391instead of 0. On most platforms C<TCSANOW> is defined to be 0, but on some
b46815fd
RS
13920 is not a valid parameter, which caused a call with defaults to fail.
1393
1394=item *
1395
23296748 1396L<Socket> has been upgraded from version 1.94 to 2.001.
b46815fd
RS
1397
1398It has new functions and constants for handling IPv6 sockets:
1399
1400 pack_ipv6_mreq
1401 unpack_ipv6_mreq
1402 IPV6_ADD_MEMBERSHIP
1403 IPV6_DROP_MEMBERSHIP
1404 IPV6_MTU
1405 IPV6_MTU_DISCOVER
1406 IPV6_MULTICAST_HOPS
1407 IPV6_MULTICAST_IF
1408 IPV6_MULTICAST_LOOP
1409 IPV6_UNICAST_HOPS
1410 IPV6_V6ONLY
1411
1412=item *
1413
91acaa3b 1414L<Storable> has been upgraded from version 2.27 to 2.34.
b46815fd
RS
1415
1416It no longer turns copy-on-write scalars into read-only scalars when
1417freezing and thawing.
1418
1419=item *
1420
91acaa3b 1421L<Sys::Syslog> has been upgraded from version 0.27 to 0.29.
b46815fd
RS
1422
1423This upgrade closes many outstanding bugs.
1424
1425=item *
1426
91acaa3b 1427L<Term::ANSIColor> has been upgraded from version 3.00 to 3.01.
b46815fd
RS
1428
1429Only interpret an initial array reference as a list of colors, not any initial
1430reference, allowing the colored function to work properly on objects with
1431stringification defined.
1432
1433=item *
1434
9d83d6bd 1435L<Term::ReadLine> has been upgraded from version 1.07 to 1.09.
b46815fd
RS
1436
1437Term::ReadLine now supports any event loop, including unpublished ones and
ebca2fa8 1438simple L<IO::Select>, loops without the need to rewrite existing code for
b46815fd
RS
1439any particular framework [perl #108470].
1440
1441=item *
1442
1443L<threads::shared> has been upgraded from version 1.37 to 1.40.
1444
1445Destructors on shared objects used to be ignored sometimes if the objects
1446were referenced only by shared data structures. This has been mostly
1447fixed, but destructors may still be ignored if the objects still exist at
1448global destruction time [perl #98204].
1449
1450=item *
1451
91acaa3b 1452L<Unicode::Collate> has been upgraded from version 0.73 to 0.89.
b46815fd
RS
1453
1454Updated to CLDR 1.9.1
1455
1456Locales updated to CLDR 2.0: mk, mt, nb, nn, ro, ru, sk, sr, sv, uk,
1457zh__pinyin, zh__stroke
1458
1459Newly supported locales: bn, fa, ml, mr, or, pa, sa, si, si__dictionary,
1460sr_Latn, sv__reformed, ta, te, th, ur, wae.
1461
1462Tailored compatibility ideographs as well as unified ideographs for the
1463locales: ja, ko, zh__big5han, zh__gb2312han, zh__pinyin, zh__stroke.
1464
1465Locale/*.pl files are now searched for in @INC.
1466
1467=item *
1468
91acaa3b 1469L<Unicode::Normalize> has been upgraded from version 1.10 to 1.14.
b46815fd
RS
1470
1471Fixes for the removal of F<unicore/CompositionExclusions.txt> from core.
1472
1473=item *
1474
91acaa3b 1475L<Unicode::UCD> has been upgraded from version 0.32 to 0.43.
b46815fd 1476
ebca2fa8 1477This adds four new functions: C<prop_aliases()> and
7af9a95b 1478C<prop_value_aliases()>, which are used to find all Unicode-approved
b46815fd 1479synonyms for property names, or to convert from one name to another;
7af9a95b 1480C<prop_invlist> which returns all code points matching a given
b46815fd
RS
1481Unicode binary property; and C<prop_invmap> which returns the complete
1482specification of a given Unicode property.
1483
1484=item *
1485
91acaa3b 1486L<Win32API::File> has been upgraded from version 0.1101 to 0.1200.
b46815fd
RS
1487
1488Added SetStdHandle and GetStdHandle functions
05c8f9ed
RS
1489
1490=back
1491
1492=head2 Removed Modules and Pragmata
1493
1494As promised in Perl 5.14.0's release notes, the following modules have
1495been removed from the core distribution, and if needed should be installed
1496from CPAN instead.
1497
1498=over
1499
1500=item *
1501
959e5662 1502L<Devel::DProf> has been removed from the Perl core. Prior version was
05c8f9ed
RS
150320110228.00.
1504
1505=item *
1506
959e5662 1507L<Shell> has been removed from the Perl core. Prior version was 0.72_01.
05c8f9ed 1508
cfcbecfc
RS
1509=item *
1510
1511Several old perl4-style libraries which have been deprecated with 5.14
1512are now removed:
1513
1514 abbrev.pl assert.pl bigfloat.pl bigint.pl bigrat.pl cacheout.pl
1515 complete.pl ctime.pl dotsh.pl exceptions.pl fastcwd.pl flush.pl
1516 getcwd.pl getopt.pl getopts.pl hostname.pl importenv.pl
1517 lib/find{,depth}.pl look.pl newgetopt.pl open2.pl open3.pl
1518 pwd.pl shellwords.pl stat.pl tainted.pl termcap.pl timelocal.pl
1519
1520They can be found on CPAN as L<Perl4::CoreLibs>.
1521
05c8f9ed
RS
1522=back
1523
1524=head1 Documentation
1525
1526=head2 New Documentation
1527
1528=head3 L<perldtrace>
1529
1530L<perldtrace> describes Perl's DTrace support, listing the provided probes
1531and gives examples of their use.
1532
1533=head3 L<perlexperiment>
1534
1535This document is intended to provide a list of experimental features in
1536Perl. It is still a work in progress.
1537
1538=head3 L<perlootut>
1539
1540This a new OO tutorial. It focuses on basic OO concepts, and then recommends
1541that readers choose an OO framework from CPAN.
1542
1543=head3 L<perlxstypemap>
1544
1545The new manual describes the XS typemapping mechanism in unprecedented
1546detail and combines new documentation with information extracted from
1547L<perlxs> and the previously unofficial list of all core typemaps.
1548
1549=head2 Changes to Existing Documentation
1550
1551=head3 L<perlapi>
1552
1553=over 4
1554
1555=item *
1556
7af9a95b
TC
1557The HV API has long accepted negative lengths to show that the key is
1558in UTF8. This is now documented.
05c8f9ed
RS
1559
1560=item *
1561
1562The C<boolSV()> macro is now documented.
1563
1564=back
1565
1566=head3 L<perlfunc>
1567
1568=over 4
1569
1570=item *
1571
1572C<dbmopen> treats a 0 mode as a special case, that prevents a nonexistent
1573file from being created. This has been the case since Perl 5.000, but was
1574never documented anywhere. Now the perlfunc entry mentions it
1575[perl #90064].
1576
1577=item *
1578
21c10391 1579As an accident of history, C<open $fh, '<:', ...> applies the default
05c8f9ed
RS
1580layers for the platform (C<:raw> on Unix, C<:crlf> on Windows), ignoring
1581whatever is declared by L<open.pm|open>. This seems such a useful feature
1582it has been documented in L<perlfunc|perlfunc/open> and L<open>.
1583
1584=item *
1585
1586The entry for C<split> has been rewritten. It is now far clearer than
1587before.
1588
1589=back
1590
1591=head3 L<perlguts>
1592
1593=over 4
1594
1595=item *
1596
1597A new section, L<Autoloading with XSUBs|perlguts/Autoloading with XSUBs>,
1598has been added, which explains the two APIs for accessing the name of the
1599autoloaded sub.
1600
1601=item *
1602
7af9a95b 1603Some function descriptions in L<perlguts> were confusing, as it was
05c8f9ed
RS
1604not clear whether they referred to the function above or below the
1605description. This has been clarified [perl #91790].
1606
1607=back
1608
1609=head3 L<perlobj>
1610
1611=over 4
1612
1613=item *
1614
1615This document has been rewritten from scratch, and its coverage of various OO
1616concepts has been expanded.
1617
1618=back
1619
1620=head3 L<perlop>
1621
1622=over 4
1623
1624=item *
1625
1626Documentation of the smartmatch operator has been reworked and moved from
1627perlsyn to perlop where it belongs.
1628
1629It has also been corrected for the case of C<undef> on the left-hand
1540de7c 1630side. The list of different smart match behaviors had an item in the
05c8f9ed
RS
1631wrong place.
1632
1633=item *
1634
1635Documentation of the ellipsis statement (C<...>) has been reworked and
1636moved from perlop to perlsyn.
1637
1638=item *
1639
1640The explanation of bitwise operators has been expanded to explain how they
1641work on Unicode strings (5.14.1).
1642
1643=item *
1644
1645More examples for C<m//g> have been added (5.14.1).
1646
1647=item *
1648
1649The C<<< <<\FOO >>> here-doc syntax has been documented (5.14.1).
1650
1651=back
1652
1653=head3 L<perlpragma>
1654
1655=over 4
1656
1657=item *
1658
1659There is now a standard convention for naming keys in the C<%^H>,
1660documented under L<Key naming|perlpragma/Key naming>.
1661
1662=back
1663
1664=head3 L<perlsec/Laundering and Detecting Tainted Data>
1665
1666=over 4
1667
1668=item *
1669
1670The example function for checking for taintedness contained a subtle
1671error. C<$@> needs to be localized to prevent its changing this
1672global's value outside the function. The preferred method to check for
1673this remains L<Scalar::Util/tainted>.
1674
1675=back
1676
1677=head3 L<perllol>
1678
1679=over
1680
1681=item *
1682
1683L<perllol> has been expanded with examples using the new C<push $scalar>
1684syntax introduced in Perl 5.14.0 (5.14.1).
1685
1686=back
1687
1688=head3 L<perlmod>
1689
1690=over
1691
1692=item *
1693
1694L<perlmod> now states explicitly that some types of explicit symbol table
1695manipulation are not supported. This codifies what was effectively already
1696the case [perl #78074].
1697
1698=back
1699
1700=head3 L<perlpodstyle>
1701
1702=over 4
1703
1704=item *
1705
1706The tips on which formatting codes to use have been corrected and greatly
1707expanded.
1708
1709=item *
1710
1711There are now a couple of example one-liners for previewing POD files after
1712they have been edited.
1713
1714=back
1715
1716=head3 L<perlre>
1717
1718=over
1719
1720=item *
1721
1722The C<(*COMMIT)> directive is now listed in the right section
1723(L<Verbs without an argument|perlre/Verbs without an argument>).
1724
1725=back
1726
1727=head3 L<perlrun>
1728
1729=over
1730
1731=item *
1732
1733L<perlrun> has undergone a significant clean-up. Most notably, the
1734B<-0x...> form of the B<-0> flag has been clarified, and the final section
1735on environment variables has been corrected and expanded (5.14.1).
1736
1737=back
1738
1739=head3 L<perlsub>
1740
1741=over
1742
1743=item *
1744
1745The ($;) prototype syntax, which has existed for rather a long time, is now
7af9a95b 1746documented in L<perlsub>. It lets a unary function have the same
05c8f9ed
RS
1747precedence as a list operator.
1748
1749=back
1750
1751=head3 L<perltie>
1752
1753=over
1754
1755=item *
1756
1757The required syntax for tying handles has been documented.
1758
1759=back
1760
1761=head3 L<perlvar>
1762
1763=over
1764
1765=item *
1766
1767The documentation for L<$!|perlvar/$!> has been corrected and clarified.
1768It used to state that $! could be C<undef>, which is not the case. It was
7af9a95b 1769also unclear whether system calls set C's C<errno> or Perl's C<$!>
05c8f9ed
RS
1770[perl #91614].
1771
1772=item *
1773
1774Documentation for L<$$|perlvar/$$> has been amended with additional
1775cautions regarding changing the process ID.
