Debian Codegraph @

upstream/debian-unstable/man1/perl5220delta.1

raw
1 .\" -*- mode: troff; coding: utf-8 -*-
2 .\" Automatically generated by Pod::Man 5.0102 (Pod::Simple 3.45)
3 .\"
4 .\" Standard preamble:
5 .\" ========================================================================
6 .de Sp \" Vertical space (when we can't use .PP)
7 .if t .sp .5v
8 .if n .sp
9 ..
10 .de Vb \" Begin verbatim text
11 .ft CW
12 .nf
13 .ne \\$1
14 ..
15 .de Ve \" End verbatim text
16 .ft R
17 .fi
18 ..
19 .\" \*(C` and \*(C' are quotes in nroff, nothing in troff, for use with C<>.
20 .ie n \{\
21 . ds C` ""
22 . ds C' ""
23 'br\}
24 .el\{\
25 . ds C`
26 . ds C'
27 'br\}
28 .\"
29 .\" Escape single quotes in literal strings from groff's Unicode transform.
30 .ie \n(.g .ds Aq \(aq
31 .el .ds Aq '
32 .\"
33 .\" If the F register is >0, we'll generate index entries on stderr for
34 .\" titles (.TH), headers (.SH), subsections (.SS), items (.Ip), and index
35 .\" entries marked with X<> in POD. Of course, you'll have to process the
36 .\" output yourself in some meaningful fashion.
37 .\"
38 .\" Avoid warning from groff about undefined register 'F'.
39 .de IX
40 ..
41 .nr rF 0
42 .if \n(.g .if rF .nr rF 1
43 .if (\n(rF:(\n(.g==0)) \{\
44 . if \nF \{\
45 . de IX
46 . tm Index:\\$1\t\\n%\t"\\$2"
47 ..
48 . if !\nF==2 \{\
49 . nr % 0
50 . nr F 2
51 . \}
52 . \}
53 .\}
54 .rr rF
55 .\" ========================================================================
56 .\"
57 .IX Title "PERL5220DELTA 1"
58 .TH PERL5220DELTA 1 2025-11-16 "perl v5.40.1" "Perl Programmers Reference Guide"
59 .\" For nroff, turn off justification. Always turn off hyphenation; it makes
60 .\" way too many mistakes in technical documents.
61 .if n .ad l
62 .nh
63 .SH NAME
64 perl5220delta \- what is new for perl v5.22.0
65 .SH DESCRIPTION
66 .IX Header "DESCRIPTION"
67 This document describes differences between the 5.20.0 release and the 5.22.0
68 release.
69 .PP
70 If you are upgrading from an earlier release such as 5.18.0, first read
71 perl5200delta, which describes differences between 5.18.0 and 5.20.0.
72 .SH "Core Enhancements"
73 .IX Header "Core Enhancements"
74 .SS "New bitwise operators"
75 .IX Subsection "New bitwise operators"
76 A new experimental facility has been added that makes the four standard
77 bitwise operators (\f(CW\*(C`& | ^ ~\*(C'\fR) treat their operands consistently as
78 numbers, and introduces four new dotted operators (\f(CW\*(C`&. |. ^. ~.\*(C'\fR) that
79 treat their operands consistently as strings. The same applies to the
80 assignment variants (\f(CW\*(C`&= |= ^= &.= |.= ^.=\*(C'\fR).
81 .PP
82 To use this, enable the "bitwise" feature and disable the
83 "experimental::bitwise" warnings category. See "Bitwise String
84 Operators" in perlop for details.
85 [GH #14348] <https://github.com/Perl/perl5/issues/14348>.
86 .SS "New double-diamond operator"
87 .IX Subsection "New double-diamond operator"
88 \&\f(CW\*(C`<<>>\*(C'\fR is like \f(CW\*(C`<>\*(C'\fR but uses three-argument \f(CW\*(C`open\*(C'\fR to open
89 each file in \f(CW@ARGV\fR. This means that each element of \f(CW@ARGV\fR will be treated
90 as an actual file name, and \f(CW"|foo"\fR won't be treated as a pipe open.
91 .ie n .SS "New ""\eb"" boundaries in regular expressions"
92 .el .SS "New \f(CW\eb\fP boundaries in regular expressions"
93 .IX Subsection "New b boundaries in regular expressions"
94 \fR\f(CI\*(C`qr/\eb{gcb}/\*(C'\fR\fI\fR
95 .IX Subsection "qr/b{gcb}/"
96 .PP
97 \&\f(CW\*(C`gcb\*(C'\fR stands for Grapheme Cluster Boundary. It is a Unicode property
98 that finds the boundary between sequences of characters that look like a
99 single character to a native speaker of a language. Perl has long had
100 the ability to deal with these through the \f(CW\*(C`\eX\*(C'\fR regular escape
101 sequence. Now, there is an alternative way of handling these. See
102 "\eb{}, \eb, \eB{}, \eB" in perlrebackslash for details.
103 .PP
104 \fR\f(CI\*(C`qr/\eb{wb}/\*(C'\fR\fI\fR
105 .IX Subsection "qr/b{wb}/"
106 .PP
107 \&\f(CW\*(C`wb\*(C'\fR stands for Word Boundary. It is a Unicode property
108 that finds the boundary between words. This is similar to the plain
109 \&\f(CW\*(C`\eb\*(C'\fR (without braces) but is more suitable for natural language
110 processing. It knows, for example, that apostrophes can occur in the
111 middle of words. See "\eb{}, \eb, \eB{}, \eB" in perlrebackslash for details.
112 .PP
113 \fR\f(CI\*(C`qr/\eb{sb}/\*(C'\fR\fI\fR
114 .IX Subsection "qr/b{sb}/"
115 .PP
116 \&\f(CW\*(C`sb\*(C'\fR stands for Sentence Boundary. It is a Unicode property
117 to aid in parsing natural language sentences.
118 See "\eb{}, \eb, \eB{}, \eB" in perlrebackslash for details.
119 .SS "Non-Capturing Regular Expression Flag"
120 .IX Subsection "Non-Capturing Regular Expression Flag"
121 Regular expressions now support a \f(CW\*(C`/n\*(C'\fR flag that disables capturing
122 and filling in \f(CW$1\fR, \f(CW$2\fR, etc inside of groups:
123 .PP
124 .Vb 1
125 \& "hello" =~ /(hi|hello)/n; # $1 is not set
126 .Ve
127 .PP
128 This is equivalent to putting \f(CW\*(C`?:\*(C'\fR at the beginning of every capturing group.
129 .PP
130 See "n" in perlre for more information.
131 .ie n .SS """use re \*(Aqstrict\*(Aq"""
132 .el .SS "\f(CWuse re \*(Aqstrict\*(Aq\fP"
133 .IX Subsection "use re strict"
134 This applies stricter syntax rules to regular expression patterns
135 compiled within its scope. This will hopefully alert you to typos and
136 other unintentional behavior that backwards-compatibility issues prevent
137 us from reporting in normal regular expression compilations. Because the
138 behavior of this is subject to change in future Perl releases as we gain
139 experience, using this pragma will raise a warning of category
140 \&\f(CW\*(C`experimental::re_strict\*(C'\fR.
141 See 'strict' in re.
142 .SS "Unicode 7.0 (with correction) is now supported"
143 .IX Subsection "Unicode 7.0 (with correction) is now supported"
144 For details on what is in this release, see
145 <http://www.unicode.org/versions/Unicode7.0.0/>.
146 The version of Unicode 7.0 that comes with Perl includes
147 a correction dealing with glyph shaping in Arabic
148 (see <http://www.unicode.org/errata/#current_errata>).
149 .ie n .SS """use\ locale"" can restrict which locale categories are affected"
150 .el .SS "\f(CWuse\ locale\fP can restrict which locale categories are affected"
151 .IX Subsection "use locale can restrict which locale categories are affected"
152 It is now possible to pass a parameter to \f(CW\*(C`use\ locale\*(C'\fR to specify
153 a subset of locale categories to be locale-aware, with the remaining
154 ones unaffected. See "The "use locale" pragma" in perllocale for details.
155 .SS "Perl now supports POSIX 2008 locale currency additions"
156 .IX Subsection "Perl now supports POSIX 2008 locale currency additions"
157 On platforms that are able to handle POSIX.1\-2008, the
158 hash returned by
159 \&\f(CWPOSIX::localeconv()\fR
160 includes the international currency fields added by that version of the
161 POSIX standard. These are
162 \&\f(CW\*(C`int_n_cs_precedes\*(C'\fR,
163 \&\f(CW\*(C`int_n_sep_by_space\*(C'\fR,
164 \&\f(CW\*(C`int_n_sign_posn\*(C'\fR,
165 \&\f(CW\*(C`int_p_cs_precedes\*(C'\fR,
166 \&\f(CW\*(C`int_p_sep_by_space\*(C'\fR,
167 and
168 \&\f(CW\*(C`int_p_sign_posn\*(C'\fR.
169 .SS "Better heuristics on older platforms for determining locale UTF\-8ness"
170 .IX Subsection "Better heuristics on older platforms for determining locale UTF-8ness"
171 On platforms that implement neither the C99 standard nor the POSIX 2001
172 standard, determining if the current locale is UTF\-8 or not depends on
173 heuristics. These are improved in this release.
174 .SS "Aliasing via reference"
175 .IX Subsection "Aliasing via reference"
176 Variables and subroutines can now be aliased by assigning to a reference:
177 .PP
178 .Vb 2
179 \& \e$c = \e$d;
180 \& \e&x = \e&y;
181 .Ve
182 .PP
183 Aliasing can also be accomplished
184 by using a backslash before a \f(CW\*(C`foreach\*(C'\fR iterator variable; this is
185 perhaps the most useful idiom this feature provides:
186 .PP
187 .Vb 1
188 \& foreach \e%hash (@array_of_hash_refs) { ... }
189 .Ve
190 .PP
191 This feature is experimental and must be enabled via \f(CWuse\ feature\ \*(Aqrefaliasing\*(Aq\fR. It will warn unless the \f(CW\*(C`experimental::refaliasing\*(C'\fR
192 warnings category is disabled.
193 .PP
194 See "Assigning to References" in perlref
195 .ie n .SS """prototype"" with no arguments"
196 .el .SS "\f(CWprototype\fP with no arguments"
197 .IX Subsection "prototype with no arguments"
198 \&\f(CWprototype()\fR with no arguments now infers \f(CW$_\fR.
199 [GH #14376] <https://github.com/Perl/perl5/issues/14376>.
200 .ie n .SS "New "":const"" subroutine attribute"
201 .el .SS "New \f(CW:const\fP subroutine attribute"
202 .IX Subsection "New :const subroutine attribute"
203 The \f(CW\*(C`const\*(C'\fR attribute can be applied to an anonymous subroutine. It
204 causes the new sub to be executed immediately whenever one is created
205 (\fIi.e.\fR when the \f(CW\*(C`sub\*(C'\fR expression is evaluated). Its value is captured
206 and used to create a new constant subroutine that is returned. This
207 feature is experimental. See "Constant Functions" in perlsub.
208 .ie n .SS """fileno"" now works on directory handles"
209 .el .SS "\f(CWfileno\fP now works on directory handles"
210 .IX Subsection "fileno now works on directory handles"
211 When the relevant support is available in the operating system, the
212 \&\f(CW\*(C`fileno\*(C'\fR builtin now works on directory handles, yielding the
213 underlying file descriptor in the same way as for filehandles. On
214 operating systems without such support, \f(CW\*(C`fileno\*(C'\fR on a directory handle
215 continues to return the undefined value, as before, but also sets \f(CW$!\fR to
216 indicate that the operation is not supported.
217 .PP
218 Currently, this uses either a \f(CW\*(C`dd_fd\*(C'\fR member in the OS \f(CW\*(C`DIR\*(C'\fR
219 structure, or a \f(CWdirfd(3)\fR function as specified by POSIX.1\-2008.
220 .SS "List form of pipe open implemented for Win32"
221 .IX Subsection "List form of pipe open implemented for Win32"
222 The list form of pipe:
223 .PP
224 .Vb 1
225 \& open my $fh, "\-|", "program", @arguments;
226 .Ve
227 .PP
228 is now implemented on Win32. It has the same limitations as \f(CW\*(C`system
229 LIST\*(C'\fR on Win32, since the Win32 API doesn't accept program arguments
230 as a list.
231 .SS "Assignment to list repetition"
232 .IX Subsection "Assignment to list repetition"
233 \&\f(CW\*(C`(...) x ...\*(C'\fR can now be used within a list that is assigned to, as long
234 as the left-hand side is a valid lvalue. This allows \f(CW\*(C`(undef,undef,$foo)\ =\ that_function()\*(C'\fR to be written as \f(CW\*(C`((undef)x2,\ $foo)\ =\ that_function()\*(C'\fR.
235 .SS "Infinity and NaN (not-a-number) handling improved"
236 .IX Subsection "Infinity and NaN (not-a-number) handling improved"
237 Floating point values are able to hold the special values infinity, negative
238 infinity, and NaN (not-a-number). Now we more robustly recognize and
239 propagate the value in computations, and on output normalize them to the strings
240 \&\f(CW\*(C`Inf\*(C'\fR, \f(CW\*(C`\-Inf\*(C'\fR, and \f(CW\*(C`NaN\*(C'\fR.
241 .PP
242 See also the POSIX enhancements.
243 .SS "Floating point parsing has been improved"
244 .IX Subsection "Floating point parsing has been improved"
245 Parsing and printing of floating point values has been improved.
246 .PP
247 As a completely new feature, hexadecimal floating point literals
248 (like \f(CW\*(C`0x1.23p\-4\*(C'\fR) are now supported, and they can be output with
249 \&\f(CW\*(C`printf\ "%a"\*(C'\fR. See "Scalar value constructors" in perldata for more
250 details.
251 .SS "Packing infinity or not-a-number into a character is now fatal"
252 .IX Subsection "Packing infinity or not-a-number into a character is now fatal"
253 Before, when trying to pack infinity or not-a-number into a
254 (signed) character, Perl would warn, and assumed you tried to
255 pack \f(CW0xFF\fR; if you gave it as an argument to \f(CW\*(C`chr\*(C'\fR,
256 \&\f(CW\*(C`U+FFFD\*(C'\fR was returned.
257 .PP
258 But now, all such actions (\f(CW\*(C`pack\*(C'\fR, \f(CW\*(C`chr\*(C'\fR, and \f(CW\*(C`print \*(Aq%c\*(Aq\*(C'\fR)
259 result in a fatal error.
260 .SS "Experimental C Backtrace API"
261 .IX Subsection "Experimental C Backtrace API"
262 Perl now supports (via a C level API) retrieving
263 the C level backtrace (similar to what symbolic debuggers like gdb do).
264 .PP
265 The backtrace returns the stack trace of the C call frames,
266 with the symbol names (function names), the object names (like "perl"),
267 and if it can, also the source code locations (file:line).
268 .PP
269 The supported platforms are Linux and OS X (some *BSD might work at
270 least partly, but they have not yet been tested).
271 .PP
272 The feature needs to be enabled with \f(CW\*(C`Configure \-Dusecbacktrace\*(C'\fR.
273 .PP
274 See "C backtrace" in perlhacktips for more information.
275 .SH Security
276 .IX Header "Security"
277 .ie n .SS "Perl is now compiled with ""\-fstack\-protector\-strong"" if available"
278 .el .SS "Perl is now compiled with \f(CW\-fstack\-protector\-strong\fP if available"
279 .IX Subsection "Perl is now compiled with -fstack-protector-strong if available"
280 Perl has been compiled with the anti-stack-smashing option
281 \&\f(CW\*(C`\-fstack\-protector\*(C'\fR since 5.10.1. Now Perl uses the newer variant
282 called \f(CW\*(C`\-fstack\-protector\-strong\*(C'\fR, if available.
283 .SS "The Safe module could allow outside packages to be replaced"
284 .IX Subsection "The Safe module could allow outside packages to be replaced"
285 Critical bugfix: outside packages could be replaced. Safe has
286 been patched to 2.38 to address this.
287 .ie n .SS "Perl is now always compiled with ""\-D_FORTIFY_SOURCE=2"" if available"
288 .el .SS "Perl is now always compiled with \f(CW\-D_FORTIFY_SOURCE=2\fP if available"
289 .IX Subsection "Perl is now always compiled with -D_FORTIFY_SOURCE=2 if available"
290 The 'code hardening' option called \f(CW\*(C`_FORTIFY_SOURCE\*(C'\fR, available in
291 gcc 4.*, is now always used for compiling Perl, if available.
292 .PP
293 Note that this isn't necessarily a huge step since in many platforms
294 the step had already been taken several years ago: many Linux
295 distributions (like Fedora) have been using this option for Perl,
296 and OS X has enforced the same for many years.
297 .SH "Incompatible Changes"
298 .IX Header "Incompatible Changes"
299 .SS "Subroutine signatures moved before attributes"
300 .IX Subsection "Subroutine signatures moved before attributes"
301 The experimental sub signatures feature, as introduced in 5.20, parsed
302 signatures after attributes. In this release, following feedback from users
303 of the experimental feature, the positioning has been moved such that
304 signatures occur after the subroutine name (if any) and before the attribute
305 list (if any).
306 .ie n .SS """&"" and ""\e&"" prototypes accepts only subs"
307 .el .SS "\f(CW&\fP and \f(CW\e&\fP prototypes accepts only subs"
308 .IX Subsection "& and & prototypes accepts only subs"
309 The \f(CW\*(C`&\*(C'\fR prototype character now accepts only anonymous subs (\f(CW\*(C`sub
310 {...}\*(C'\fR), things beginning with \f(CW\*(C`\e&\*(C'\fR, or an explicit \f(CW\*(C`undef\*(C'\fR. Formerly
311 it erroneously also allowed references to arrays, hashes, and lists.
312 [GH #2776] <https://github.com/Perl/perl5/issues/2776>.
313 [GH #14186] <https://github.com/Perl/perl5/issues/14186>.
314 [GH #14353] <https://github.com/Perl/perl5/issues/14353>.
315 .PP
316 In addition, the \f(CW\*(C`\e&\*(C'\fR prototype was allowing subroutine calls, whereas
317 now it only allows subroutines: \f(CW&foo\fR is still permitted as an argument,
318 while \f(CW&foo()\fR and \f(CWfoo()\fR no longer are.
319 [GH #10633] <https://github.com/Perl/perl5/issues/10633>.
320 .ie n .SS """use encoding"" is now lexical"
321 .el .SS "\f(CWuse encoding\fP is now lexical"
322 .IX Subsection "use encoding is now lexical"
323 The encoding pragma's effect is now limited to lexical scope. This
324 pragma is deprecated, but in the meantime, it could adversely affect
325 unrelated modules that are included in the same program; this change
326 fixes that.
327 .SS "List slices returning empty lists"
328 .IX Subsection "List slices returning empty lists"
329 List slices now return an empty list only if the original list was empty
330 (or if there are no indices). Formerly, a list slice would return an empty
331 list if all indices fell outside the original list; now it returns a list
332 of \f(CW\*(C`undef\*(C'\fR values in that case.
333 [GH #12335] <https://github.com/Perl/perl5/issues/12335>.
334 .ie n .SS """\eN{}"" with a sequence of multiple spaces is now a fatal error"
335 .el .SS "\f(CW\eN{}\fP with a sequence of multiple spaces is now a fatal error"
336 .IX Subsection "N{} with a sequence of multiple spaces is now a fatal error"
337 E.g. \f(CW\*(C`\eN{TOO\ \ MANY\ SPACES}\*(C'\fR or \f(CW\*(C`\eN{TRAILING\ SPACE\ }\*(C'\fR.
338 This has been deprecated since v5.18.
339 .ie n .SS """use\ UNIVERSAL\ \*(Aq...\*(Aq"" is now a fatal error"
340 .el .SS "\f(CWuse\ UNIVERSAL\ \*(Aq...\*(Aq\fP is now a fatal error"
341 .IX Subsection "use UNIVERSAL ... is now a fatal error"
342 Importing functions from \f(CW\*(C`UNIVERSAL\*(C'\fR has been deprecated since v5.12, and
343 is now a fatal error. \f(CW\*(C`use\ UNIVERSAL\*(C'\fR without any arguments is still
344 allowed.
345 .ie n .SS "In double-quotish ""\ec\fIX\fP"", \fIX\fP must now be a printable ASCII character"
346 .el .SS "In double-quotish \f(CW\ec\fP\f(CIX\fP\f(CW\fP, \fIX\fP must now be a printable ASCII character"
347 .IX Subsection "In double-quotish cX, X must now be a printable ASCII character"
348 In prior releases, failure to do this raised a deprecation warning.
349 .ie n .SS "Splitting the tokens ""(?"" and ""(*"" in regular expressions is now a fatal compilation error"
350 .el .SS "Splitting the tokens \f(CW(?\fP and \f(CW(*\fP in regular expressions is now a fatal compilation error"
351 .IX Subsection "Splitting the tokens (? and (* in regular expressions is now a fatal compilation error"
352 These had been deprecated since v5.18.
353 .ie n .SS """qr/foo/x"" now ignores all Unicode pattern white space"
354 .el .SS "\f(CWqr/foo/x\fP now ignores all Unicode pattern white space"
355 .IX Subsection "qr/foo/x now ignores all Unicode pattern white space"
356 The \f(CW\*(C`/x\*(C'\fR regular expression modifier allows the pattern to contain
357 white space and comments (both of which are ignored) for improved
358 readability. Until now, not all the white space characters that Unicode
359 designates for this purpose were handled. The additional ones now
360 recognized are:
361 .PP
362 .Vb 5
363 \& U+0085 NEXT LINE
364 \& U+200E LEFT\-TO\-RIGHT MARK
365 \& U+200F RIGHT\-TO\-LEFT MARK
366 \& U+2028 LINE SEPARATOR
367 \& U+2029 PARAGRAPH SEPARATOR
368 .Ve
369 .PP
370 The use of these characters with \f(CW\*(C`/x\*(C'\fR outside bracketed character
371 classes and when not preceded by a backslash has raised a deprecation
372 warning since v5.18. Now they will be ignored.