1776
1777=back
1778
1779=head3 Other Changes
1780
1781=over 4
1782
1783=item *
1784
1785L<perlxs> was extended with documentation on inline typemaps.
1786
1787=item *
1788
1789L<perlref> has a new L<Circular References|perlref/Circular References>
1790section explaining how circularities may not be freed and how to solve that
1791with weak references.
1792
1793=item *
1794
1795Parts of L<perlapi> were clarified, and Perl equivalents of some C
1796functions have been added as an additional mode of exposition.
1797
1798=item *
1799
1800A few parts of L<perlre> and L<perlrecharclass> were clarified.
1801
1802=back
1803
1804=head2 Removed Documentation
1805
1806=head3 Old OO Documentation
1807
7af9a95b 1808The old OO tutorials, perltoot, perltooc, and perlboot, have been
05c8f9ed
RS
1809removed. The perlbot (bag of object tricks) document has been removed
1810as well.
1811
1812=head3 Development Deltas
1813
1814The perldelta files for development releases are no longer packaged with
1815perl. These can still be found in the perl source code repository.
1816
1817=head1 Diagnostics
1818
1819The following additions or changes have been made to diagnostic output,
1820including warnings and fatal error messages. For the complete list of
1821diagnostic messages, see L<perldiag>.
1822
1823=head2 New Diagnostics
1824
1825=head3 New Errors
1826
1827=over 4
1828
1829=item *
1830
1831L<Cannot set tied @DB::args|perldiag/"Cannot set tied @DB::args">
1832
1833This error occurs when C<caller> tries to set C<@DB::args> but finds it
1834tied. Before this error was added, it used to crash instead.
1835
1836=item *
1837
1838L<Cannot tie unreifiable array|perldiag/"Cannot tie unreifiable array">
1839
1840This error is part of a safety check that the C<tie> operator does before
1841tying a special array like C<@_>. You should never see this message.
1842
1843=item *
1844
1845L<&CORE::%s cannot be called directly|perldiag/"&CORE::%s cannot be called directly">
1846
1847This occurs when a subroutine in the C<CORE::> namespace is called
1848with C<&foo> syntax or through a reference. Some subroutines
1849in this package cannot yet be called that way, but must be
1850called as barewords. See L</Subroutines in the C<CORE> namespace>, above.
1851
1852=item *
1853
1854L<Source filters apply only to byte streams|perldiag/"Source filters apply only to byte streams">
1855
1856This new error occurs when you try to activate a source filter (usually by
1857loading a source filter module) within a string passed to C<eval> under the
1858C<unicode_eval> feature.
1859
1860=back
1861
1862=head3 New Warnings
1863
1864=over 4
1865
1866=item *
1867
1868L<defined(@array) is deprecated|perldiag/"defined(@array) is deprecated">
1869
1870The long-deprecated C<defined(@array)> now also warns for package variables.
7af9a95b 1871Previously it issued a warning for lexical variables only.
05c8f9ed
RS
1872
1873=item *
1874
1875L<length() used on %s|perldiag/length() used on %s>
1876
1877This new warning occurs when C<length> is used on an array or hash, instead
1878of C<scalar(@array)> or C<scalar(keys %hash)>.
1879
1880=item *
1881
1882L<lvalue attribute %s already-defined subroutine|perldiag/"lvalue attribute %s already-defined subroutine">
1883
1884L<attributes.pm|attributes> now emits this warning when the :lvalue
1885attribute is applied to a Perl subroutine that has already been defined, as
1886doing so can have unexpected side-effects.
1887
1888=item *
1889
1890L<overload arg '%s' is invalid|perldiag/"overload arg '%s' is invalid">
1891
1892This warning, in the "overload" category, is produced when the overload
1893pragma is given an argument it doesn't recognize, presumably a mistyped
1894operator.
1895
1896=item *
1897
1898L<$[ used in %s (did you mean $] ?)|perldiag/"$[ used in %s (did you mean $] ?)">
1899
1900This new warning exists to catch the mistaken use of C<$[> in version
1901checks. C<$]>, not C<$[>, contains the version number.
1902
1903=item *
1904
1905L<Useless assignment to a temporary|perldiag/"Useless assignment to a temporary">
1906
1907Assigning to a temporary scalar returned
1908from an lvalue subroutine now produces this
1909warning [perl #31946].
1910
1911=item *
1912
1913L<Useless use of \E|perldiag/"Useless use of \E">
1914
1915C<\E> does nothing unless preceded by C<\Q>, C<\L> or C<\U>.
1916
1917=back
1918
1919=head2 Removed Errors
1920
1921=over
1922
1923=item *
1924
1925"sort is now a reserved word"
1926
1927This error used to occur when C<sort> was called without arguments,
1928followed by C<;> or C<)>. (E.g., C<sort;> would die, but C<{sort}> was
1929OK.) This error message was added in Perl 3 to catch code like
1930C<close(sort)> which would no longer work. More than two decades later,
1931this message is no longer appropriate. Now C<sort> without arguments is
1932always allowed, and returns an empty list, as it did in those cases
1933where it was already allowed [perl #90030].
1934
1935=back
1936
1937=head2 Changes to Existing Diagnostics
1938
1939=over 4
1940
1941=item *
1942
1943The "Applying pattern match..." or similar warning produced when an
1944array or hash is on the left-hand side of the C<=~> operator now
1945mentions the name of the variable.
1946
1947=item *
1948
1949The "Attempt to free non-existent shared string" has had the spelling
1950of "non-existent" corrected to "nonexistent". It was already listed
1951with the correct spelling in L<perldiag>.
1952
1953=item *
1954
7af9a95b 1955The error messages for using C<default> and C<when> outside a
1540de7c 1956topicalizer have been standardized to match the messages for C<continue>
05c8f9ed
RS
1957and loop controls. They now read 'Can't "default" outside a
1958topicalizer' and 'Can't "when" outside a topicalizer'. They both used
1959to be 'Can't use when() outside a topicalizer' [perl #91514].
1960
1961=item *
1962
1963The message, "Code point 0x%X is not Unicode, no properties match it;
1964all inverse properties do" has been changed to "Code point 0x%X is not
1965Unicode, all \p{} matches fail; all \P{} matches succeed".
1966
1967=item *
1968
1969Redefinition warnings for constant subroutines used to be mandatory,
1970even occurring under C<no warnings>. Now they respect the L<warnings>
1971pragma.
1972
1973=item *
1974
1975The "glob failed" warning message is now suppressible via C<no warnings>
1976[perl #111656].
1977
1978=item *
1979
1980The L<Invalid version format|perldiag/"Invalid version format (%s)">
1981error message now says "negative version number" within the parentheses,
1982rather than "non-numeric data", for negative numbers.
1983
1984=item *
1985
1986The two warnings
1987L<Possible attempt to put comments in qw() list|perldiag/"Possible attempt to put comments in qw() list">
1988and
1989L<Possible attempt to separate words with commas|perldiag/"Possible attempt to separate words with commas">
1990are no longer mutually exclusive: the same C<qw> construct may produce
1991both.
1992
1993=item *
1994
1995The uninitialized warning for C<y///r> when C<$_> is implicit and
1996undefined now mentions the variable name, just like the non-/r variation
1997of the operator.
1998
1999=item *
2000
2001The 'Use of "foo" without parentheses is ambiguous' warning has been
2002extended to apply also to user-defined subroutines with a (;$)
2003prototype, and not just to built-in functions.
2004
2005=item *
2006
2007Warnings that mention the names of lexical (C<my>) variables with
2008Unicode characters in them now respect the presence or absence of the
2009C<:utf8> layer on the output handle, instead of outputting UTF8
2010regardless. Also, the correct names are included in the strings passed
2011to C<$SIG{__WARN__}> handlers, rather than the raw UTF8 bytes.
2012
2013=back
2014
2015=head1 Utility Changes
2016
2017=head3 L<h2ph>
2018
2019=over 4
2020
2021=item *
2022
2023L<h2ph> used to generate code of the form
2024
2025 unless(defined(&FOO)) {
2026 sub FOO () {42;}
2027 }
2028
2029But the subroutine is a compile-time declaration, and is hence unaffected
2030by the condition. It has now been corrected to emit a string C<eval>
2031around the subroutine [perl #99368].
2032
2033=back
2034
2035=head3 L<splain>
2036
2037=over 4
2038
2039=item *
2040
2041F<splain> no longer emits backtraces with the first line number repeated.
2042
2043This:
2044
2045 Uncaught exception from user code:
2046 Cannot fwiddle the fwuddle at -e line 1.
2047 at -e line 1
2048 main::baz() called at -e line 1
2049 main::bar() called at -e line 1
2050 main::foo() called at -e line 1
2051
2052has become this:
2053
2054 Uncaught exception from user code:
2055 Cannot fwiddle the fwuddle at -e line 1.
2056 main::baz() called at -e line 1
2057 main::bar() called at -e line 1
2058 main::foo() called at -e line 1
2059
2060=item *
2061
2062Some error messages consist of multiple lines that are listed as separate
2063entries in L<perldiag>. splain has been taught to find the separate
2064entries in these cases, instead of simply failing to find the message.
2065
2066=back
2067
2068=head3 L<zipdetails>
2069
2070=over 4
2071
2072=item *
2073
2074This is a new utility, included as part of an
2075L<IO::Compress::Base> upgrade.
2076
2077L<zipdetails> displays information about the internal record structure
2078of the zip file. It is not concerned with displaying any details of
2079the compressed data stored in the zip file.
2080
2081=back
2082
2083=head1 Configuration and Compilation
2084
2085=over 4
2086
2087=item *
2088
2089F<regexp.h> has been modified for compatibility with GCC's B<-Werror>
2090option, as used by some projects that include perl's header files (5.14.1).
2091
2092=item *
2093
2094C<USE_LOCALE{,_COLLATE,_CTYPE,_NUMERIC}> have been added the output of perl -V
1540de7c 2095as they have affect the behavior of the interpreter binary (albeit
7af9a95b 2096in only a small area).
05c8f9ed
RS
2097
2098=item *
2099
2100The code and tests for L<IPC::Open2> have been moved from F<ext/IPC-Open2>
2101into F<ext/IPC-Open3>, as C<IPC::Open2::open2()> is implemented as a thin
2102wrapper around C<IPC::Open3::_open3()>, and hence is very tightly coupled to
2103it.
2104
2105=item *
2106
2107The magic types and magic vtables are now generated from data in a new script
2108F<regen/mg_vtable.pl>, instead of being maintained by hand. As different
2109EBCDIC variants can't agree on the code point for '~', the character to code
2110point conversion is done at build time by F<generate_uudmap> to a new generated
2111header F<mg_data.h>. C<PL_vtbl_bm> and C<PL_vtbl_fm> are now defined by the
2112pre-processor as C<PL_vtbl_regexp>, instead of being distinct C variables.
2113C<PL_vtbl_sig> has been removed.
2114
2115=item *
2116
2117Building with C<-DPERL_GLOBAL_STRUCT> works again. This configuration is not
2118generally used.
2119
2120=item *
2121
2122Perl configured with I<MAD> now correctly frees C<MADPROP> structures when
2123OPs are freed. C<MADPROP>s are now allocated with C<PerlMemShared_malloc()>
2124
2125=item *
2126
2127F<makedef.pl> has been refactored. This should have no noticeable affect on
2128any of the platforms that use it as part of their build (AIX, VMS, Win32).
2129
2130=item *
2131
2132C<useperlio> can no longer be disabled.
2133
2134=item *
2135
2136The file F<global.sym> is no longer needed, and has been removed. It
2137contained a list of all exported functions, one of the files generated by
2138F<regen/embed.pl> from data in F<embed.fnc> and F<regen/opcodes>. The code
2139has been refactored so that the only user of F<global.sym>, F<makedef.pl>,
2140now reads F<embed.fnc> and F<regen/opcodes> directly, removing the need to
2141store the list of exported functions in an intermediate file.
2142
2143As F<global.sym> was never installed, this change should not be visible
2144outside the build process.
2145
2146=item *
2147
2148F<pod/buildtoc>, used by the build process to build L<perltoc>, has been
7af9a95b 2149refactored and simplified. It now contains only code to build L<perltoc>;
05c8f9ed
RS
2150the code to regenerate Makefiles has been moved to F<Porting/pod_rules.pl>.
2151It's a bug if this change has any material effect on the build process.
2152
2153=item *
2154
2155F<pod/roffitall> is now built by F<pod/buildtoc>, instead of being
2156shipped with the distribution. Its list of manpages is now generated
2157(and therefore current). See also RT #103202 for an unresolved related
2158issue.
2159
2160=item *
2161
2162The man page for C<XS::Typemap> is no longer installed. C<XS::Typemap>
2163is a test module which is not installed, hence installing its
2164documentation makes no sense.
2165
2166=item *
2167
2168The -Dusesitecustomize and -Duserelocatableinc options now work
2169together properly.
2170
2171=back
2172
2173=head1 Platform Support
2174
2175=head2 Platform-Specific Notes
2176
2177=head3 Cygwin
2178
2179=over 4
2180
2181=item *
2182
2183Since version 1.7, Cygwin supports native UTF-8 paths. If Perl is built
2184under that environment, directory and filenames will be UTF-8 encoded.
2185
959e5662
BO
2186=item *
2187
05c8f9ed
RS
2188Cygwin does not initialize all original Win32 environment variables. See
2189F<README.cygwin> for a discussion of the newly-added
2190C<Cygwin::sync_winenv()> function [perl #110190] and for
2191further links.
2192
2193=back
2194
2195=head3 HP-UX
2196
2197=over 4
2198
2199=item *
2200
2201HP-UX PA-RISC/64 now supports gcc-4.x
2202
2203A fix to correct the socketsize now makes the test suite pass on HP-UX
006b5090 2204PA-RISC for 64bitall builds. (5.14.2)
05c8f9ed
RS
2205
2206=back
2207
2208=head3 VMS
2209
2210=over 4
2211
2212=item *
2213
2214Remove unnecessary includes, fix miscellaneous compiler warnings and
2215close some unclosed comments on F<vms/vms.c>.
2216
959e5662
BO
2217=item *
2218
05c8f9ed
RS
2219Remove sockadapt layer from the VMS build.
2220
2221=item *
2222
7af9a95b
TC
2223Explicit support for VMS versions before v7.0 and DEC C versions
2224before v6.0 has been removed.
05c8f9ed
RS
2225
2226=item *
2227
2228Since Perl 5.10.1, the home-grown C<stat> wrapper has been unable to
2229distinguish between a directory name containing an underscore and an
2230otherwise-identical filename containing a dot in the same position
2231(e.g., t/test_pl as a directory and t/test.pl as a file). This problem
2232has been corrected.
2233
2234=item *
2235
7af9a95b 2236The build on VMS now permits names of the resulting symbols in C code for
05c8f9ed
RS
2237Perl longer than 31 characters. Symbols like
2238C<Perl__it_was_the_best_of_times_it_was_the_worst_of_times> can now be
2239created freely without causing the VMS linker to seize up.
2240
2241=back
2242
2243=head3 GNU/Hurd
2244
959e5662
BO
2245=over 4
2246
2247=item *
2248
05c8f9ed
RS
2249Numerous build and test failures on GNU/Hurd have been resolved with hints
2250for building DBM modules, detection of the library search path, and enabling
2251of large file support.
2252
959e5662
BO
2253=back
2254
05c8f9ed
RS
2255=head3 OpenVOS
2256
959e5662
BO
2257=over 4
2258
2259=item *
2260
05c8f9ed
RS
2261Perl is now built with dynamic linking on OpenVOS, the minimum supported
2262version of which is now Release 17.1.0.
2263
959e5662
BO
2264=back
2265
05c8f9ed
RS
2266=head3 SunOS
2267
2268The CC workshop C++ compiler is now detected and used on systems that ship
2269without cc.
2270
2271=head1 Internal Changes
2272
2273=over 4
2274
2275=item *
2276
2277The compiled representation of formats is now stored via the C<mg_ptr> of
2278their C<PERL_MAGIC_fm>. Previously it was stored in the string buffer,
2279beyond C<SvLEN()>, the regular end of the string. C<SvCOMPILED()> and
2280C<SvCOMPILED_{on,off}()> now exist solely for compatibility for XS code.
2281The first is always 0, the other two now no-ops. (5.14.1)
2282
2283=item *
2284
2285Some global variables have been marked C<const>, members in the interpreter
2286structure have been re-ordered, and the opcodes have been re-ordered. The
2287op C<OP_AELEMFAST> has been split into C<OP_AELEMFAST> and C<OP_AELEMFAST_LEX>.
2288
2289=item *
2290
7af9a95b 2291When empting a hash of its elements (e.g., via undef(%h), or %h=()), HvARRAY
05c8f9ed
RS
2292field is no longer temporarily zeroed. Any destructors called on the freed
2293elements see the remaining elements. Thus, %h=() becomes more like
2294C<delete $h{$_} for keys %h>.
2295
2296=item *
2297
2298Boyer-Moore compiled scalars are now PVMGs, and the Boyer-Moore tables are now
2299stored via the mg_ptr of their C<PERL_MAGIC_bm>.
2300Previously they were PVGVs, with the tables stored in
2301the string buffer, beyond C<SvLEN()>. This eliminates
2302the last place where the core stores data beyond C<SvLEN()>.
2303
2304=item *
2305
2306Simplified logic in C<Perl_sv_magic()> introduces a small change of
1540de7c 2307behavior for error cases involving unknown magic types. Previously, if
05c8f9ed
RS
2308C<Perl_sv_magic()> was passed a magic type unknown to it, it would
2309
2310=over
2311
2312=item 1.
2313
2314Croak "Modification of a read-only value attempted" if read only
2315
2316=item 2.
2317
2318Return without error if the SV happened to already have this magic
2319
2320=item 3.
2321
2322otherwise croak "Don't know how to handle magic of type \\%o"
2323
2324=back
2325
2326Now it will always croak "Don't know how to handle magic of type \\%o", even
7af9a95b 2327on read-only values, or SVs which already have the unknown magic type.
05c8f9ed
RS
2328
2329=item *
2330
2331The experimental C<fetch_cop_label> function has been renamed to
2332C<cop_fetch_label>.
2333
2334=item *
2335
2336The C<cop_store_label> function has been added to the API, but is
2337experimental.
2338
2339=item *
2340
2341F<embedvar.h> has been simplified, and one level of macro indirection for
2342PL_* variables has been removed for the default (non-multiplicity)
2343configuration. PERLVAR*() macros now directly expand their arguments to
2344tokens such as C<PL_defgv>, instead of expanding to C<PL_Idefgv>, with
2345F<embedvar.h> defining a macro to map C<PL_Idefgv> to C<PL_defgv>. XS code
2346which has unwarranted chumminess with the implementation may need updating.
2347
2348=item *
2349
7af9a95b 2350An API has been added to explicitly choose whether to export XSUB
05c8f9ed
RS
2351symbols. More detail can be found in the comments for commit e64345f8.
2352
2353=item *
2354
2355The C<is_gv_magical_sv> function has been eliminated and merged with
2356C<gv_fetchpvn_flags>. It used to be called to determine whether a GV
2357should be autovivified in rvalue context. Now it has been replaced with a
2358new C<GV_ADDMG> flag (not part of the API).
2359
2360=item *
2361
2362The returned code point from the function C<utf8n_to_uvuni()>
449d5120
RS
2363when the input is malformed UTF-8, malformations are allowed, and
2364C<utf8> warnings are off is now the Unicode REPLACEMENT CHARACTER
05c8f9ed
RS
2365whenever the malformation is such that no well-defined code point can be
2366computed. Previously the returned value was essentially garbage. The
2367only malformations that have well-defined values are a zero-length
2368string (0 is the return), and overlong UTF-8 sequences.
2369
2370=item *
2371
2372Padlists are now marked C<AvREAL>; i.e., reference-counted. They have
2373always been reference-counted, but were not marked real, because F<pad.c>
2374did its own clean-up, instead of using the usual clean-up code in F<sv.c>.
2375That caused problems in thread cloning, so now the C<AvREAL> flag is on,
2376but is turned off in F<pad.c> right before the padlist is freed (after
2377F<pad.c> has done its custom freeing of the pads).
2378
2379=item *
2380
7af9a95b 2381All C files that make up the Perl core have been converted to UTF-8.
05c8f9ed
RS
2382
2383=item *
2384
2385These new functions have been added as part of the work on Unicode symbols:
2386
2387 HvNAMELEN
2388 HvNAMEUTF8
2389 HvENAMELEN
2390 HvENAMEUTF8
2391 gv_init_pv
2392 gv_init_pvn
2393 gv_init_pvsv
2394 gv_fetchmeth_pv
2395 gv_fetchmeth_pvn
2396 gv_fetchmeth_sv
2397 gv_fetchmeth_pv_autoload
2398 gv_fetchmeth_pvn_autoload
2399 gv_fetchmeth_sv_autoload
2400 gv_fetchmethod_pv_flags
2401 gv_fetchmethod_pvn_flags
2402 gv_fetchmethod_sv_flags
2403 gv_autoload_pv
2404 gv_autoload_pvn
2405 gv_autoload_sv
2406 newGVgen_flags
2407 sv_derived_from_pv
2408 sv_derived_from_pvn
2409 sv_derived_from_sv
2410 sv_does_pv
2411 sv_does_pvn
2412 sv_does_sv
2413 whichsig_pv
2414 whichsig_pvn
2415 whichsig_sv
2416 newCONSTSUB_flags
2417
2418The gv_fetchmethod_*_flags functions, like gv_fetchmethod_flags, are
2419experimental and may change in a future release.
2420
2421=item *
2422
2423The following functions were added. These are I<not> part of the API:
2424
2425 GvNAMEUTF8
2426 GvENAMELEN
2427 GvENAME_HEK
2428 CopSTASH_flags
2429 CopSTASH_flags_set
2430 PmopSTASH_flags
2431 PmopSTASH_flags_set
2432 sv_sethek
2433 HEKfARG
2434
2435There is also a C<HEKf> macro corresponding to C<SVf>, for
2436interpolating HEKs in formatted strings.
2437
2438=item *
2439
2440C<sv_catpvn_flags> takes a couple of new internal-only flags,
2441C<SV_CATBYTES> and C<SV_CATUTF8>, which tell it whether the char array to
2442be concatenated is UTF8. This allows for more efficient concatenation than
2443creating temporary SVs to pass to C<sv_catsv>.
2444
2445=item *
2446
2447For XS AUTOLOAD subs, $AUTOLOAD is set once more, as it was in 5.6.0. This
2448is in addition to setting C<SvPVX(cv)>, for compatibility with 5.8 to 5.14.
2449See L<perlguts/Autoloading with XSUBs>.
2450
2451=item *
2452
1540de7c 2453Perl now checks whether the array (the linearized isa) returned by a MRO
05c8f9ed
RS
2454plugin begins with the name of the class itself, for which the array was
2455created, instead of assuming that it does. This prevents the first element
2456from being skipped during method lookup. It also means that
2457C<mro::get_linear_isa> may return an array with one more element than the
2458MRO plugin provided [perl #94306].
2459
2460=item *
2461
2462C<PL_curstash> is now reference-counted.
2463
2464=item *
2465
2466There are now feature bundle hints in C<PL_hints> (C<$^H>) that version
2467declarations use, to avoid having to load F<feature.pm>. One setting of
2468the hint bits indicates a "custom" feature bundle, which means that the
2469entries in C<%^H> still apply. F<feature.pm> uses that.