373 .ie n .SS "Comment lines within ""(?[\ ])"" are now ended only by a ""\en"""
374 .el .SS "Comment lines within \f(CW(?[\ ])\fP are now ended only by a \f(CW\en\fP"
375 .IX Subsection "Comment lines within (?[ ]) are now ended only by a n"
376 \&\f(CW\*(C`(?[\ ])\*(C'\fR is an experimental feature, introduced in v5.18. It operates
377 as if \f(CW\*(C`/x\*(C'\fR is always enabled. But there was a difference: comment
378 lines (following a \f(CW\*(C`#\*(C'\fR character) were terminated by anything matching
379 \&\f(CW\*(C`\eR\*(C'\fR which includes all vertical whitespace, such as form feeds. For
380 consistency, this is now changed to match what terminates comment lines
381 outside \f(CW\*(C`(?[\ ])\*(C'\fR, namely a \f(CW\*(C`\en\*(C'\fR (even if escaped), which is the
382 same as what terminates a heredoc string and formats.
383 .ie n .SS """(?[...])"" operators now follow standard Perl precedence"
384 .el .SS "\f(CW(?[...])\fP operators now follow standard Perl precedence"
385 .IX Subsection "(?[...]) operators now follow standard Perl precedence"
386 This experimental feature allows set operations in regular expression patterns.
387 Prior to this, the intersection operator had the same precedence as the other
388 binary operators. Now it has higher precedence. This could lead to different
389 outcomes than existing code expects (though the documentation has always noted
390 that this change might happen, recommending fully parenthesizing the
391 expressions). See "Extended Bracketed Character Classes" in perlrecharclass.
392 .ie n .SS "Omitting ""%"" and ""@"" on hash and array names is no longer permitted"
393 .el .SS "Omitting \f(CW%\fP and \f(CW@\fP on hash and array names is no longer permitted"
394 .IX Subsection "Omitting % and @ on hash and array names is no longer permitted"
395 Really old Perl let you omit the \f(CW\*(C`@\*(C'\fR on array names and the \f(CW\*(C`%\*(C'\fR on hash
396 names in some spots. This has issued a deprecation warning since Perl
397 5.000, and is no longer permitted.
398 .ie n .SS """$!"" text is now in English outside the scope of ""use locale"""
399 .el .SS "\f(CW""$!""\fP text is now in English outside the scope of \f(CWuse locale\fP"
400 .IX Subsection """$!"" text is now in English outside the scope of use locale"
401 Previously, the text, unlike almost everything else, always came out
402 based on the current underlying locale of the program. (Also affected
403 on some systems is \f(CW"$^E"\fR.) For programs that are unprepared to
404 handle locale differences, this can cause garbage text to be displayed.
405 It's better to display text that is translatable via some tool than
406 garbage text which is much harder to figure out.
407 .ie n .SS """$!"" text will be returned in UTF\-8 when appropriate"
408 .el .SS "\f(CW""$!""\fP text will be returned in UTF\-8 when appropriate"
409 .IX Subsection """$!"" text will be returned in UTF-8 when appropriate"
410 The stringification of \f(CW$!\fR and \f(CW$^E\fR will have the UTF\-8 flag set
411 when the text is actually non-ASCII UTF\-8. This will enable programs
412 that are set up to be locale-aware to properly output messages in the
413 user's native language. Code that needs to continue the 5.20 and
414 earlier behavior can do the stringification within the scopes of both
415 \&\f(CW\*(C`use\ bytes\*(C'\fR and \f(CW\*(C`use\ locale\ ":messages"\*(C'\fR. Within these two
416 scopes, no other Perl operations will
417 be affected by locale; only \f(CW$!\fR and \f(CW$^E\fR stringification. The
418 \&\f(CW\*(C`bytes\*(C'\fR pragma causes the UTF\-8 flag to not be set, just as in previous
419 Perl releases. This resolves
420 [GH #12035] <https://github.com/Perl/perl5/issues/12035>.
421 .ie n .SS "Support for ""?PATTERN?"" without explicit operator has been removed"
422 .el .SS "Support for \f(CW?PATTERN?\fP without explicit operator has been removed"
423 .IX Subsection "Support for ?PATTERN? without explicit operator has been removed"
424 The \f(CW\*(C`m?PATTERN?\*(C'\fR construct, which allows matching a regex only once,
425 previously had an alternative form that was written directly with a question
426 mark delimiter, omitting the explicit \f(CW\*(C`m\*(C'\fR operator. This usage has produced
427 a deprecation warning since 5.14.0. It is now a syntax error, so that the
428 question mark can be available for use in new operators.
429 .ie n .SS "defined(@array) and defined(%hash) are now fatal errors"
430 .el .SS "\f(CWdefined(@array)\fP and \f(CWdefined(%hash)\fP are now fatal errors"
431 .IX Subsection "defined(@array) and defined(%hash) are now fatal errors"
432 These have been deprecated since v5.6.1 and have raised deprecation
433 warnings since v5.16.
434 .SS "Using a hash or an array as a reference are now fatal errors"
435 .IX Subsection "Using a hash or an array as a reference are now fatal errors"
436 For example, \f(CW\*(C`%foo\->{"bar"}\*(C'\fR now causes a fatal compilation
437 error. These have been deprecated since before v5.8, and have raised
438 deprecation warnings since then.
439 .ie n .SS "Changes to the ""*"" prototype"
440 .el .SS "Changes to the \f(CW*\fP prototype"
441 .IX Subsection "Changes to the * prototype"
442 The \f(CW\*(C`*\*(C'\fR character in a subroutine's prototype used to allow barewords to take
443 precedence over most, but not all, subroutine names. It was never
444 consistent and exhibited buggy behavior.
445 .PP
446 Now it has been changed, so subroutines always take precedence over barewords,
447 which brings it into conformity with similarly prototyped built-in functions:
448 .PP
449 .Vb 6
450 \& sub splat(*) { ... }
451 \& sub foo { ... }
452 \& splat(foo); # now always splat(foo())
453 \& splat(bar); # still splat(\*(Aqbar\*(Aq) as before
454 \& close(foo); # close(foo())
455 \& close(bar); # close(\*(Aqbar\*(Aq)
456 .Ve
457 .SH Deprecations
458 .IX Header "Deprecations"
459 .ie n .SS "Setting ""${^ENCODING}"" to anything but ""undef"""
460 .el .SS "Setting \f(CW${^ENCODING}\fP to anything but \f(CWundef\fP"
461 .IX Subsection "Setting ${^ENCODING} to anything but undef"
462 This variable allows Perl scripts to be written in an encoding other than
463 ASCII or UTF\-8. However, it affects all modules globally, leading
464 to wrong answers and segmentation faults. New scripts should be written
465 in UTF\-8; old scripts should be converted to UTF\-8, which is easily done
466 with the piconv utility.
467 .SS "Use of non-graphic characters in single-character variable names"
468 .IX Subsection "Use of non-graphic characters in single-character variable names"
469 The syntax for single-character variable names is more lenient than
470 for longer variable names, allowing the one-character name to be a
471 punctuation character or even invisible (a non-graphic). Perl v5.20
472 deprecated the ASCII-range controls as such a name. Now, all
473 non-graphic characters that formerly were allowed are deprecated.
474 The practical effect of this occurs only when not under \f(CW\*(C`use\ utf8\*(C'\fR, and affects just the C1 controls (code points 0x80 through
475 0xFF), NO-BREAK SPACE, and SOFT HYPHEN.
476 .ie n .SS "Inlining of ""sub () { $var }"" with observable side-effects"
477 .el .SS "Inlining of \f(CWsub () { $var }\fP with observable side-effects"
478 .IX Subsection "Inlining of sub () { $var } with observable side-effects"
479 In many cases Perl makes \f(CW\*(C`sub\ ()\ {\ $var\ }\*(C'\fR into an inlinable constant
480 subroutine, capturing the value of \f(CW$var\fR at the time the \f(CW\*(C`sub\*(C'\fR expression
481 is evaluated. This can break the closure behavior in those cases where
482 \&\f(CW$var\fR is subsequently modified, since the subroutine won't return the
483 changed value. (Note that this all only applies to anonymous subroutines
484 with an empty prototype (\f(CW\*(C`sub\ ()\*(C'\fR).)
485 .PP
486 This usage is now deprecated in those cases where the variable could be
487 modified elsewhere. Perl detects those cases and emits a deprecation
488 warning. Such code will likely change in the future and stop producing a
489 constant.
490 .PP
491 If your variable is only modified in the place where it is declared, then
492 Perl will continue to make the sub inlinable with no warnings.
493 .PP
494 .Vb 4
495 \& sub make_constant {
496 \& my $var = shift;
497 \& return sub () { $var }; # fine
498 \& }
499 \&
500 \& sub make_constant_deprecated {
501 \& my $var;
502 \& $var = shift;
503 \& return sub () { $var }; # deprecated
504 \& }
505 \&
506 \& sub make_constant_deprecated2 {
507 \& my $var = shift;
508 \& log_that_value($var); # could modify $var
509 \& return sub () { $var }; # deprecated
510 \& }
511 .Ve
512 .PP
513 In the second example above, detecting that \f(CW$var\fR is assigned to only once
514 is too hard to detect. That it happens in a spot other than the \f(CW\*(C`my\*(C'\fR
515 declaration is enough for Perl to find it suspicious.
516 .PP
517 This deprecation warning happens only for a simple variable for the body of
518 the sub. (A \f(CW\*(C`BEGIN\*(C'\fR block or \f(CW\*(C`use\*(C'\fR statement inside the sub is ignored,
519 because it does not become part of the sub's body.) For more complex
520 cases, such as \f(CW\*(C`sub\ ()\ {\ do_something()\ if\ 0;\ $var\ }\*(C'\fR the behavior has
521 changed such that inlining does not happen if the variable is modifiable
522 elsewhere. Such cases should be rare.
523 .ie n .SS "Use of multiple ""/x"" regexp modifiers"
524 .el .SS "Use of multiple \f(CW/x\fP regexp modifiers"
525 .IX Subsection "Use of multiple /x regexp modifiers"
526 It is now deprecated to say something like any of the following:
527 .PP
528 .Vb 3
529 \& qr/foo/xx;
530 \& /(?xax:foo)/;
531 \& use re qw(/amxx);
532 .Ve
533 .PP
534 That is, now \f(CW\*(C`x\*(C'\fR should only occur once in any string of contiguous
535 regular expression pattern modifiers. We do not believe there are any
536 occurrences of this in all of CPAN. This is in preparation for a future
537 Perl release having \f(CW\*(C`/xx\*(C'\fR permit white-space for readability in
538 bracketed character classes (those enclosed in square brackets:
539 \&\f(CW\*(C`[...]\*(C'\fR).