2470
2471The C<HINT_FEATURE_MASK> macro is defined in F<perl.h> along with other
2472hints. Other macros for setting and testing features and bundles are in
2473the new F<feature.h>. C<FEATURE_IS_ENABLED> (which has moved to
2474F<feature.h>) is no longer used throughout the codebase, but more specific
2475macros, e.g., C<FEATURE_SAY_IS_ENABLED>, that are defined in F<feature.h>.
2476
2477=item *
2478
2479F<lib/feature.pm> is now a generated file, created by the new
2480F<regen/feature.pl> script, which also generates F<feature.h>.
2481
2482=item *
2483
2484Tied arrays are now always C<AvREAL>. If C<@_> or C<DB::args> is tied, it
2485is reified first, to make sure this is always the case.
2486
2487=item *
2488
2489Two new functions C<utf8_to_uvchr_buf()> and C<utf8_to_uvuni_buf()> have
2490been added. These are the same as C<utf8_to_uvchr> and
2491C<utf8_to_uvuni> (which are now deprecated), but take an extra parameter
2492that is used to guard against reading beyond the end of the input
2493string.
2494See L<perlapi/utf8_to_uvchr_buf> and L<perlapi/utf8_to_uvuni_buf>.
2495
2496=item *
2497
2498The regular expression engine now does TRIE case insensitive matches
2499under Unicode. This may change the output of C<< use re 'debug'; >>,
2500and will speed up various things.
2501
2502=item *
2503
2504There is a new C<wrap_op_checker()> function, which provides a thread-safe
2505alternative to writing to C<PL_check> directly.
2506
2507=back
2508
2509=head1 Selected Bug Fixes
2510
2511=head2 Array and hash
2512
2513=over
2514
2515=item *
2516
2517A bug has been fixed that would cause a "Use of freed value in iteration"
2518error if the next two hash elements that would be iterated over are
2519deleted [perl #85026]. (5.14.1)
2520
2521=item *
2522
7af9a95b 2523Deleting the current hash iterator (the hash element that would be returned
05c8f9ed
RS
2524by the next call to C<each>) in void context used not to free it
2525[perl #85026].
2526
2527=item *
2528
2529Deletion of methods via C<delete $Class::{method}> syntax used to update
2530method caches if called in void context, but not scalar or list context.
2531
2532=item *
2533
2534When hash elements are deleted in void context, the internal hash entry is
2535now freed before the value is freed, to prevent destructors called by that
2536latter freeing from seeing the hash in an inconsistent state. It was
2537possible to cause double-frees if the destructor freed the hash itself
2538[perl #100340].
2539
2540=item *
2541
1540de7c 2542A C<keys> optimization in Perl 5.12.0 to make it faster on empty hashes
05c8f9ed
RS
2543caused C<each> not to reset the iterator if called after the last element
2544was deleted.
2545
2546=item *
2547
2548Freeing deeply nested hashes no longer crashes [perl #44225].
2549
2550=item *
2551
2552It is possible from XS code to create hashes with elements that have no
2553values. The hash element and slice operators used to crash
2554when handling these in lvalue context. They now
2555produce a "Modification of non-creatable hash value attempted" error
2556message.
2557
2558=item *
2559
2560If list assignment to a hash or array triggered destructors that freed the
2561hash or array itself, a crash would ensue. This is no longer the case
2562[perl #107440].
2563
2564=item *
2565
1540de7c 2566It used to be possible to free the typeglob of a localized array or hash
05c8f9ed
RS
2567(e.g., C<local @{"x"}; delete $::{x}>), resulting in a crash on scope exit.
2568
2569=item *
2570
2571Some core bugs affecting L<Hash::Util> have been fixed: locking a hash
7af9a95b 2572element that is a glob copy no longer causes the next assignment to it to
006b5090
FC
2573corrupt the glob (5.14.2), and unlocking a hash element that holds a
2574copy-on-write scalar no longer causes modifications to that scalar to
2575modify other scalars that were sharing the same string buffer.
05c8f9ed
RS
2576
2577=back
2578
2579=head2 C API fixes
2580
2581=over
2582
2583=item *
2584
2585The C<newHVhv> XS function now works on tied hashes, instead of crashing or
2586returning an empty hash.
2587
2588=item *
2589
2590The C<SvIsCOW> C macro now returns false for read-only copies of typeglobs,
2591such as those created by:
2592
2593 $hash{elem} = *foo;
2594 Hash::Util::lock_value %hash, 'elem';
2595
2596It used to return true.
2597
2598=item *
2599
2600The C<SvPVutf8> C function no longer tries to modify its argument,
2601resulting in errors [perl #108994].
2602
2603=item *
2604
2605C<SvPVutf8> now works properly with magical variables.
2606
2607=item *
2608
2609C<SvPVbyte> now works properly non-PVs.
2610
2611=item *
2612
2613When presented with malformed UTF-8 input, the XS-callable functions
2614C<is_utf8_string()>, C<is_utf8_string_loc()>, and
2615C<is_utf8_string_loclen()> could read beyond the end of the input
2616string by up to 12 bytes. This no longer happens. [perl #32080].
2617However, currently, C<is_utf8_char()> still has this defect, see
2618L</is_utf8_char()> above.
2619
2620=item *
2621
7af9a95b 2622The C-level C<pregcomp> function could become confused about whether the
05c8f9ed
RS
2623pattern was in UTF8 if the pattern was an overloaded, tied, or otherwise
2624magical scalar [perl #101940].
2625
2626=back
2627
2628=head2 Compile-time hints
2629
2630=over
2631
2632=item *
2633
2634Tying C<%^H> no longer causes perl to crash or ignore the contents of
2635C<%^H> when entering a compilation scope [perl #106282].
2636
2637=item *
2638
2639C<eval $string> and C<require> used not to
1540de7c 2640localize C<%^H> during compilation if it
05c8f9ed
RS
2641was empty at the time the C<eval> call itself was compiled. This could
2642lead to scary side effects, like C<use re "/m"> enabling other flags that
2643the surrounding code was trying to enable for its caller [perl #68750].
2644
2645=item *
2646
1540de7c 2647C<eval $string> and C<require> no longer localize hints (C<$^H> and C<%^H>)
05c8f9ed
RS
2648at run time, but only during compilation of the $string or required file.
2649This makes C<BEGIN { $^H{foo}=7 }> equivalent to
2650C<BEGIN { eval '$^H{foo}=7' }> [perl #70151].
2651
2652=item *
2653
2654Creating a BEGIN block from XS code (via C<newXS> or C<newATTRSUB>) would,
2655on completion, make the hints of the current compiling code the current
2656hints. This could cause warnings to occur in a non-warning scope.
2657
2658=back
2659
2660=head2 Copy-on-write scalars
2661
2662Copy-on-write or shared hash key scalars
2663were introduced in 5.8.0, but most Perl code
2664did not encounter them (they were used mostly internally). Perl
26655.10.0 extended them, such that assigning C<__PACKAGE__> or a
2666hash key to a scalar would make it copy-on-write. Several parts
2667of Perl were not updated to account for them, but have now been fixed.
2668
2669=over
2670
2671=item *
2672
2673C<utf8::decode> had a nasty bug that would modify copy-on-write scalars'
2674string buffers in place (i.e., skipping the copy). This could result in
006b5090 2675hashes having two elements with the same key [perl #91834]. (5.14.2)
05c8f9ed
RS
2676
2677=item *
2678
2679Lvalue subroutines were not allowing COW scalars to be returned. This was
2680fixed for lvalue scalar context in Perl 5.12.3 and 5.14.0, but list context
2681was not fixed until this release.
2682
2683=item *
2684
2685Elements of restricted hashes (see the L<fields> pragma) containing
2686copy-on-write values couldn't be deleted, nor could such hashes be cleared
006b5090 2687(C<%hash = ()>). (5.14.2)
05c8f9ed
RS
2688
2689=item *
2690
1540de7c 2691Localizing a tied variable used to make it read-only if it contained a
006b5090 2692copy-on-write string. (5.14.2)
05c8f9ed
RS
2693
2694=item *
2695
2696Assigning a copy-on-write string to a stash
2697element no longer causes a double free. Regardless of this change, the
2698results of such assignments are still undefined.
2699
2700=item *
2701
2702Assigning a copy-on-write string to a tied variable no longer stops that
2703variable from being tied if it happens to be a PVMG or PVLV internally.
2704
2705=item *
2706
2707Doing a substitution on a tied variable returning a copy-on-write
2708scalar used to cause an assertion failure or an "Attempt to free
2709nonexistent shared string" warning.
2710
2711=item *
2712
2713This one is a regression from 5.12: In 5.14.0, the bitwise assignment
2714operators C<|=>, C<^=> and C<&=> started leaving the left-hand side
2715undefined if it happened to be a copy-on-write string [perl #108480].
2716
2717=item *
2718
2719L<Storable>, L<Devel::Peek> and L<PerlIO::scalar> had similar problems.
2720See L</Updated Modules and Pragmata>, above.
2721
2722=back
2723
2724=head2 The debugger
2725
2726=over
2727
2728=item *
2729
7af9a95b 2730F<dumpvar.pl>, and therefore the C<x> command in the debugger, have been
05c8f9ed
RS
2731fixed to handle objects blessed into classes whose names contain "=". The
2732contents of such objects used not to be dumped [perl #101814].
2733
2734=item *
2735
2736The "R" command for restarting a debugger session has been fixed to work on
2737Windows, or any other system lacking a C<POSIX::_SC_OPEN_MAX> constant
2738[perl #87740].
2739
2740=item *
2741
2742The C<#line 42 foo> directive used not to update the arrays of lines used
2743by the debugger if it occurred in a string eval. This was partially fixed
7af9a95b 2744in 5.14, but it worked only for a single C<#line 42 foo> in each eval. Now
05c8f9ed
RS
2745it works for multiple.
2746
2747=item *
2748
2749When subroutine calls are intercepted by the debugger, the name of the
2750subroutine or a reference to it is stored in C<$DB::sub>, for the debugger
7af9a95b 2751to access. Sometimes (such as C<$foo = *bar; undef *bar; &$foo>)
05c8f9ed
RS
2752C<$DB::sub> would be set to a name that could not be used to find the
2753subroutine, and so the debugger's attempt to call it would fail. Now the
2754check to see whether a reference is needed is more robust, so those
2755problems should not happen anymore [rt.cpan.org #69862].
2756
2757=item *
2758
2759Every subroutine has a filename associated with it that the debugger uses.
2760The one associated with constant subroutines used to be misallocated when
2761cloned under threads. Consequently, debugging threaded applications could
2762result in memory corruption [perl #96126].
2763
2764=back
2765
2766=head2 Dereferencing operators
2767
2768=over
2769
2770=item *
2771
2772C<defined(${"..."})>, C<defined(*{"..."})>, etc., used to
2773return true for most, but not all built-in variables, if
2774they had not been used yet. This bug affected C<${^GLOBAL_PHASE}> and
2775C<${^UTF8CACHE}>, among others. It also used to return false if the
2776package name was given as well (C<${"::!"}>) [perl #97978, #97492].
2777
2778=item *
2779
2780Perl 5.10.0 introduced a similar bug: C<defined(*{"foo"})> where "foo"
2781represents the name of a built-in global variable used to return false if
2782the variable had never been used before, but only on the I<first> call.
2783This, too, has been fixed.
2784
2785=item *
2786
2787Since 5.6.0, C<*{ ... }> has been inconsistent in how it treats undefined
2788values. It would die in strict mode or lvalue context for most undefined
2789values, but would be treated as the empty string (with a warning) for the
2790specific scalar return by C<undef()> (C<&PL_sv_undef> internally). This
2791has been corrected. C<undef()> is now treated like other undefined
2792scalars, as in Perl 5.005.
2793
2794=back
2795
2796=head2 Filehandle, last-accessed
2797
2798Perl has an internal variable that stores the last filehandle to be
2799accessed. It is used by C<$.> and by C<tell> and C<eof> without
2800arguments.