540 .ie n .SS "Using a NO-BREAK space in a character alias for ""\eN{...}"" is now deprecated"
541 .el .SS "Using a NO-BREAK space in a character alias for \f(CW\eN{...}\fP is now deprecated"
542 .IX Subsection "Using a NO-BREAK space in a character alias for N{...} is now deprecated"
543 This non-graphic character is essentially indistinguishable from a
544 regular space, and so should not be allowed. See
545 "CUSTOM ALIASES" in charnames.
546 .ie n .SS "A literal ""{"" should now be escaped in a pattern"
547 .el .SS "A literal \f(CW""{""\fP should now be escaped in a pattern"
548 .IX Subsection "A literal ""{"" should now be escaped in a pattern"
549 If you want a literal left curly bracket (also called a left brace) in a
550 regular expression pattern, you should now escape it by either
551 preceding it with a backslash (\f(CW"\e{"\fR) or enclosing it within square
552 brackets \f(CW"[{]"\fR, or by using \f(CW\*(C`\eQ\*(C'\fR; otherwise a deprecation warning
553 will be raised. This was first announced as forthcoming in the v5.16
554 release; it will allow future extensions to the language to happen.
555 .SS "Making all warnings fatal is discouraged"
556 .IX Subsection "Making all warnings fatal is discouraged"
557 The documentation for fatal warnings notes that
558 \&\f(CW\*(C`use warnings FATAL => \*(Aqall\*(Aq\*(C'\fR is discouraged, and provides stronger
559 language about the risks of fatal warnings in general.
560 .SH "Performance Enhancements"
561 .IX Header "Performance Enhancements"
562 .IP \(bu 4
563 If a method or class name is known at compile time, a hash is precomputed
564 to speed up run-time method lookup. Also, compound method names like
565 \&\f(CW\*(C`SUPER::new\*(C'\fR are parsed at compile time, to save having to parse them at
566 run time.
567 .IP \(bu 4
568 Array and hash lookups (especially nested ones) that use only constants
569 or simple variables as keys, are now considerably faster. See
570 "Internal Changes" for more details.
571 .IP \(bu 4
572 \&\f(CW\*(C`(...)x1\*(C'\fR, \f(CW\*(C`("constant")x0\*(C'\fR and \f(CW\*(C`($scalar)x0\*(C'\fR are now optimised in list
573 context. If the right-hand argument is a constant 1, the repetition
574 operator disappears. If the right-hand argument is a constant 0, the whole
575 expression is optimised to the empty list, so long as the left-hand
576 argument is a simple scalar or constant. (That is, \f(CW\*(C`(foo())x0\*(C'\fR is not
577 subject to this optimisation.)
578 .IP \(bu 4
579 \&\f(CW\*(C`substr\*(C'\fR assignment is now optimised into 4\-argument \f(CW\*(C`substr\*(C'\fR at the end
580 of a subroutine (or as the argument to \f(CW\*(C`return\*(C'\fR). Previously, this
581 optimisation only happened in void context.
582 .IP \(bu 4
583 In \f(CW"\eL..."\fR, \f(CW"\eQ..."\fR, etc., the extra "stringify" op is now optimised
584 away, making these just as fast as \f(CW\*(C`lcfirst\*(C'\fR, \f(CW\*(C`quotemeta\*(C'\fR, etc.
585 .IP \(bu 4
586 Assignment to an empty list is now sometimes faster. In particular, it
587 never calls \f(CW\*(C`FETCH\*(C'\fR on tied arguments on the right-hand side, whereas it
588 used to sometimes.
589 .IP \(bu 4
590 There is a performance improvement of up to 20% when \f(CW\*(C`length\*(C'\fR is applied to
591 a non-magical, non-tied string, and either \f(CW\*(C`use bytes\*(C'\fR is in scope or the
592 string doesn't use UTF\-8 internally.
593 .IP \(bu 4
594 On most perl builds with 64\-bit integers, memory usage for non-magical,
595 non-tied scalars containing only a floating point value has been reduced
596 by between 8 and 32 bytes, depending on OS.
597 .IP \(bu 4
598 In \f(CW\*(C`@array = split\*(C'\fR, the assignment can be optimized away, so that \f(CW\*(C`split\*(C'\fR
599 writes directly to the array. This optimisation was happening only for
600 package arrays other than \f(CW@_\fR, and only sometimes. Now this
601 optimisation happens almost all the time.
602 .IP \(bu 4
603 \&\f(CW\*(C`join\*(C'\fR is now subject to constant folding. So for example
604 \&\f(CW\*(C`join\ "\-",\ "a",\ "b"\*(C'\fR is converted at compile-time to \f(CW"a\-b"\fR.
605 Moreover, \f(CW\*(C`join\*(C'\fR with a scalar or constant for the separator and a
606 single-item list to join is simplified to a stringification, and the
607 separator doesn't even get evaluated.
608 .IP \(bu 4
609 \&\f(CWqq(@array)\fR is implemented using two ops: a stringify op and a join op.
610 If the \f(CW\*(C`qq\*(C'\fR contains nothing but a single array, the stringification is
611 optimized away.
612 .IP \(bu 4
613 \&\f(CW\*(C`our\ $var\*(C'\fR and \f(CW\*(C`our($s,@a,%h)\*(C'\fR in void context are no longer evaluated at
614 run time. Even a whole sequence of \f(CW\*(C`our\ $foo;\*(C'\fR statements will simply be
615 skipped over. The same applies to \f(CW\*(C`state\*(C'\fR variables.
616 .IP \(bu 4
617 Many internal functions have been refactored to improve performance and reduce
618 their memory footprints.
619 [GH #13659] <https://github.com/Perl/perl5/issues/13659>
620 [GH #13856] <https://github.com/Perl/perl5/issues/13856>
621 [GH #13874] <https://github.com/Perl/perl5/issues/13874>
622 .IP \(bu 4
623 \&\f(CW\*(C`\-T\*(C'\fR and \f(CW\*(C`\-B\*(C'\fR filetests will return sooner when an empty file is detected.
624 [GH #13686] <https://github.com/Perl/perl5/issues/13686>
625 .IP \(bu 4
626 Hash lookups where the key is a constant are faster.
627 .IP \(bu 4
628 Subroutines with an empty prototype and a body containing just \f(CW\*(C`undef\*(C'\fR are now
629 eligible for inlining.
630 [GH #14077] <https://github.com/Perl/perl5/issues/14077>
631 .IP \(bu 4
632 Subroutines in packages no longer need to be stored in typeglobs:
633 declaring a subroutine will now put a simple sub reference directly in the
634 stash if possible, saving memory. The typeglob still notionally exists,
635 so accessing it will cause the stash entry to be upgraded to a typeglob
636 (\fIi.e.\fR this is just an internal implementation detail).
637 This optimization does not currently apply to XSUBs or exported
638 subroutines, and method calls will undo it, since they cache things in
639 typeglobs.
640 [GH #13392] <https://github.com/Perl/perl5/issues/13392>
641 .IP \(bu 4
642 The functions \f(CWutf8::native_to_unicode()\fR and \f(CWutf8::unicode_to_native()\fR
643 (see utf8) are now optimized out on ASCII platforms. There is now not even
644 a minimal performance hit in writing code portable between ASCII and EBCDIC
645 platforms.
646 .IP \(bu 4
647 Win32 Perl uses 8 KB less of per-process memory than before for every perl
648 process, because some data is now memory mapped from disk and shared
649 between processes from the same perl binary.
650 .SH "Modules and Pragmata"
651 .IX Header "Modules and Pragmata"
652 .SS "Updated Modules and Pragmata"
653 .IX Subsection "Updated Modules and Pragmata"
654 Many of the libraries distributed with perl have been upgraded since v5.20.0.
655 For a complete list of changes, run:
656 .PP
657 .Vb 1
658 \& corelist \-\-diff 5.20.0 5.22.0
659 .Ve
660 .PP
661 You can substitute your favorite version in place of 5.20.0, too.
662 .PP
663 Some notable changes include:
664 .IP \(bu 4
665 Archive::Tar has been upgraded to version 2.04.
666 .Sp
667 Tests can now be run in parallel.
668 .IP \(bu 4
669 attributes has been upgraded to version 0.27.
670 .Sp
671 The usage of \f(CW\*(C`memEQs\*(C'\fR in the XS has been corrected.
672 [GH #14072] <https://github.com/Perl/perl5/issues/14072>
673 .Sp
674 Avoid reading beyond the end of a buffer. [perl #122629]
675 .IP \(bu 4
676 B has been upgraded to version 1.58.
677 .Sp
678 It provides a new \f(CW\*(C`B::safename\*(C'\fR function, based on the existing
679 \&\f(CW\*(C`B::GV\->SAFENAME\*(C'\fR, that converts \f(CW\*(C`\ecOPEN\*(C'\fR to \f(CW\*(C`^OPEN\*(C'\fR.
680 .Sp
681 Nulled COPs are now of class \f(CW\*(C`B::COP\*(C'\fR, rather than \f(CW\*(C`B::OP\*(C'\fR.
682 .Sp
683 \&\f(CW\*(C`B::REGEXP\*(C'\fR objects now provide a \f(CW\*(C`qr_anoncv\*(C'\fR method for accessing the
684 implicit CV associated with \f(CW\*(C`qr//\*(C'\fR things containing code blocks, and a
685 \&\f(CW\*(C`compflags\*(C'\fR method that returns the pertinent flags originating from the
686 \&\f(CW\*(C`qr//blahblah\*(C'\fR op.
687 .Sp
688 \&\f(CW\*(C`B::PMOP\*(C'\fR now provides a \f(CW\*(C`pmregexp\*(C'\fR method returning a \f(CW\*(C`B::REGEXP\*(C'\fR object.
689 Two new classes, \f(CW\*(C`B::PADNAME\*(C'\fR and \f(CW\*(C`B::PADNAMELIST\*(C'\fR, have been introduced.
690 .Sp
691 A bug where, after an ithread creation or pseudofork, special/immortal SVs in
692 the child ithread/pseudoprocess did not have the correct class of
693 \&\f(CW\*(C`B::SPECIAL\*(C'\fR, has been fixed.
694 The \f(CW\*(C`id\*(C'\fR and \f(CW\*(C`outid\*(C'\fR PADLIST methods have been added.
695 .IP \(bu 4
696 B::Concise has been upgraded to version 0.996.
697 .Sp
698 Null ops that are part of the execution chain are now given sequence
699 numbers.
700 .Sp
701 Private flags for nulled ops are now dumped with mnemonics as they would be
702 for the non-nulled counterparts.
703 .IP \(bu 4
704 B::Deparse has been upgraded to version 1.35.
705 .Sp
706 It now deparses \f(CW\*(C`+sub : attr { ... }\*(C'\fR correctly at the start of a
707 statement. Without the initial \f(CW\*(C`+\*(C'\fR, \f(CW\*(C`sub\*(C'\fR would be a statement label.
708 .Sp
709 \&\f(CW\*(C`BEGIN\*(C'\fR blocks are now emitted in the right place most of the time, but
710 the change unfortunately introduced a regression, in that \f(CW\*(C`BEGIN\*(C'\fR blocks
711 occurring just before the end of the enclosing block may appear below it
712 instead.