2801
2802=over
2803
2804=item *
2805
2806It used to be possible to set this internal variable to a glob copy and
2807then modify that glob copy to be something other than a glob, and still
2808have the last-accessed filehandle associated with the variable after
2809assigning a glob to it again:
2810
2811 my $foo = *STDOUT; # $foo is a glob copy
2812 <$foo>; # $foo is now the last-accessed handle
2813 $foo = 3; # no longer a glob
2814 $foo = *STDERR; # still the last-accessed handle
2815
2816Now the C<$foo = 3> assignment unsets that internal variable, so there
2817is no last-accessed filehandle, just as if C<< <$foo> >> had never
2818happened.
2819
2820This also prevents some unrelated handle from becoming the last-accessed
2821handle if $foo falls out of scope and the same internal SV gets used for
2822another handle [perl #97988].
2823
2824=item *
2825
2826A regression in 5.14 caused these statements not to set that internal
2827variable:
2828
2829 my $fh = *STDOUT;
2830 tell $fh;
2831 eof $fh;
2832 seek $fh, 0,0;
2833 tell *$fh;
2834 eof *$fh;
2835 seek *$fh, 0,0;
2836 readline *$fh;
2837
2838This is now fixed, but C<tell *{ *$fh }> still has the problem, and it
2839is not clear how to fix it [perl #106536].
2840
2841=back
2842
2843=head2 Filetests and C<stat>
2844
2845The term "filetests" refers to the operators that consist of a hyphen
2846followed by a single letter: C<-r>, C<-x>, C<-M>, etc. The term "stacked"
2847when applied to filetests means followed by another filetest operator
2848sharing the same operand, as in C<-r -x -w $fooo>.
2849
2850=over
2851
2852=item *
2853
2854C<stat> produces more consistent warnings. It no longer warns for "_"
2855[perl #71002] and no longer skips the warning at times for other unopened
2856handles. It no longer warns about an unopened handle when the operating
2857system's C<fstat> function fails.
2858
2859=item *
2860
2861C<stat> would sometimes return negative numbers for large inode numbers,
2862because it was using the wrong internal C type. [perl #84590]
2863
2864=item *
2865
2866C<lstat> is documented to fall back to C<stat> (with a warning) when given
2867a filehandle. When passed an IO reference, it was actually doing the
2868equivalent of S<C<stat _>> and ignoring the handle.
2869
2870=item *
2871
2872C<-T _> with no preceding C<stat> used to produce a
2873confusing "uninitialized" warning, even though there
2874is no visible uninitialized value to speak of.
2875
2876=item *
2877
2878C<-T>, C<-B>, C<-l> and C<-t> now work
2879when stacked with other filetest operators
2880[perl #77388].
2881
2882=item *
2883
2884In 5.14.0, filetest ops (C<-r>, C<-x>, etc.) started calling FETCH on a
2885tied argument belonging to the previous argument to a list operator, if
2886called with a bareword argument or no argument at all. This has been
2887fixed, so C<push @foo, $tied, -r> no longer calls FETCH on C<$tied>.
2888
2889=item *
2890
2891In Perl 5.6, C<-l> followed by anything other than a bareword would treat
2892its argument as a file name. That was changed in 5.8 for glob references
2893(C<\*foo>), but not for globs themselves (C<*foo>). C<-l> started
2894returning C<undef> for glob references without setting the last
2895stat buffer that the "_" handle uses, but only if warnings
2896were turned on. With warnings off, it was the same as 5.6.
2897In other words, it was simply buggy and inconsistent. Now the 5.6
1540de7c 2898behavior has been restored.
05c8f9ed
RS
2899
2900=item *
2901
2902C<-l> followed by a bareword no longer "eats" the previous argument to
2903the list operator in whose argument list it resides. Hence,
2904C<print "bar", -l foo> now actually prints "bar", because C<-l>
2905on longer eats it.
2906
2907=item *
2908
2909Perl keeps several internal variables to keep track of the last stat
2910buffer, from which file(handle) it originated, what type it was, and
2911whether the last stat succeeded.
2912
2913There were various cases where these could get out of synch, resulting in
1540de7c 2914inconsistent or erratic behavior in edge cases (every mention of C<-T>
05c8f9ed
RS
2915applies to C<-B> as well):
2916
2917=over
2918
2919=item *
2920
2921C<-T I<HANDLE>>, even though it does a C<stat>, was not resetting the last
2922stat type, so an C<lstat _> following it would merrily return the wrong
2923results. Also, it was not setting the success status.
2924
2925=item *
2926
2927Freeing the handle last used by C<stat> or a filetest could result in
2928S<C<-T _>> using an unrelated handle.
2929
2930=item *
2931
2932C<stat> with an IO reference would not reset the stat type or record the
2933filehandle for S<C<-T _>> to use.
2934
2935=item *
2936
2937Fatal warnings could cause the stat buffer not to be reset
2938for a filetest operator on an unopened filehandle or C<-l> on any handle.
2939Fatal warnings also stopped C<-T> from setting C<$!>.
2940
2941=item *
2942
2943When the last stat was on an unreadable file, C<-T _> is supposed to
2944return C<undef>, leaving the last stat buffer unchanged. But it was
2945setting the stat type, causing C<lstat _> to stop working.
2946
2947=item *
2948
2949C<-T I<FILENAME>> was not resetting the internal stat buffers for
2950unreadable files.
2951
2952=back
2953
2954These have all been fixed.
2955
2956=back
2957
2958=head2 Formats
2959
2960=over
2961
2962=item *
2963
7af9a95b 2964Several edge cases have been fixed with formats and C<formline>;
05c8f9ed
RS
2965in particular, where the format itself is potentially variable (such as
2966with ties and overloading), and where the format and data differ in their
2967encoding. In both these cases, it used to possible for the output to be
2968corrupted [perl #91032].
2969
2970=item *
2971
2972C<formline> no longer converts its argument into a string in-place. So
2973passing a reference to C<formline> no longer destroys the reference
2974[perl #79532].
2975
2976=item *
2977
2978Assignment to C<$^A> (the format output accumulator) now recalculates
2979the number of lines output.
2980
2981=back
2982
2983=head2 C<given> and C<when>
2984
2985=over
2986
2987=item *
2988
2989C<given> was not scoping its implicit $_ properly, resulting in memory
2990leaks or "Variable is not available" warnings [perl #94682].
2991
2992=item *
2993
2994C<given> was not calling set-magic on the implicit lexical C<$_> that it
2995uses. This meant, for example, that C<pos> would be remembered from one
2996execution of the same C<given> block to the next, even if the input were a
2997different variable [perl #84526].
2998
2999=item *
3000
3001C<when> blocks are now capable of returning variables declared inside the
3002enclosing C<given> block [perl #93548].
3003
3004=back
3005
3006=head2 The C<glob> operator
3007
3008=over
3009
3010=item *
3011
3012On OSes other than VMS, Perl's C<glob> operator (and the C<< <...> >> form)
3013use L<File::Glob> underneath. L<File::Glob> splits the pattern into words,
3014before feeding each word to its C<bsd_glob> function.
3015
3016There were several inconsistencies in the way the split was done. Now
3017quotation marks (' and ") are always treated as shell-style word delimiters
3018(that allow whitespace as part of a word) and backslashes are always
3019preserved, unless they exist to escape quotation marks. Before, those
3020would only sometimes be the case, depending on whether the pattern
3021contained whitespace. Also, escaped whitespace at the end of the pattern
3022is no longer stripped [perl #40470].
3023
3024=item *
3025
3026C<CORE::glob> now works as a way to call the default globbing function. It
3027used to respect overrides, despite the C<CORE::> prefix.
3028
3029=item *
3030
3031Under miniperl (used to configure modules when perl itself is built),
3032C<glob> now clears %ENV before calling csh, since the latter croaks on some
7af9a95b 3033systems if it does not like the contents of the LS_COLORS environment
05c8f9ed
RS
3034variable [perl #98662].
3035
3036=back
3037
3038=head2 Lvalue subroutines
3039
3040=over
3041
3042=item *
3043
3044Explicit return now returns the actual argument passed to return, instead
3045of copying it [perl #72724, #72706].
3046
3047=item *
3048
3049Lvalue subroutines used to enforce lvalue syntax (i.e., whatever can go on
3050the left-hand side of C<=>) for the last statement and the arguments to
3051return. Since lvalue subroutines are not always called in lvalue context,
3052this restriction has been lifted.
3053
3054=item *
3055
7af9a95b 3056Lvalue subroutines are less restrictive about what values can be returned.
05c8f9ed
RS
3057It used to croak on values returned by C<shift> and C<delete> and from
3058other subroutines, but no longer does so [perl #71172].
3059
3060=item *
3061
3062Empty lvalue subroutines (C<sub :lvalue {}>) used to return C<@_> in list
7af9a95b 3063context. All subroutines used to do this, but regular subs were fixed in
05c8f9ed
RS
3064Perl 5.8.2. Now lvalue subroutines have been likewise fixed.
3065
3066=item *
3067
3068Autovivification now works on values returned from lvalue subroutines
3069[perl #7946], as does returning C<keys> in lvalue context.
3070
3071=item *
3072
3073Lvalue subroutines used to copy their return values in rvalue context. Not
3074only was this a waste of CPU cycles, but it also caused bugs. A C<($)>
3075prototype would cause an lvalue sub to copy its return value [perl #51408],
3076and C<while(lvalue_sub() =~ m/.../g) { ... }> would loop endlessly
3077[perl #78680].
3078
3079=item *
3080
3081When called in potential lvalue context
3082(e.g., subroutine arguments or a list
3083passed to C<for>), lvalue subroutines used to copy
3084any read-only value that was returned. E.g., C< sub :lvalue { $] } >
3085would not return C<$]>, but a copy of it.
3086
3087=item *
3088
3089When called in potential lvalue context, an lvalue subroutine returning
3090arrays or hashes used to bind the arrays or hashes to scalar variables,
3091resulting in bugs. This was fixed in 5.14.0 if an array were the first
3092thing returned from the subroutine (but not for C<$scalar, @array> or
3093hashes being returned). Now a more general fix has been applied
3094[perl #23790].
3095
3096=item *
3097
3098Method calls whose arguments were all surrounded with C<my()> or C<our()>
3099(as in C<< $object->method(my($a,$b)) >>) used to force lvalue context on
3100the subroutine. This would prevent lvalue methods from returning certain
3101values.
3102
3103=item *
3104
3105Lvalue sub calls that are not determined to be such at compile time
3106(C<&$name> or &{"name"}) are no longer exempt from strict refs if they
3107occur in the last statement of an lvalue subroutine [perl #102486].
3108
3109=item *
3110
3111Sub calls whose subs are not visible at compile time, if
3112they occurred in the last statement of an lvalue subroutine,
3113would reject non-lvalue subroutines and die with "Can't modify non-lvalue
3114subroutine call" [perl #102486].
3115
3116Non-lvalue sub calls whose subs I<are> visible at compile time exhibited
3117the opposite bug. If the call occurred in the last statement of an lvalue
3118subroutine, there would be no error when the lvalue sub was called in
3119lvalue context. Perl would blindly assign to the temporary value returned
3120by the non-lvalue subroutine.
3121
3122=item *
3123
3124C<AUTOLOAD> routines used to take precedence over the actual sub being
3125called (i.e., when autoloading wasn't needed), for sub calls in lvalue or
3126potential lvalue context, if the subroutine was not visible at compile
3127time.
3128
3129=item *
3130
3131Applying the C<:lvalue> attribute to an XSUB or to an aliased subroutine
3132stub with C<< sub foo :lvalue; >> syntax stopped working in Perl 5.12.
3133This has been fixed.
3134
3135=item *
3136
3137Applying the :lvalue attribute to subroutine that is already defined does
3138not work properly, as the attribute changes the way the sub is compiled.
3139Hence, Perl 5.12 began warning when an attempt is made to apply the
3140attribute to an already defined sub. In such cases, the attribute is
3141discarded.
3142
3143But the change in 5.12 missed the case where custom attributes are also
3144present: that case still silently and ineffectively applied the attribute.
3145That omission has now been corrected. C<sub foo :lvalue :Whatever> (when
3146C<foo> is already defined) now warns about the :lvalue attribute, and does
3147not apply it.
3148
3149=item *
3150
3151A bug affecting lvalue context propagation through nested lvalue subroutine
3152calls has been fixed. Previously, returning a value in nested rvalue
3153context would be treated as lvalue context by the inner subroutine call,
3154resulting in some values (such as read-only values) being rejected.
3155
3156=back
3157
3158=head2 Overloading
3159
3160=over
3161
3162=item *
3163
3164Arithmetic assignment (C<$left += $right>) involving overloaded objects
3165that rely on the 'nomethod' override no longer segfault when the left
3166operand is not overloaded.
3167
3168=item *
3169
3170Errors that occur when methods cannot be found during overloading now
3171mention the correct package name, as they did in 5.8.x, instead of
3172erroneously mentioning the "overload" package, as they have since 5.10.0.
3173
3174=item *
3175
3176Undefining C<%overload::> no longer causes a crash.
3177
3178=back
3179
3180=head2 Prototypes of built-in keywords
3181
3182=over
3183
3184=item *
3185
3186The C<prototype> function no longer dies for the C<__FILE__>, C<__LINE__>
3187and C<__PACKAGE__> directives. It now returns an empty-string prototype
3188for them, because they are syntactically indistinguishable from nullary
3189functions like C<time>.
3190
3191=item *
3192
3193C<prototype> now returns C<undef> for all overridable infix operators,
3194such as C<eq>, which are not callable in any way resembling functions.
3195It used to return incorrect prototypes for some and die for others
3196[perl #94984].
3197
3198=item *
3199
3200The prototypes of several built-in functions--C<getprotobynumber>, C<lock>,
3201C<not> and C<select>--have been corrected, or at least are now closer to
3202reality than before.
3203
3204=back
3205
3206=head2 Regular expressions
3207
3208=for comment Is it possible to merge some of these items?
3209
3210=over 4
3211
3212=item *
3213
3214C</[[:ascii:]]/> and C</[[:blank:]]/> now use locale rules under
3215C<use locale> when the platform supports that. Previously, they used
3216the platform's native character set.
3217
3218=item *
3219
3220C<m/[[:ascii:]]/i> and C</\p{ASCII}/i> now match identically (when not
3221under a differing locale). This fixes a regression introduced in 5.14
3222in which the first expression could match characters outside of ASCII,
3223such as the KELVIN SIGN.
3224
3225=item *
3226
3227C</.*/g> would sometimes refuse to match at the end of a string that ends
3228with "\n". This has been fixed [perl #109206].
3229
3230=item *
3231
3232Starting with 5.12.0, Perl used to get its internal bookkeeping muddled up
3233after assigning C<${ qr// }> to a hash element and locking it with
7af9a95b 3234L<Hash::Util>. This could result in double frees, crashes, or erratic
1540de7c 3235behavior.
05c8f9ed
RS
3236
3237=item *
3238
3239The new (in 5.14.0) regular expression modifier C</a> when repeated like
3240C</aa> forbids the characters outside the ASCII range that match
3241characters inside that range from matching under C</i>. This did not
3242work under some circumstances, all involving alternation, such as:
3243
3244 "\N{KELVIN SIGN}" =~ /k|foo/iaa;
3245
3246succeeded inappropriately. This is now fixed.
3247
3248=item *
3249
32505.14.0 introduced some memory leaks in regular expression character
3251classes such as C<[\w\s]>, which have now been fixed. (5.14.1)
3252
3253=item *
3254
3255An edge case in regular expression matching could potentially loop.
3256This happened only under C</i> in bracketed character classes that have
3257characters with multi-character folds, and the target string to match
3258against includes the first portion of the fold, followed by another
3259character that has a multi-character fold that begins with the remaining
3260portion of the fold, plus some more.
3261
3262 "s\N{U+DF}" =~ /[\x{DF}foo]/i
3263
3264is one such case. C<\xDF> folds to C<"ss">. (5.14.1)
3265
3266=item *
3267
3268A few characters in regular expression pattern matches did not
3269match correctly in some circumstances, all involving C</i>. The
3270affected characters are:
3271COMBINING GREEK YPOGEGRAMMENI,
3272GREEK CAPITAL LETTER IOTA,
3273GREEK CAPITAL LETTER UPSILON,
3274GREEK PROSGEGRAMMENI,
3275GREEK SMALL LETTER IOTA WITH DIALYTIKA AND OXIA,
3276GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS,
3277GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND OXIA,
3278GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS,
3279LATIN SMALL LETTER LONG S,
3280LATIN SMALL LIGATURE LONG S T,
3281and
3282LATIN SMALL LIGATURE ST.
3283
3284=item *
3285
3286A memory leak regression in regular expression compilation
3287under threading has been fixed.
3288
3289=item *
3290
3620abf9 3291A regression introduced in 5.14.0 has
05c8f9ed
RS
3292been fixed. This involved an inverted
3293bracketed character class in a regular expression that consisted solely
3294of a Unicode property. That property wasn't getting inverted outside the
3295Latin1 range.
3296
3297=item *
3298
7af9a95b 3299Three problematic Unicode characters now work better in regex pattern matching under C</i>.
05c8f9ed
RS
3300
3301In the past, three Unicode characters:
3302LATIN SMALL LETTER SHARP S,
3303GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS,
3304and
3305GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS,
3306along with the sequences that they fold to
7af9a95b 3307(including "ss" for LATIN SMALL LETTER SHARP S),
05c8f9ed
RS
3308did not properly match under C</i>. 5.14.0 fixed some of these cases,
3309but introduced others, including a panic when one of the characters or
3310sequences was used in the C<(?(DEFINE)> regular expression predicate.
3311The known bugs that were introduced in 5.14 have now been fixed; as well
7af9a95b 3312as some other edge cases that have never worked until now. These all
05c8f9ed
RS
3313involve using the characters and sequences outside bracketed character
3314classes under C</i>. This closes [perl #98546].
3315
3316There remain known problems when using certain characters with
3317multi-character folds inside bracketed character classes, including such
3318constructs as C<qr/[\N{LATIN SMALL LETTER SHARP}a-z]/i>. These
3319remaining bugs are addressed in [perl #89774].
3320
3321=item *
3322
3323RT #78266: The regex engine has been leaking memory when accessing
3324named captures that weren't matched as part of a regex ever since 5.10
7af9a95b 3325when they were introduced; e.g., this would consume over a hundred MB of
05c8f9ed
RS
3326memory:
3327
3328 for (1..10_000_000) {
3329 if ("foo" =~ /(foo|(?<capture>bar))?/) {
3330 my $capture = $+{capture}
3331 }
3332 }
3333 system "ps -o rss $$"'
3334
3335=item *
3336
3337In 5.14, C</[[:lower:]]/i> and C</[[:upper:]]/i> no longer matched the
3338opposite case. This has been fixed [perl #101970].
3339
3340=item *
3341
3342A regular expression match with an overloaded object on the right-hand side
7af9a95b 3343would sometimes stringify the object too many times.
05c8f9ed
RS
3344
3345=item *
3346
3347A regression has been fixed that was introduced in 5.14, in C</i>
3348regular expression matching, in which a match improperly fails if the
3349pattern is in UTF-8, the target string is not, and a Latin-1 character
3350precedes a character in the string that should match the pattern.
3351[perl #101710]
3352
3353=item *
3354
3355In case-insensitive regular expression pattern matching, no longer on
7af9a95b 3356UTF-8 encoded strings does the scan for the start of match look only at
05c8f9ed
RS
3357the first possible position. This caused matches such as
3358C<"f\x{FB00}" =~ /ff/i> to fail.
3359
3360=item *
3361
1540de7c 3362The regexp optimizer no longer crashes on debugging builds when merging
05c8f9ed
RS
3363fixed-string nodes with inconvenient contents.
3364
3365=item *
3366
3367A panic involving the combination of the regular expression modifiers
3368C</aa> and the C<\b> escape sequence introduced in 5.14.0 has been
006b5090 3369fixed [perl #95964]. (5.14.2)
05c8f9ed
RS
3370
3371=item *
3372
3373The combination of the regular expression modifiers C</aa> and the C<\b>
3374and C<\B> escape sequences did not work properly on UTF-8 encoded
3375strings. All non-ASCII characters under C</aa> should be treated as
3376non-word characters, but what was happening was that Unicode rules were
3377used to determine wordness/non-wordness for non-ASCII characters. This
3378is now fixed [perl #95968].
3379
3380=item *
3381
3382C<< (?foo: ...) >> no longer loses passed in character set.
3383
3384=item *
3385
1540de7c 3386The trie optimization used to have problems with alternations containing
05c8f9ed
RS
3387an empty C<(?:)>, causing C<< "x" =~ /\A(?>(?:(?:)A|B|C?x))\z/ >> not to
3388match, whereas it should [perl #111842].
3389
3390=item *
3391
3392Use of lexical (C<my>) variables in code blocks embedded in regular
3393expressions will no longer result in memory corruption or crashes.
3394
3395Nevertheless, these code blocks are still experimental, as there are still
3396problems with the wrong variables being closed over (in loops for instance)
3397and with abnormal exiting (e.g., C<die>) causing memory corruption.
3398
3399=item *
3400
3401The C<\h>, C<\H>, C<\v> and C<\V> regular expression metacharacters used to
7af9a95b 3402cause a panic error message when trying to match at the end of the
05c8f9ed
RS
3403string [perl #96354].
3404
3405=item *
3406
3407The abbreviations for four C1 control characters C<MW> C<PM>, C<RI>, and
3408C<ST> were previously unrecognized by C<\N{}>, vianame(), and
3409string_vianame().
3410
3411=item *
3412
3413Mentioning a variable named "&" other than C<$&> (i.e., C<@&> or C<%&>) no
3414longer stops C<$&> from working. The same applies to variables named "'"
3415and "`" [perl #24237].
3416
3417=item *
3418
3419Creating a C<UNIVERSAL::AUTOLOAD> sub no longer stops C<%+>, C<%-> and
3420C<%!> from working some of the time [perl #105024].
3421
3422=back
3423
3424=head2 Smartmatching
3425
3426=over
3427
3428=item *
3429
3430C<~~> now correctly handles the precedence of Any~~Object, and is not tricked
3431by an overloaded object on the left-hand side.
3432
3433=item *
3434
3435In Perl 5.14.0, C<$tainted ~~ @array> stopped working properly. Sometimes
3436it would erroneously fail (when C<$tainted> contained a string that occurs
3437in the array I<after> the first element) or erroneously succeed (when
3438C<undef> occurred after the first element) [perl #93590].
3439
3440=back
3441
3442=head2 The C<sort> operator
3443
3444=over
3445
3446=item *
3447
3448C<sort> was not treating C<sub {}> and C<sub {()}> as equivalent when
3449such a sub was provided as the comparison routine. It used to croak on
3450C<sub {()}>.
3451
3452=item *
3453
3454C<sort> now works once more with custom sort routines that are XSUBs. It
3455stopped working in 5.10.0.
3456
3457=item *
3458
3459C<sort> with a constant for a custom sort routine, although it produces
3460unsorted results, no longer crashes. It started crashing in 5.10.0.
3461
3462=item *
3463
3464Warnings emitted by C<sort> when a custom comparison routine returns a
3465non-numeric value now contain "in sort" and show the line number of the
3466C<sort> operator, rather than the last line of the comparison routine. The
7af9a95b 3467warnings also now occur only if warnings are enabled in the scope where
05c8f9ed
RS
3468C<sort> occurs. Previously the warnings would occur if enabled in the
3469comparison routine's scope.