713 .Sp
714 \&\f(CW\*(C`B::Deparse\*(C'\fR no longer puts erroneous \f(CW\*(C`local\*(C'\fR here and there, such as for
715 \&\f(CW\*(C`LIST = tr/a//d\*(C'\fR. [perl #119815]
716 .Sp
717 Adjacent \f(CW\*(C`use\*(C'\fR statements are no longer accidentally nested if one
718 contains a \f(CW\*(C`do\*(C'\fR block. [perl #115066]
719 .Sp
720 Parenthesised arrays in lists passed to \f(CW\*(C`\e\*(C'\fR are now correctly deparsed
721 with parentheses (\fIe.g.\fR, \f(CW\*(C`\e(@a, (@b), @c)\*(C'\fR now retains the parentheses
722 around \f(CW@b\fR), thus preserving the flattening behavior of referenced
723 parenthesised arrays. Formerly, it only worked for one array: \f(CW\*(C`\e(@a)\*(C'\fR.
724 .Sp
725 \&\f(CW\*(C`local our\*(C'\fR is now deparsed correctly, with the \f(CW\*(C`our\*(C'\fR included.
726 .Sp
727 \&\f(CW\*(C`for($foo; !$bar; $baz) {...}\*(C'\fR was deparsed without the \f(CW\*(C`!\*(C'\fR (or \f(CW\*(C`not\*(C'\fR).
728 This has been fixed.
729 .Sp
730 Core keywords that conflict with lexical subroutines are now deparsed with
731 the \f(CW\*(C`CORE::\*(C'\fR prefix.
732 .Sp
733 \&\f(CW\*(C`foreach state $x (...) {...}\*(C'\fR now deparses correctly with \f(CW\*(C`state\*(C'\fR and
734 not \f(CW\*(C`my\*(C'\fR.
735 .Sp
736 \&\f(CW\*(C`our @array = split(...)\*(C'\fR now deparses correctly with \f(CW\*(C`our\*(C'\fR in those
737 cases where the assignment is optimized away.
738 .Sp
739 It now deparses \f(CWour(\fR\f(CILIST\fR\f(CW)\fR and typed lexical (\f(CW\*(C`my Dog $spot\*(C'\fR) correctly.
740 .Sp
741 Deparse \f(CW$#_\fR as that instead of as \f(CW$#{_}\fR.
742 [GH #14545] <https://github.com/Perl/perl5/issues/14545>
743 .Sp
744 BEGIN blocks at the end of the enclosing scope are now deparsed in the
745 right place. [perl #77452]
746 .Sp
747 BEGIN blocks were sometimes deparsed as _\|_ANON_\|_, but are now always called
748 BEGIN.
749 .Sp
750 Lexical subroutines are now fully deparsed. [perl #116553]
751 .Sp
752 \&\f(CW\*(C`Anything =~ y///r\*(C'\fR with \f(CW\*(C`/r\*(C'\fR no longer omits the left-hand operand.
753 .Sp
754 The op trees that make up regexp code blocks are now deparsed for real.
755 Formerly, the original string that made up the regular expression was used.
756 That caused problems with \f(CW\*(C`qr/(?{<<heredoc})/\*(C'\fR and multiline code blocks,
757 which were deparsed incorrectly. [perl #123217] [perl #115256]
758 .Sp
759 \&\f(CW$;\fR at the end of a statement no longer loses its semicolon.
760 [perl #123357]
761 .Sp
762 Some cases of subroutine declarations stored in the stash in shorthand form
763 were being omitted.
764 .Sp
765 Non-ASCII characters are now consistently escaped in strings, instead of
766 some of the time. (There are still outstanding problems with regular
767 expressions and identifiers that have not been fixed.)
768 .Sp
769 When prototype sub calls are deparsed with \f(CW\*(C`&\*(C'\fR (\fIe.g.\fR, under the \fB\-P\fR
770 option), \f(CW\*(C`scalar\*(C'\fR is now added where appropriate, to force the scalar
771 context implied by the prototype.
772 .Sp
773 \&\f(CW\*(C`require(foo())\*(C'\fR, \f(CW\*(C`do(foo())\*(C'\fR, \f(CW\*(C`goto(foo())\*(C'\fR and similar constructs with
774 loop controls are now deparsed correctly. The outer parentheses are not
775 optional.
776 .Sp
777 Whitespace is no longer escaped in regular expressions, because it was
778 getting erroneously escaped within \f(CW\*(C`(?x:...)\*(C'\fR sections.
779 .Sp
780 \&\f(CW\*(C`sub foo { foo() }\*(C'\fR is now deparsed with those mandatory parentheses.
781 .Sp
782 \&\f(CW\*(C`/@array/\*(C'\fR is now deparsed as a regular expression, and not just
783 \&\f(CW@array\fR.
784 .Sp
785 \&\f(CW\*(C`/@{\-}/\*(C'\fR, \f(CW\*(C`/@{+}/\*(C'\fR and \f(CW$#{1}\fR are now deparsed with the braces, which
786 are mandatory in these cases.
787 .Sp
788 In deparsing feature bundles, \f(CW\*(C`B::Deparse\*(C'\fR was emitting \f(CW\*(C`no feature;\*(C'\fR first
789 instead of \f(CW\*(C`no feature \*(Aq:all\*(Aq;\*(C'\fR. This has been fixed.
790 .Sp
791 \&\f(CW\*(C`chdir FH\*(C'\fR is now deparsed without quotation marks.
792 .Sp
793 \&\f(CW\*(C`\emy @a\*(C'\fR is now deparsed without parentheses. (Parenthese would flatten
794 the array.)
795 .Sp
796 \&\f(CW\*(C`system\*(C'\fR and \f(CW\*(C`exec\*(C'\fR followed by a block are now deparsed correctly.
797 Formerly there was an erroneous \f(CW\*(C`do\*(C'\fR before the block.
798 .Sp
799 \&\f(CW\*(C`use constant QR => qr/.../flags\*(C'\fR followed by \f(CW\*(C`"" =~ QR\*(C'\fR is no longer
800 without the flags.
801 .Sp
802 Deparsing \f(CW\*(C`BEGIN { undef &foo }\*(C'\fR with the \fB\-w\fR switch enabled started to
803 emit 'uninitialized' warnings in Perl 5.14. This has been fixed.
804 .Sp
805 Deparsing calls to subs with a \f(CW\*(C`(;+)\*(C'\fR prototype resulted in an infinite
806 loop. The \f(CW\*(C`(;$\*(C'\fR) \f(CW\*(C`(_)\*(C'\fR and \f(CW\*(C`(;_)\*(C'\fR prototypes were given the wrong
807 precedence, causing \f(CWfoo($a<$b)\fR to be deparsed without the parentheses.
808 .Sp
809 Deparse now provides a defined state sub in inner subs.
810 .IP \(bu 4
811 B::Op_private has been added.
812 .Sp
813 B::Op_private provides detailed information about the flags used in the
814 \&\f(CW\*(C`op_private\*(C'\fR field of perl opcodes.
815 .IP \(bu 4
816 bigint, bignum, bigrat have been upgraded to version 0.39.
817 .Sp
818 Document in CAVEATS that using strings as numbers won't always invoke
819 the big number overloading, and how to invoke it. [rt.perl.org #123064]
820 .IP \(bu 4
821 Carp has been upgraded to version 1.36.
822 .Sp
823 \&\f(CW\*(C`Carp::Heavy\*(C'\fR now ignores version mismatches with Carp if Carp is newer
824 than 1.12, since \f(CW\*(C`Carp::Heavy\*(C'\fR's guts were merged into Carp at that
825 point.
826 [GH #13708] <https://github.com/Perl/perl5/issues/13708>
827 .Sp
828 Carp now handles non-ASCII platforms better.
829 .Sp
830 Off-by-one error fix for Perl < 5.14.
831 .IP \(bu 4
832 constant has been upgraded to version 1.33.
833 .Sp
834 It now accepts fully-qualified constant names, allowing constants to be defined
835 in packages other than the caller.
836 .IP \(bu 4
837 CPAN has been upgraded to version 2.11.
838 .Sp
839 Add support for \f(CWCwd::getdcwd()\fR and introduce workaround for a misbehavior
840 seen on Strawberry Perl 5.20.1.
841 .Sp
842 Fix \f(CWchdir()\fR after building dependencies bug.
843 .Sp
844 Introduce experimental support for plugins/hooks.
845 .Sp
846 Integrate the \f(CW\*(C`App::Cpan\*(C'\fR sources.
847 .Sp
848 Do not check recursion on optional dependencies.
849 .Sp
850 Sanity check \fIMETA.yml\fR to contain a hash.
851 [cpan #95271] <https://rt.cpan.org/Ticket/Display.html?id=95271>
852 .IP \(bu 4
853 CPAN::Meta::Requirements has been upgraded to version 2.132.
854 .Sp
855 Works around limitations in \f(CW\*(C`version::vpp\*(C'\fR detecting v\-string magic and adds
856 support for forthcoming ExtUtils::MakeMaker bootstrap \fIversion.pm\fR for
857 Perls older than 5.10.0.
858 .IP \(bu 4
859 Data::Dumper has been upgraded to version 2.158.
860 .Sp
861 Fixes CVE\-2014\-4330 by adding a configuration variable/option to limit
862 recursion when dumping deep data structures.
863 .Sp
864 Changes to resolve Coverity issues.
865 XS dumps incorrectly stored the name of code references stored in a
866 GLOB.
867 [GH #13911] <https://github.com/Perl/perl5/issues/13911>
868 .IP \(bu 4
869 DynaLoader has been upgraded to version 1.32.
870 .Sp
871 Remove \f(CW\*(C`dl_nonlazy\*(C'\fR global if unused in Dynaloader. [perl #122926]
872 .IP \(bu 4
873 Encode has been upgraded to version 2.72.
874 .Sp
875 \&\f(CW\*(C`piconv\*(C'\fR now has better error handling when the encoding name is nonexistent,
876 and a build breakage when upgrading Encode in perl\-5.8.2 and earlier has
877 been fixed.
878 .Sp
879 Building in C++ mode on Windows now works.
880 .IP \(bu 4
881 Errno has been upgraded to version 1.23.
882 .Sp
883 Add \f(CW\*(C`\-P\*(C'\fR to the preprocessor command-line on GCC 5. GCC added extra
884 line directives, breaking parsing of error code definitions. [rt.perl.org
885 #123784]
886 .IP \(bu 4
887 experimental has been upgraded to version 0.013.
888 .Sp
889 Hardcodes features for Perls older than 5.15.7.
890 .IP \(bu 4
891 ExtUtils::CBuilder has been upgraded to version 0.280221.
892 .Sp
893 Fixes a regression on Android.
894 [GH #14064] <https://github.com/Perl/perl5/issues/14064>
895 .IP \(bu 4
896 ExtUtils::Manifest has been upgraded to version 1.70.
897 .Sp
898 Fixes a bug with \f(CWmaniread()\fR's handling of quoted filenames and improves
899 \&\f(CWmanifind()\fR to follow symlinks.
900 [GH #14003] <https://github.com/Perl/perl5/issues/14003>
901 .IP \(bu 4
902 ExtUtils::ParseXS has been upgraded to version 3.28.
903 .Sp
904 Only declare \f(CW\*(C`file\*(C'\fR unused if we actually define it.
905 Improve generated \f(CW\*(C`RETVAL\*(C'\fR code generation to avoid repeated
906 references to \f(CWST(0)\fR. [perl #123278]
907 Broaden and document the \f(CW\*(C`/OBJ$/\*(C'\fR to \f(CW\*(C`/REF$/\*(C'\fR typemap optimization
908 for the \f(CW\*(C`DESTROY\*(C'\fR method. [perl #123418]
909 .IP \(bu 4
910 Fcntl has been upgraded to version 1.13.
911 .Sp
912 Add support for the Linux pipe buffer size \f(CWfcntl()\fR commands.
913 .IP \(bu 4
914 File::Find has been upgraded to version 1.29.
915 .Sp
916 \&\f(CWfind()\fR and \f(CWfinddepth()\fR will now warn if passed inappropriate or
917 misspelled options.
918 .IP \(bu 4
919 File::Glob has been upgraded to version 1.24.
920 .Sp
921 Avoid \f(CWSvIV()\fR expanding to call \f(CWget_sv()\fR three times in a few
922 places. [perl #123606]
923 .IP \(bu 4
924 HTTP::Tiny has been upgraded to version 0.054.
925 .Sp
926 \&\f(CW\*(C`keep_alive\*(C'\fR is now fork-safe and thread-safe.
927 .IP \(bu 4
928 IO has been upgraded to version 1.35.
929 .Sp
930 The XS implementation has been fixed for the sake of older Perls.
931 .IP \(bu 4
932 IO::Socket has been upgraded to version 1.38.
933 .Sp
934 Document the limitations of the \f(CWconnected()\fR method. [perl #123096]
935 .IP \(bu 4
936 IO::Socket::IP has been upgraded to version 0.37.
937 .Sp
938 A better fix for subclassing \f(CWconnect()\fR.
939 [cpan #95983] <https://rt.cpan.org/Ticket/Display.html?id=95983>
940 [cpan #97050] <https://rt.cpan.org/Ticket/Display.html?id=97050>
941 .Sp
942 Implements Timeout for \f(CWconnect()\fR.
943 [cpan #92075] <https://rt.cpan.org/Ticket/Display.html?id=92075>
944 .IP \(bu 4
945 The libnet collection of modules has been upgraded to version 3.05.
946 .Sp
947 Support for IPv6 and SSL to \f(CW\*(C`Net::FTP\*(C'\fR, \f(CW\*(C`Net::NNTP\*(C'\fR, \f(CW\*(C`Net::POP3\*(C'\fR and \f(CW\*(C`Net::SMTP\*(C'\fR.
948 Improvements in \f(CW\*(C`Net::SMTP\*(C'\fR authentication.
949 .IP \(bu 4
950 Locale::Codes has been upgraded to version 3.34.
951 .Sp
952 Fixed a bug in the scripts used to extract data from spreadsheets that
953 prevented the SHP currency code from being found.
954 [cpan #94229] <https://rt.cpan.org/Ticket/Display.html?id=94229>
955 .Sp
956 New codes have been added.
957 .IP \(bu 4
958 Math::BigInt has been upgraded to version 1.9997.
959 .Sp
960 Synchronize POD changes from the CPAN release.
961 \&\f(CW\*(C`Math::BigFloat\->blog(x)\*(C'\fR would sometimes return \f(CWblog(2*x)\fR when
962 the accuracy was greater than 70 digits.
963 The result of \f(CW\*(C`Math::BigFloat\->bdiv()\*(C'\fR in list context now
964 satisfies \f(CW\*(C`x = quotient * divisor + remainder\*(C'\fR.
965 .Sp
966 Correct handling of subclasses.
967 [cpan #96254] <https://rt.cpan.org/Ticket/Display.html?id=96254>
968 [cpan #96329] <https://rt.cpan.org/Ticket/Display.html?id=96329>
969 .IP \(bu 4
970 Module::Metadata has been upgraded to version 1.000026.
971 .Sp
972 Support installations on older perls with an ExtUtils::MakeMaker earlier
973 than 6.63_03
974 .IP \(bu 4
975 overload has been upgraded to version 1.26.
976 .Sp
977 A redundant \f(CW\*(C`ref $sub\*(C'\fR check has been removed.
978 .IP \(bu 4
979 The PathTools module collection has been upgraded to version 3.56.
980 .Sp
981 A warning from the \fBgcc\fR compiler is now avoided when building the XS.
982 .Sp
983 Don't turn leading \f(CW\*(C`//\*(C'\fR into \f(CW\*(C`/\*(C'\fR on Cygwin. [perl #122635]
984 .IP \(bu 4
985 perl5db.pl has been upgraded to version 1.49.
986 .Sp
987 The debugger would cause an assertion failure.
988 [GH #14605] <https://github.com/Perl/perl5/issues/14605>
989 .Sp
990 \&\f(CWfork()\fR in the debugger under \f(CW\*(C`tmux\*(C'\fR will now create a new window for
991 the forked process. [GH #13602] <https://github.com/Perl/perl5/issues/13602>
992 .Sp
993 The debugger now saves the current working directory on startup and
994 restores it when you restart your program with \f(CW\*(C`R\*(C'\fR or \f(CW\*(C`rerun\*(C'\fR.
995 [GH #13691] <https://github.com/Perl/perl5/issues/13691>
996 .IP \(bu 4
997 PerlIO::scalar has been upgraded to version 0.22.
998 .Sp
999 Reading from a position well past the end of the scalar now correctly
1000 returns end of file. [perl #123443]
1001 .Sp
1002 Seeking to a negative position still fails, but no longer leaves the
1003 file position set to a negation location.
1004 .Sp
1005 \&\f(CWeof()\fR on a \f(CW\*(C`PerlIO::scalar\*(C'\fR handle now properly returns true when
1006 the file position is past the 2GB mark on 32\-bit systems.
1007 .Sp
1008 Attempting to write at file positions impossible for the platform now
1009 fail early rather than wrapping at 4GB.
1010 .IP \(bu 4
1011 Pod::Perldoc has been upgraded to version 3.25.
1012 .Sp
1013 Filehandles opened for reading or writing now have \f(CW:encoding(UTF\-8)\fR set.
1014 [cpan #98019] <https://rt.cpan.org/Ticket/Display.html?id=98019>
1015 .IP \(bu 4
1016 POSIX has been upgraded to version 1.53.
1017 .Sp
1018 The C99 math functions and constants (for example \f(CW\*(C`acosh\*(C'\fR, \f(CW\*(C`isinf\*(C'\fR, \f(CW\*(C`isnan\*(C'\fR, \f(CW\*(C`round\*(C'\fR,
1019 \&\f(CW\*(C`trunc\*(C'\fR; \f(CW\*(C`M_E\*(C'\fR, \f(CW\*(C`M_SQRT2\*(C'\fR, \f(CW\*(C`M_PI\*(C'\fR) have been added.
1020 .Sp
1021 \&\f(CWPOSIX::tmpnam()\fR now produces a deprecation warning. [perl #122005]
1022 .IP \(bu 4
1023 Safe has been upgraded to version 2.39.
1024 .Sp
1025 \&\f(CW\*(C`reval\*(C'\fR was not propagating void context properly.
1026 .IP \(bu 4
1027 Scalar-List-Utils has been upgraded to version 1.41.
1028 .Sp
1029 A new module, Sub::Util, has been added, containing functions related to
1030 CODE refs, including \f(CW\*(C`subname\*(C'\fR (inspired by \f(CW\*(C`Sub::Identity\*(C'\fR) and \f(CW\*(C`set_subname\*(C'\fR
1031 (copied and renamed from \f(CW\*(C`Sub::Name\*(C'\fR).
1032 The use of \f(CW\*(C`GetMagic\*(C'\fR in \f(CWList::Util::reduce()\fR has also been fixed.
1033 [cpan #63211] <https://rt.cpan.org/Ticket/Display.html?id=63211>
1034 .IP \(bu 4
1035 SDBM_File has been upgraded to version 1.13.
1036 .Sp
1037 Simplified the build process. [perl #123413]
1038 .IP \(bu 4
1039 Time::Piece has been upgraded to version 1.29.
1040 .Sp
1041 When pretty printing negative \f(CW\*(C`Time::Seconds\*(C'\fR, the "minus" is no longer lost.
1042 .IP \(bu 4
1043 Unicode::Collate has been upgraded to version 1.12.
1044 .Sp
1045 Version 0.67's improved discontiguous contractions is invalidated by default
1046 and is supported as a parameter \f(CW\*(C`long_contraction\*(C'\fR.
1047 .IP \(bu 4
1048 Unicode::Normalize has been upgraded to version 1.18.
1049 .Sp
1050 The XSUB implementation has been removed in favor of pure Perl.
1051 .IP \(bu 4
1052 Unicode::UCD has been upgraded to version 0.61.
1053 .Sp
1054 A new function \fBproperty_values()\fR
1055 has been added to return a given property's possible values.
1056 .Sp
1057 A new function \fBcharprop()\fR
1058 has been added to return the value of a given property for a given code
1059 point.
1060 .Sp
1061 A new function \fBcharprops_all()\fR
1062 has been added to return the values of all Unicode properties for a
1063 given code point.
1064 .Sp
1065 A bug has been fixed so that \fBpropaliases()\fR
1066 returns the correct short and long names for the Perl extensions where
1067 it was incorrect.
1068 .Sp
1069 A bug has been fixed so that
1070 \&\fBprop_value_aliases()\fR
1071 returns \f(CW\*(C`undef\*(C'\fR instead of a wrong result for properties that are Perl
1072 extensions.
1073 .Sp
1074 This module now works on EBCDIC platforms.
1075 .IP \(bu 4
1076 utf8 has been upgraded to version 1.17
1077 .Sp
1078 A mismatch between the documentation and the code in \f(CWutf8::downgrade()\fR
1079 was fixed in favor of the documentation. The optional second argument
1080 is now correctly treated as a perl boolean (true/false semantics) and
1081 not as an integer.
1082 .IP \(bu 4
1083 version has been upgraded to version 0.9909.
1084 .Sp
1085 Numerous changes. See the \fIChanges\fR file in the CPAN distribution for
1086 details.
1087 .IP \(bu 4
1088 Win32 has been upgraded to version 0.51.
1089 .Sp
1090 \&\f(CWGetOSName()\fR now supports Windows 8.1, and building in C++ mode now works.