3470
3471=item *
3472
1540de7c 3473C<< sort { $a <=> $b } >>, which is optimized internally, now produces
05c8f9ed
RS
3474"uninitialized" warnings for NaNs (not-a-number values), since C<< <=> >>
3475returns C<undef> for those. This brings it in line with
3476S<C<< sort { 1; $a <=> $b } >>> and other more complex cases, which are not
1540de7c 3477optimized [perl #94390].
05c8f9ed
RS
3478
3479=back
3480
3481=head2 The C<substr> operator
3482
3483=over
3484
3485=item *
88c5c971 3486
05c8f9ed
RS
3487Tied (and otherwise magical) variables are no longer exempt from the
3488"Attempt to use reference as lvalue in substr" warning.
8f12b018 3489
05c8f9ed 3490=item *
8f12b018 3491
05c8f9ed 3492That warning now occurs when the returned lvalue is assigned to, not
7af9a95b
TC
3493when C<substr> itself is called. This makes a difference only if the
3494return value of C<substr> is referenced and later assigned to.
26afcec5 3495
05c8f9ed 3496=item *
8f12b018 3497
05c8f9ed
RS
3498Passing a substring of a read-only value or a typeglob to a function
3499(potential lvalue context) no longer causes an immediate "Can't coerce"
7af9a95b
TC
3500or "Modification of a read-only value" error. That error occurs only
3501if the passed value is assigned to.
d7fbd56d 3502
05c8f9ed 3503The same thing happens with the "substr outside of string" error. If
7af9a95b 3504the lvalue is only read from, not written to, it is now just a warning, as
05c8f9ed 3505with rvalue C<substr>.
d7fbd56d 3506
05c8f9ed 3507=item *
d5dc7001 3508
05c8f9ed
RS
3509C<substr> assignments no longer call FETCH twice if the first argument
3510is a tied variable, just once.
d5dc7001 3511
05c8f9ed 3512=back
711a3903 3513
05c8f9ed 3514=head2 Support for embedded nulls
977c1d31 3515
05c8f9ed
RS
3516Some parts of Perl did not work correctly with nulls (C<chr 0>) embedded in
3517strings. That meant that, for instance, C<< $m = "a\0b"; foo->$m >> would
3518call the "a" method, instead of the actual method name contained in $m.
3519These parts of perl have been fixed to support nulls:
27f00e3d 3520
05c8f9ed 3521=over
6ba817f3 3522
05c8f9ed 3523=item *
2630d42b 3524
05c8f9ed 3525Method names
7620cb10 3526
05c8f9ed 3527=item *
d7c042c9 3528
05c8f9ed 3529Typeglob names (including filehandle and subroutine names)
d7c042c9 3530
05c8f9ed 3531=item *
d7c042c9 3532
05c8f9ed 3533Package names, including the return value of C<ref()>
985213f2 3534
05c8f9ed 3535=item *
977c1d31 3536
05c8f9ed 3537Typeglob elements (C<*foo{"THING\0stuff"}>)
977c1d31 3538
05c8f9ed 3539=item *
985213f2 3540
05c8f9ed 3541Signal names
985213f2 3542
204b72a4 3543=item *
2e2b2571 3544
05c8f9ed
RS
3545Various warnings and error messages that mention variable names or values,
3546methods, etc.
2e2b2571 3547
204b72a4 3548=back
b240fc0f 3549
05c8f9ed
RS
3550One side effect of these changes is that blessing into "\0" no longer
3551causes C<ref()> to return false.
27f00e3d 3552
05c8f9ed 3553=head2 Threading bugs
27f00e3d 3554
05c8f9ed 3555=over
cadced9f 3556
05c8f9ed 3557=item *
cadced9f 3558
05c8f9ed
RS
3559Typeglobs returned from threads are no longer cloned if the parent thread
3560already has a glob with the same name. This means that returned
3561subroutines will now assign to the right package variables [perl #107366].
3562
3563=item *
3564
3565Some cases of threads crashing due to memory allocation during cloning have
3566been fixed [perl #90006].
632c5d30
NC
3567
3568=item *
3569
05c8f9ed 3570Thread joining would sometimes emit "Attempt to free unreferenced scalar"
7af9a95b 3571warnings if C<caller> had been used from the C<DB> package before thread
05c8f9ed
RS
3572creation [perl #98092].
3573
3574=item *
3575
3576Locking a subroutine (via C<lock &sub>) is no longer a compile-time error
3577for regular subs. For lvalue subroutines, it no longer tries to return the
3578sub as a scalar, resulting in strange side effects like C<ref \$_>
3579returning "CODE" in some instances.
3580
3581C<lock &sub> is now a run-time error if L<threads::shared> is loaded (a
3582no-op otherwise), but that may be rectified in a future version.
95ce428c 3583
2630d42b 3584=back
95ce428c 3585
05c8f9ed 3586=head2 Tied variables
4e6e9b23 3587
05c8f9ed 3588=over
4e6e9b23 3589
c88a046d 3590=item *
4e6e9b23 3591
05c8f9ed
RS
3592Various cases in which FETCH was being ignored or called too many times
3593have been fixed:
1887da8c 3594
05c8f9ed 3595=over
58856662 3596
05c8f9ed 3597=item *
c88a046d 3598
05c8f9ed 3599C<PerlIO::get_layers> [perl #97956]
58856662
NC
3600
3601=item *
3602
05c8f9ed
RS
3603C<$tied =~ y/a/b/>, C<chop $tied> and C<chomp $tied> when $tied holds a
3604reference.
d333a655 3605
05c8f9ed 3606=item *
589c1691 3607
05c8f9ed 3608When calling C<local $_> [perl #105912]
fae9e8f4 3609
05c8f9ed 3610=item *
fae9e8f4 3611
05c8f9ed 3612Four-argument C<select>
58856662 3613
05c8f9ed
RS
3614=item *
3615
3616A tied buffer passed to C<sysread>
c11980ad 3617
05c8f9ed 3618=item *
c11980ad 3619
05c8f9ed 3620C<< $tied .= <> >>
c88a046d 3621
05c8f9ed 3622=item *
c88a046d 3623
05c8f9ed
RS
3624Three-argument C<open>, the third being a tied file handle
3625(as in C<< open $fh, ">&", $tied >>)
2630d42b 3626
05c8f9ed 3627=item *
c88a046d 3628
05c8f9ed 3629C<sort> with a reference to a tied glob for the comparison routine.
c88a046d
A
3630
3631=item *
3632
05c8f9ed 3633C<..> and C<...> in list context [perl #53554].
c88a046d 3634
05c8f9ed 3635=item *
c88a046d 3636
05c8f9ed
RS
3637C<${$tied}>, C<@{$tied}>, C<%{$tied}> and C<*{$tied}> where the tied
3638variable returns a string (C<&{}> was unaffected)
c88a046d 3639
05c8f9ed 3640=item *
c88a046d 3641
05c8f9ed
RS
3642C<defined ${ $tied_variable }>
3643
3644=item *
c88a046d 3645
05c8f9ed
RS
3646Various functions that take a filehandle argument in rvalue context
3647(C<close>, C<readline>, etc.) [perl #97482]
c88a046d 3648
05c8f9ed 3649=item *
c11980ad 3650
05c8f9ed
RS
3651Some cases of dereferencing a complex expression, such as
3652C<${ (), $tied } = 1>, used to call C<FETCH> multiple times, but now call
3653it once.
c11980ad 3654
05c8f9ed 3655=item *
c88a046d 3656
05c8f9ed
RS
3657C<$tied-E<gt>method> where $tied returns a package name--even resulting in
3658a failure to call the method, due to memory corruption
6c69e197 3659
05c8f9ed 3660=item *
c88a046d 3661
05c8f9ed 3662Assignments like C<*$tied = \&{"..."}> and C<*glob = $tied>
c88a046d
A
3663
3664=item *
3665
05c8f9ed
RS
3666C<chdir>, C<chmod>, C<chown>, C<utime>, C<truncate>, C<stat>, C<lstat> and
3667the filetest ops (C<-r>, C<-x>, etc.)
c88a046d 3668
2630d42b 3669=back
c88a046d 3670
05c8f9ed 3671=item *
2630d42b 3672
05c8f9ed
RS
3673C<caller> sets C<@DB::args> to the subroutine arguments when called from
3674the DB package. It used to crash when doing so if C<@DB::args> happened to
3675be tied. Now it croaks instead.
84ecb73f 3676
fae9e8f4
A
3677=item *
3678
05c8f9ed
RS
3679Tying an element of %ENV or C<%^H> and then deleting that element would
3680result in a call to the tie object's DELETE method, even though tying the
3681element itself is supposed to be equivalent to tying a scalar (the element
3682is, of course, a scalar) [perl #67490].
fae9e8f4 3683
05c8f9ed 3684=item *
a3cc0403 3685
05c8f9ed
RS
3686When Perl autovivifies an element of a tied array or hash (which entails
3687calling STORE with a new reference), it now calls FETCH immediately after
3688the STORE, instead of assuming that FETCH would have returned the same
3689reference. This can make it easier to implement tied objects [perl #35865, #43011].
c88a046d 3690
05c8f9ed 3691=item *
75ff5956 3692
05c8f9ed
RS
3693Four-argument C<select> no longer produces its "Non-string passed as
3694bitmask" warning on tied or tainted variables that are strings.
3695
3696=item *
3697
1540de7c 3698Localizing a tied scalar that returns a typeglob no longer stops it from
05c8f9ed
RS
3699being tied till the end of the scope.
3700
3701=item *
3702
3703Attempting to C<goto> out of a tied handle method used to cause memory
3704corruption or crashes. Now it produces an error message instead
3705[perl #8611].
75ff5956 3706
2630d42b 3707=item *
c11980ad 3708
05c8f9ed
RS
3709A bug has been fixed that occurs when a tied variable is used as a
3710subroutine reference: if the last thing assigned to or returned from the
3711variable was a reference or typeglob, the C<\&$tied> could either crash or
3712return the wrong subroutine. The reference case is a regression introduced
3713in Perl 5.10.0. For typeglobs, it has probably never worked till now.
843331c7 3714
2630d42b 3715=back
c11980ad 3716
05c8f9ed 3717=head2 Version objects and vstrings
2a7afa74 3718
05c8f9ed 3719=over
ecd144ea 3720
05c8f9ed 3721=item *
c11980ad 3722
05c8f9ed
RS
3723The bitwise complement operator (and possibly other operators, too) when
3724passed a vstring would leave vstring magic attached to the return value,
3725even though the string had changed. This meant that
3726C<< version->new(~v1.2.3) >> would create a version looking like "v1.2.3"
3727even though the string passed to C<< version->new >> was actually
3728"\376\375\374". This also caused L<B::Deparse> to deparse C<~v1.2.3>
3729incorrectly, without the C<~> [perl #29070].
c11980ad 3730
05c8f9ed
RS
3731=item *
3732
3733Assigning a vstring to a magic (e.g., tied, C<$!>) variable and then
7af9a95b 3734assigning something else used to blow away all magic. This meant that
05c8f9ed
RS
3735tied variables would come undone, C<$!> would stop getting updated on
3736failed system calls, C<$|> would stop setting autoflush, and other
3737mischief would take place. This has been fixed.
3738
3739=item *
3740
3741C<< version->new("version") >> and C<printf "%vd", "version"> no longer
3742crash [perl #102586].
ecd144ea
FC
3743
3744=item *
3745
05c8f9ed
RS
3746Version comparisons, such as those that happen implicitly with C<use
3747v5.43>, no longer cause locale settings to change [perl #105784].
3748
3749=item *
3750
3751Version objects no longer cause memory leaks in boolean context
3752[perl #109762].
9dea6244 3753
204b72a4 3754=back
9dea6244 3755
05c8f9ed 3756=head2 Warnings, redefinition
c11980ad 3757
05c8f9ed 3758=over
2a7afa74 3759
05c8f9ed 3760=item *
e9e4ee62 3761
05c8f9ed
RS
3762Subroutines from the C<autouse> namespace are once more exempt from
3763redefinition warnings. This used to work in 5.005, but was broken in
37645.6 for most subroutines. For subs created via XS that redefine
3765subroutines from the C<autouse> package, this stopped working in 5.10.
937a45d0 3766
ef337e16
CBW
3767=item *
3768
05c8f9ed
RS
3769New XSUBs now produce redefinition warnings if they overwrite existing
3770subs, as they did in 5.8.x. (The C<autouse> logic was reversed in
37715.10-14. Only subroutines from the C<autouse> namespace would warn
3772when clobbered.)
ef337e16 3773
05c8f9ed 3774=item *
679b54e7 3775
05c8f9ed
RS
3776C<newCONSTSUB> used to use compile-time warning hints, instead of
3777run-time hints. The following code should never produce a redefinition
3778warning, but it used to, if C<newCONSTSUB> redefined an existing
3779subroutine:
39de7394 3780
05c8f9ed
RS
3781 use warnings;
3782 BEGIN {
3783 no warnings;
3784 some_XS_function_that_calls_new_CONSTSUB();
3785 }
2630d42b 3786
05c8f9ed 3787=item *
2630d42b 3788
05c8f9ed 3789Redefinition warnings for constant subroutines are on by default (what
7af9a95b 3790are known as severe warnings in L<perldiag>). This occurred only
05c8f9ed
RS
3791when it was a glob assignment or declaration of a Perl subroutine that
3792caused the warning. If the creation of XSUBs triggered the warning, it
3793was not a default warning. This has been corrected.
52272450 3794
84ecb73f
SM
3795=item *
3796
05c8f9ed
RS
3797The internal check to see whether a redefinition warning should occur
3798used to emit "uninitialized" warnings in cases like this:
3799
3800 use warnings "uninitialized";
3801 use constant {u => undef, v => undef};
3802 sub foo(){u}
3803 sub foo(){v}
84ecb73f 3804
52deee2e 3805=back
5dd80d85 3806
05c8f9ed 3807=head2 Warnings, "Uninitialized"
52272450 3808
05c8f9ed 3809=over
249950d7 3810
05c8f9ed
RS
3811=item *
3812
3813Various functions that take a filehandle argument in rvalue context
3814(C<close>, C<readline>, etc.) used to warn twice for an undefined handle
3815[perl #97482].
a1d95121 3816
05c8f9ed 3817=item *
977c1d31 3818
05c8f9ed
RS
3819C<dbmopen> now only warns once, rather than three times, if the mode
3820argument is C<undef> [perl #90064].
977c1d31 3821
05c8f9ed
RS
3822=item *
3823
3824The C<+=> operator does not usually warn when the left-hand side is
3825C<undef>, but it was doing so for tied variables. This has been fixed
3826[perl #44895].
3827
3828=item *
977c1d31 3829
05c8f9ed
RS
3830A bug fix in Perl 5.14 introduced a new bug, causing "uninitialized"
3831warnings to report the wrong variable if the operator in question had
3832two operands and one was C<%{...}> or C<@{...}>. This has been fixed
3833[perl #103766].
39ea6a4b 3834
05c8f9ed
RS
3835=item *
3836
3837C<..> and C<...> in list context now mention the name of the variable in
3838"uninitialized" warnings for string (as opposed to numeric) ranges.
39ea6a4b 3839
977c1d31 3840=back
ea317ccb 3841
05c8f9ed 3842=head2 Weak references
a7bff800 3843
05c8f9ed 3844=over
0aae26c1 3845
05c8f9ed
RS
3846=item *
3847
3848Weakening the first argument to an automatically-invoked C<DESTROY> method
3849could result in erroneous "DESTROY created new reference" errors or
3850crashes. Now it is an error to weaken a read-only reference.
3851
3852=item *
3853
3854Weak references to lexical hashes going out of scope were not going stale
3855(becoming undefined), but continued to point to the hash.
0aae26c1 3856
05c8f9ed
RS
3857=item *
3858
3859Weak references to lexical variables going out of scope are now broken
3860before any magical methods (e.g., DESTROY on a tie object) are called.
3861This prevents such methods from modifying the variable that will be seen
3862the next time the scope is entered.
3863
3864=item *
d5dc7001 3865
05c8f9ed
RS
3866Creating a weak reference to an @ISA array or accessing the array index
3867(C<$#ISA>) could result in confused internal bookkeeping for elements
7af9a95b 3868later added to the @ISA array. For instance, creating a weak
05c8f9ed
RS
3869reference to the element itself could push that weak reference on to @ISA;
3870and elements added after use of C<$#ISA> would be ignored by method lookup
3871[perl #85670].
d5dc7001 3872
2630d42b 3873=back
d5dc7001 3874
05c8f9ed 3875=head2 Other notable fixes
d5dc7001 3876
05c8f9ed 3877=over
d5dc7001 3878
05c8f9ed 3879=item *
d5dc7001 3880
05c8f9ed
RS
3881C<quotemeta> now quotes consistently the same non-ASCII characters under
3882C<use feature 'unicode_strings'>, regardless of whether the string is
3883encoded in UTF-8 or not, hence fixing the last vestiges (we hope) of the
7af9a95b 3884notorious L<perlunicode/The "Unicode Bug">. [perl #77654].
d5dc7001 3885
05c8f9ed
RS
3886Which of these code points is quoted has changed, based on Unicode's
3887recommendations. See L<perlfunc/quotemeta> for details.
d5dc7001 3888
05c8f9ed 3889=item *
d5dc7001 3890
e58efd23
RS
3891C<study> is now a no-op, presumably fixing all outstanding bugs related to
3892study causing regex matches to behave incorrectly!
3893
3894=item *
3895
05c8f9ed
RS
3896When one writes C<open foo || die>, which used to work in Perl 4, a
3897"Precedence problem" warning is produced. This warning used erroneously to
3898apply to fully-qualified bareword handle names not followed by C<||>. This
3899has been corrected.
d5dc7001 3900
05c8f9ed 3901=item *
d5dc7001 3902
05c8f9ed
RS
3903After package aliasing (C<*foo:: = *bar::>), C<select> with 0 or 1 argument
3904would sometimes return a name that could not be used to refer to the
3905filehandle, or sometimes it would return C<undef> even when a filehandle
3906was selected. Now it returns a typeglob reference in such cases.
d5dc7001 3907
05c8f9ed
RS
3908=item *
3909
3910C<PerlIO::get_layers> no longer ignores some arguments that it thinks are
3911numeric, while treating others as filehandle names. It is now consistent
3912for flat scalars (i.e., not references).
d5dc7001
A
3913
3914=item *
3915
1540de7c 3916Unrecognized switches on C<#!> line
0aae26c1 3917
05c8f9ed
RS
3918If a switch, such as B<-x>, that cannot occur on the C<#!> line is used
3919there, perl dies with "Can't emulate...".
0aae26c1 3920
05c8f9ed 3921It used to produce the same message for switches that perl did not
1540de7c 3922recognize at all, whether on the command line or the C<#!> line.
e2e06450 3923
05c8f9ed 3924Now it produces the "Unrecognized switch" error message [perl #104288].
ccfdda5c 3925
05c8f9ed 3926=item *
e2e06450 3927
05c8f9ed
RS
3928C<system> now temporarily blocks the SIGCHLD signal handler, to prevent the
3929signal handler from stealing the exit status [perl #105700].
3930
3931=item *
3932
3933The %n formatting code for C<printf> and C<sprintf>, which causes the number
3934of characters to be assigned to the next argument, now actually
3935assigns the number of characters, instead of the number of bytes.
3936
3937It also works now with special lvalue functions like C<substr> and with
3938nonexistent hash and array elements [perl #3471, #103492].
3939
3940=item *
3941
3942Perl skips copying values returned from a subroutine, for the sake of
7af9a95b 3943speed, if doing so would make no observable difference. Because of faulty
05c8f9ed
RS
3944logic, this would happen with the
3945result of C<delete>, C<shift> or C<splice>, even if the result was
3946referenced elsewhere. It also did so with tied variables about to be freed
3947[perl #91844, #95548].
3948
3949=item *
3950
3951C<utf8::decode> now refuses to modify read-only scalars [perl #91850].
3952
3953=item *
3954
3955Freeing $_ inside a C<grep> or C<map> block, a code block embedded in a
3956regular expression, or an @INC filter (a subroutine returned by a
3957subroutine in @INC) used to result in double frees or crashes
3958[perl #91880, #92254, #92256].
3959
3960=item *
3961
3962C<eval> returns C<undef> in scalar context or an empty list in list
3963context when there is a run-time error. When C<eval> was passed a
3964string in list context and a syntax error occurred, it used to return a
3965list containing a single undefined element. Now it returns an empty
3966list in list context for all errors [perl #80630].
3967
3968=item *
3969
3970C<goto &func> no longer crashes, but produces an error message, when
3971the unwinding of the current subroutine's scope fires a destructor that
3972undefines the subroutine being "goneto" [perl #99850].
3973
3974=item *
3975
3976Perl now holds an extra reference count on the package that code is
3977currently compiling in. This means that the following code no longer
3978crashes [perl #101486]:
3979
3980 package Foo;
3981 BEGIN {*Foo:: = *Bar::}
3982 sub foo;
3983
3984=item *
3985
3986The C<x> repetition operator no longer crashes on 64-bit builds with large
3987repeat counts [perl #94560].
3988
3989=item *
3990
3991Calling C<require> on an implicit C<$_> when C<*CORE::GLOBAL::require> has
3992been overridden does not segfault anymore, and C<$_> is now passed to the
3993overriding subroutine [perl #78260].
3994
3995=item *
3996
3997C<use> and C<require> are no longer affected by the I/O layers active in
3998the caller's scope (enabled by L<open.pm|open>) [perl #96008].
3999
4000=item *
4001
4002C<our $::é; $é> (which is invalid) no longer produces the "Compilation
4003error at lib/utf8_heavy.pl..." error message, which it started emitting in
40045.10.0 [perl #99984].
4005
4006=item *
4007
4008On 64-bit systems, C<read()> now understands large string offsets beyond
4009the 32-bit range.
4010
4011=item *
4012
4013Errors that occur when processing subroutine attributes no longer cause the
4014subroutine's op tree to leak.
4015
4016=item *
4017
4018Passing the same constant subroutine to both C<index> and C<formline> no
4019longer causes one or the other to fail [perl #89218]. (5.14.1)
4020
4021=item *
4022
4023List assignment to lexical variables declared with attributes in the same
4024statement (C<my ($x,@y) : blimp = (72,94)>) stopped working in Perl 5.8.0.
4025It has now been fixed.
4026
4027=item *
4028
4029Perl 5.10.0 introduced some faulty logic that made "U*" in the middle of
4030a pack template equivalent to "U0" if the input string was empty. This has
006b5090 4031been fixed [perl #90160]. (5.14.2)
05c8f9ed
RS
4032
4033=item *
4034
4035Destructors on objects were not called during global destruction on objects
4036that were not referenced by any scalars. This could happen if an array
4037element were blessed (e.g., C<bless \$a[0]>) or if a closure referenced a
4038blessed variable (C<bless \my @a; sub foo { @a }>).
4039
4040Now there is an extra pass during global destruction to fire destructors on
4041any objects that might be left after the usual passes that check for
4042objects referenced by scalars [perl #36347].
4043
4044=item *
4045
4046Fixed a case where it was possible that a freed buffer may have been read
4047from when parsing a here document [perl #90128]. (5.14.1)
4048
4049=item *
4050
4051C<each(I<ARRAY>)> is now wrapped in C<defined(...)>, like C<each(I<HASH>)>,
4052inside a C<while> condition [perl #90888].
4053
4054=item *
4055
4056A problem with context propagation when a C<do> block is an argument to
4057C<return> has been fixed. It used to cause C<undef> to be returned in
7af9a95b 4058certain cases of a C<return> inside an C<if> block which itself is followed by
05c8f9ed