1091 .IP \(bu 4
1092 Win32API::File has been upgraded to version 0.1202
1093 .Sp
1094 Building in C++ mode now works.
1095 .IP \(bu 4
1096 XSLoader has been upgraded to version 0.20.
1097 .Sp
1098 Allow XSLoader to load modules from a different namespace.
1099 [perl #122455]
1100 .SS "Removed Modules and Pragmata"
1101 .IX Subsection "Removed Modules and Pragmata"
1102 The following modules (and associated modules) have been removed from the core
1103 perl distribution:
1104 .IP \(bu 4
1105 CGI
1106 .IP \(bu 4
1107 Module::Build
1108 .SH Documentation
1109 .IX Header "Documentation"
1110 .SS "New Documentation"
1111 .IX Subsection "New Documentation"
1112 \fIperlunicook\fR
1113 .IX Subsection "perlunicook"
1114 .PP
1115 This document, by Tom Christiansen, provides examples of handling Unicode in
1116 Perl.
1117 .SS "Changes to Existing Documentation"
1118 .IX Subsection "Changes to Existing Documentation"
1119 \fIperlaix\fR
1120 .IX Subsection "perlaix"
1121 .IP \(bu 4
1122 A note on long doubles has been added.
1123 .PP
1124 \fIperlapi\fR
1125 .IX Subsection "perlapi"
1126 .IP \(bu 4
1127 Note that \f(CW\*(C`SvSetSV\*(C'\fR doesn't do set magic.
1128 .IP \(bu 4
1129 \&\f(CW\*(C`sv_usepvn_flags\*(C'\fR \- fix documentation to mention the use of \f(CW\*(C`Newx\*(C'\fR instead of
1130 \&\f(CW\*(C`malloc\*(C'\fR.
1131 .Sp
1132 [GH #13835] <https://github.com/Perl/perl5/issues/13835>
1133 .IP \(bu 4
1134 Clarify where \f(CW\*(C`NUL\*(C'\fR may be embedded or is required to terminate a string.
1135 .IP \(bu 4
1136 Some documentation that was previously missing due to formatting errors is
1137 now included.
1138 .IP \(bu 4
1139 Entries are now organized into groups rather than by the file where they
1140 are found.
1141 .IP \(bu 4
1142 Alphabetical sorting of entries is now done consistently (automatically
1143 by the POD generator) to make entries easier to find when scanning.
1144 .PP
1145 \fIperldata\fR
1146 .IX Subsection "perldata"
1147 .IP \(bu 4
1148 The syntax of single-character variable names has been brought
1149 up-to-date and more fully explained.
1150 .IP \(bu 4
1151 Hexadecimal floating point numbers are described, as are infinity and
1152 NaN.
1153 .PP
1154 \fIperlebcdic\fR
1155 .IX Subsection "perlebcdic"
1156 .IP \(bu 4
1157 This document has been significantly updated in the light of recent
1158 improvements to EBCDIC support.
1159 .PP
1160 \fIperlfilter\fR
1161 .IX Subsection "perlfilter"
1162 .IP \(bu 4
1163 Added a LIMITATIONS section.
1164 .PP
1165 \fIperlfunc\fR
1166 .IX Subsection "perlfunc"
1167 .IP \(bu 4
1168 Mention that \f(CWstudy()\fR is currently a no-op.
1169 .IP \(bu 4
1170 Calling \f(CW\*(C`delete\*(C'\fR or \f(CW\*(C`exists\*(C'\fR on array values is now described as "strongly
1171 discouraged" rather than "deprecated".
1172 .IP \(bu 4
1173 Improve documentation of \f(CW\*(C`our\*(C'\fR.
1174 .IP \(bu 4
1175 \&\f(CW\*(C`\-l\*(C'\fR now notes that it will return false if symlinks aren't supported by the
1176 file system.
1177 [GH #13695] <https://github.com/Perl/perl5/issues/13695>
1178 .IP \(bu 4
1179 Note that \f(CW\*(C`exec LIST\*(C'\fR and \f(CW\*(C`system LIST\*(C'\fR may fall back to the shell on
1180 Win32. Only the indirect-object syntax \f(CW\*(C`exec PROGRAM LIST\*(C'\fR and
1181 \&\f(CW\*(C`system PROGRAM LIST\*(C'\fR will reliably avoid using the shell.
1182 .Sp
1183 This has also been noted in perlport.
1184 .Sp
1185 [GH #13907] <https://github.com/Perl/perl5/issues/13907>
1186 .PP
1187 \fIperlguts\fR
1188 .IX Subsection "perlguts"
1189 .IP \(bu 4
1190 The OOK example has been updated to account for COW changes and a change in the
1191 storage of the offset.
1192 .IP \(bu 4
1193 Details on C level symbols and libperl.t added.
1194 .IP \(bu 4
1195 Information on Unicode handling has been added
1196 .IP \(bu 4
1197 Information on EBCDIC handling has been added
1198 .PP
1199 \fIperlhack\fR
1200 .IX Subsection "perlhack"
1201 .IP \(bu 4
1202 A note has been added about running on platforms with non-ASCII
1203 character sets
1204 .IP \(bu 4
1205 A note has been added about performance testing
1206 .PP
1207 \fIperlhacktips\fR
1208 .IX Subsection "perlhacktips"
1209 .IP \(bu 4
1210 Documentation has been added illustrating the perils of assuming that
1211 there is no change to the contents of static memory pointed to by the
1212 return values of Perl's wrappers for C library functions.
1213 .IP \(bu 4
1214 Replacements for \f(CW\*(C`tmpfile\*(C'\fR, \f(CW\*(C`atoi\*(C'\fR, \f(CW\*(C`strtol\*(C'\fR, and \f(CW\*(C`strtoul\*(C'\fR are now
1215 recommended.
1216 .IP \(bu 4
1217 Updated documentation for the \f(CW\*(C`test.valgrind\*(C'\fR \f(CW\*(C`make\*(C'\fR target.
1218 [GH #13658] <https://github.com/Perl/perl5/issues/13658>
1219 .IP \(bu 4
1220 Information is given about writing test files portably to non-ASCII
1221 platforms.
1222 .IP \(bu 4
1223 A note has been added about how to get a C language stack backtrace.
1224 .PP
1225 \fIperlhpux\fR
1226 .IX Subsection "perlhpux"
1227 .IP \(bu 4
1228 Note that the message "Redeclaration of "sendpath" with a different
1229 storage class specifier" is harmless.
1230 .PP
1231 \fIperllocale\fR
1232 .IX Subsection "perllocale"
1233 .IP \(bu 4
1234 Updated for the enhancements in v5.22, along with some clarifications.
1235 .PP
1236 \fIperlmodstyle\fR
1237 .IX Subsection "perlmodstyle"
1238 .IP \(bu 4
1239 Instead of pointing to the module list, we are now pointing to
1240 PrePAN <http://prepan.org/>.
1241 .PP
1242 \fIperlop\fR
1243 .IX Subsection "perlop"
1244 .IP \(bu 4
1245 Updated for the enhancements in v5.22, along with some clarifications.
1246 .PP
1247 \fIperlpodspec\fR
1248 .IX Subsection "perlpodspec"
1249 .IP \(bu 4
1250 The specification of the pod language is changing so that the default
1251 encoding of pods that aren't in UTF\-8 (unless otherwise indicated) is
1252 CP1252 instead of ISO 8859\-1 (Latin1).
1253 .PP
1254 \fIperlpolicy\fR
1255 .IX Subsection "perlpolicy"
1256 .IP \(bu 4
1257 We now have a code of conduct for the \fIp5p\fR mailing list, as documented
1258 in "STANDARDS OF CONDUCT" in perlpolicy.
1259 .IP \(bu 4
1260 The conditions for marking an experimental feature as non-experimental are now
1261 set out.
1262 .IP \(bu 4
1263 Clarification has been made as to what sorts of changes are permissible in
1264 maintenance releases.
1265 .PP
1266 \fIperlport\fR
1267 .IX Subsection "perlport"
1268 .IP \(bu 4
1269 Out-of-date VMS-specific information has been fixed and/or simplified.
1270 .IP \(bu 4
1271 Notes about EBCDIC have been added.
1272 .PP
1273 \fIperlre\fR
1274 .IX Subsection "perlre"
1275 .IP \(bu 4
1276 The description of the \f(CW\*(C`/x\*(C'\fR modifier has been clarified to note that
1277 comments cannot be continued onto the next line by escaping them; and
1278 there is now a list of all the characters that are considered whitespace
1279 by this modifier.
1280 .IP \(bu 4
1281 The new \f(CW\*(C`/n\*(C'\fR modifier is described.
1282 .IP \(bu 4
1283 A note has been added on how to make bracketed character class ranges
1284 portable to non-ASCII machines.
1285 .PP
1286 \fIperlrebackslash\fR
1287 .IX Subsection "perlrebackslash"
1288 .IP \(bu 4
1289 Added documentation of \f(CW\*(C`\eb{sb}\*(C'\fR, \f(CW\*(C`\eb{wb}\*(C'\fR, \f(CW\*(C`\eb{gcb}\*(C'\fR, and \f(CW\*(C`\eb{g}\*(C'\fR.
1290 .PP
1291 \fIperlrecharclass\fR
1292 .IX Subsection "perlrecharclass"
1293 .IP \(bu 4
1294 Clarifications have been added to "Character Ranges" in perlrecharclass
1295 to the effect \f(CW\*(C`[A\-Z]\*(C'\fR, \f(CW\*(C`[a\-z]\*(C'\fR, \f(CW\*(C`[0\-9]\*(C'\fR and
1296 any subranges thereof in regular expression bracketed character classes
1297 are guaranteed to match exactly what a naive English speaker would
1298 expect them to match, even on platforms (such as EBCDIC) where perl
1299 has to do extra work to accomplish this.
1300 .IP \(bu 4
1301 The documentation of Bracketed Character Classes has been expanded to cover the
1302 improvements in \f(CW\*(C`qr/[\eN{named sequence}]/\*(C'\fR (see under "Selected Bug Fixes").
1303 .PP
1304 \fIperlref\fR
1305 .IX Subsection "perlref"
1306 .IP \(bu 4
1307 A new section has been added
1308 Assigning to References
1309 .PP
1310 \fIperlsec\fR
1311 .IX Subsection "perlsec"
1312 .IP \(bu 4
1313 Comments added on algorithmic complexity and tied hashes.
1314 .PP
1315 \fIperlsyn\fR
1316 .IX Subsection "perlsyn"
1317 .IP \(bu 4
1318 An ambiguity in the documentation of the \f(CW\*(C`...\*(C'\fR statement has been corrected.
1319 [GH #14054] <https://github.com/Perl/perl5/issues/14054>
1320 .IP \(bu 4
1321 The empty conditional in \f(CW\*(C`for\*(C'\fR and \f(CW\*(C`while\*(C'\fR is now documented
1322 in perlsyn.
1323 .PP
1324 \fIperlunicode\fR
1325 .IX Subsection "perlunicode"
1326 .IP \(bu 4
1327 This has had extensive revisions to bring it up-to-date with current
1328 Unicode support and to make it more readable. Notable is that Unicode
1329 7.0 changed what it should do with non-characters. Perl retains the old
1330 way of handling for reasons of backward compatibility. See
1331 "Noncharacter code points" in perlunicode.
1332 .PP
1333 \fIperluniintro\fR
1334 .IX Subsection "perluniintro"
1335 .IP \(bu 4
1336 Advice for how to make sure your strings and regular expression patterns are
1337 interpreted as Unicode has been updated.
1338 .PP
1339 \fIperlvar\fR
1340 .IX Subsection "perlvar"
1341 .IP \(bu 4
1342 \&\f(CW$]\fR is no longer listed as being deprecated. Instead, discussion has
1343 been added on the advantages and disadvantages of using it versus
1344 \&\f(CW$^V\fR. \f(CW$OLD_PERL_VERSION\fR was re-added to the documentation as the long
1345 form of \f(CW$]\fR.
1346 .IP \(bu 4
1347 \&\f(CW\*(C`${^ENCODING}\*(C'\fR is now marked as deprecated.
1348 .IP \(bu 4
1349 The entry for \f(CW\*(C`%^H\*(C'\fR has been clarified to indicate it can only handle
1350 simple values.
1351 .PP
1352 \fIperlvms\fR
1353 .IX Subsection "perlvms"
1354 .IP \(bu 4
1355 Out-of-date and/or incorrect material has been removed.
1356 .IP \(bu 4
1357 Updated documentation on environment and shell interaction in VMS.
1358 .PP
1359 \fIperlxs\fR
1360 .IX Subsection "perlxs"
1361 .IP \(bu 4
1362 Added a discussion of locale issues in XS code.
1363 .SH Diagnostics
1364 .IX Header "Diagnostics"
1365 The following additions or changes have been made to diagnostic output,
1366 including warnings and fatal error messages. For the complete list of
1367 diagnostic messages, see perldiag.
1368 .SS "New Diagnostics"
1369 .IX Subsection "New Diagnostics"
1370 \fINew Errors\fR
1371 .IX Subsection "New Errors"
1372 .IP \(bu 4
1373 Bad symbol for scalar
1374 .Sp
1375 (P) An internal request asked to add a scalar entry to something that
1376 wasn't a symbol table entry.
1377 .IP \(bu 4
1378 Can't use a hash as a reference
1379 .Sp
1380 (F) You tried to use a hash as a reference, as in
1381 \&\f(CW\*(C`%foo\->{"bar"}\*(C'\fR or \f(CW\*(C`%$ref\->{"hello"}\*(C'\fR. Versions of perl <= 5.6.1
1382 used to allow this syntax, but shouldn't have.
1383 .IP \(bu 4
1384 Can't use an array as a reference
1385 .Sp
1386 (F) You tried to use an array as a reference, as in
1387 \&\f(CW\*(C`@foo\->[23]\*(C'\fR or \f(CW\*(C`@$ref\->[99]\*(C'\fR. Versions of perl <= 5.6.1 used to
1388 allow this syntax, but shouldn't have.
1389 .IP \(bu 4
1390 Can't use 'defined(@array)' (Maybe you should just omit the \fBdefined()\fR?)
1391 .Sp
1392 (F) \f(CWdefined()\fR is not useful on arrays because it
1393 checks for an undefined \fIscalar\fR value. If you want to see if the
1394 array is empty, just use \f(CW\*(C`if\ (@array)\ {\ #\ not\ empty\ }\*(C'\fR for example.
1395 .IP \(bu 4
1396 Can't use 'defined(%hash)' (Maybe you should just omit the \fBdefined()\fR?)
1397 .Sp
1398 (F) \f(CWdefined()\fR is not usually right on hashes.
1399 .Sp
1400 Although \f(CW\*(C`defined\ %hash\*(C'\fR is false on a plain not-yet-used hash, it
1401 becomes true in several non-obvious circumstances, including iterators,
1402 weak references, stash names, even remaining true after \f(CW\*(C`undef\ %hash\*(C'\fR.
1403 These things make \f(CW\*(C`defined\ %hash\*(C'\fR fairly useless in practice, so it now
1404 generates a fatal error.
1405 .Sp
1406 If a check for non-empty is what you wanted then just put it in boolean
1407 context (see "Scalar values" in perldata):
1408 .Sp
1409 .Vb 3
1410 \& if (%hash) {
1411 \& # not empty
1412 \& }
1413 .Ve
1414 .Sp
1415 If you had \f(CW\*(C`defined\ %Foo::Bar::QUUX\*(C'\fR to check whether such a package
1416 variable exists then that's never really been reliable, and isn't
1417 a good way to enquire about the features of a package, or whether
1418 it's loaded, etc.
1419 .IP \(bu 4
1420 Cannot chr \f(CW%f\fR
1421 .Sp
1422 (F) You passed an invalid number (like an infinity or not-a-number) to
1423 \&\f(CW\*(C`chr\*(C'\fR.
1424 .IP \(bu 4
1425 Cannot compress \f(CW%f\fR in pack
1426 .Sp
1427 (F) You tried converting an infinity or not-a-number to an unsigned
1428 character, which makes no sense.
1429 .IP \(bu 4
1430 Cannot pack \f(CW%f\fR with '%c'
1431 .Sp
1432 (F) You tried converting an infinity or not-a-number to a character,
1433 which makes no sense.
1434 .IP \(bu 4
1435 Cannot print \f(CW%f\fR with '%c'
1436 .Sp
1437 (F) You tried printing an infinity or not-a-number as a character (\f(CW%c\fR),
1438 which makes no sense. Maybe you meant \f(CW\*(Aq%s\*(Aq\fR, or just stringifying it?
1439 .IP \(bu 4
1440 charnames alias definitions may not contain a sequence of multiple spaces
1441 .Sp
1442 (F) You defined a character name which had multiple space
1443 characters in a row. Change them to single spaces. Usually these
1444 names are defined in the \f(CW\*(C`:alias\*(C'\fR import argument to \f(CW\*(C`use charnames\*(C'\fR, but
1445 they could be defined by a translator installed into \f(CW$^H{charnames}\fR.
1446 See "CUSTOM ALIASES" in charnames.
1447 .IP \(bu 4
1448 charnames alias definitions may not contain trailing white-space
1449 .Sp
1450 (F) You defined a character name which ended in a space
1451 character. Remove the trailing space(s). Usually these names are
1452 defined in the \f(CW\*(C`:alias\*(C'\fR import argument to \f(CW\*(C`use charnames\*(C'\fR, but they
1453 could be defined by a translator installed into \f(CW$^H{charnames}\fR.
1454 See "CUSTOM ALIASES" in charnames.
1455 .IP \(bu 4
1456 :const is not permitted on named subroutines
1457 .Sp
1458 (F) The \f(CW\*(C`const\*(C'\fR attribute causes an anonymous subroutine to be run and
1459 its value captured at the time that it is cloned. Named subroutines are
1460 not cloned like this, so the attribute does not make sense on them.
1461 .IP \(bu 4
1462 Hexadecimal float: internal error
1463 .Sp
1464 (F) Something went horribly bad in hexadecimal float handling.
1465 .IP \(bu 4
1466 Hexadecimal float: unsupported long double format
1467 .Sp
1468 (F) You have configured Perl to use long doubles but
1469 the internals of the long double format are unknown,
1470 therefore the hexadecimal float output is impossible.
1471 .IP \(bu 4
1472 Illegal suidscript
1473 .Sp
1474 (F) The script run under suidperl was somehow illegal.
1475 .IP \(bu 4
1476 In '(?...)', the '(' and '?' must be adjacent in regex; marked by <\-\-\ HERE in m/%s/
1477 .Sp
1478 (F) The two-character sequence \f(CW"(?"\fR in
1479 this context in a regular expression pattern should be an
1480 indivisible token, with nothing intervening between the \f(CW"("\fR
1481 and the \f(CW"?"\fR, but you separated them.
1482 .IP \(bu 4
1483 In '(*VERB...)', the '(' and '*' must be adjacent in regex; marked by <\-\-\ HERE in m/%s/
1484 .Sp
1485 (F) The two-character sequence \f(CW"(*"\fR in
1486 this context in a regular expression pattern should be an
1487 indivisible token, with nothing intervening between the \f(CW"("\fR
1488 and the \f(CW"*"\fR, but you separated them.
1489 .IP \(bu 4
1490 Invalid quantifier in {,} in regex; marked by <\-\- HERE in m/%s/
1491 .Sp
1492 (F) The pattern looks like a {min,max} quantifier, but the min or max could not
1493 be parsed as a valid number: either it has leading zeroes, or it represents
1494 too big a number to cope with. The <\-\-\ HERE shows where in the regular
1495 expression the problem was discovered. See perlre.
1496 .IP \(bu 4
1497 \&'%s' is an unknown bound type in regex
1498 .Sp
1499 (F) You used \f(CW\*(C`\eb{...}\*(C'\fR or \f(CW\*(C`\eB{...}\*(C'\fR and the \f(CW\*(C`...\*(C'\fR is not known to
1500 Perl. The current valid ones are given in
1501 "\eb{}, \eb, \eB{}, \eB" in perlrebackslash.
1502 .IP \(bu 4
1503 Missing or undefined argument to require
1504 .Sp
1505 (F) You tried to call \f(CW\*(C`require\*(C'\fR with no argument or with an undefined
1506 value as an argument. \f(CW\*(C`require\*(C'\fR expects either a package name or a
1507 file-specification as an argument. See "require" in perlfunc.
1508 .Sp
1509 Formerly, \f(CW\*(C`require\*(C'\fR with no argument or \f(CW\*(C`undef\*(C'\fR warned about a Null filename.
1510 .PP
1511 \fINew Warnings\fR
1512 .IX Subsection "New Warnings"
1513 .IP \(bu 4
1514 \&\eC is deprecated in regex
1515 .Sp
1516 (D deprecated) The \f(CW\*(C`/\eC/\*(C'\fR character class was deprecated in v5.20, and
1517 now emits a warning. It is intended that it will become an error in v5.24.
1518 This character class matches a single byte even if it appears within a
1519 multi-byte character, breaks encapsulation, and can corrupt UTF\-8
1520 strings.
1521 .IP \(bu 4
1522 "%s" is more clearly written simply as "%s" in regex; marked by <\-\- HERE in m/%s/
1523 .Sp
1524 (W regexp) (only under \f(CW\*(C`use\ re\ \*(Aqstrict\*(Aq\*(C'\fR or within \f(CW\*(C`(?[...])\*(C'\fR)
1525 .Sp
1526 You specified a character that has the given plainer way of writing it,
1527 and which is also portable to platforms running with different character
1528 sets.
1529 .IP \(bu 4
1530 Argument "%s" treated as 0 in increment (++)
1531 .Sp
1532 (W numeric) The indicated string was fed as an argument to the \f(CW\*(C`++\*(C'\fR operator
1533 which expects either a number or a string matching \f(CW\*(C`/^[a\-zA\-Z]*[0\-9]*\ez/\*(C'\fR.
1534 See "Auto-increment and Auto-decrement" in perlop for details.
1535 .IP \(bu 4
1536 Both or neither range ends should be Unicode in regex; marked by <\-\- HERE in m/%s/
1537 .Sp
1538 (W regexp) (only under \f(CW\*(C`use\ re\ \*(Aqstrict\*(Aq\*(C'\fR or within \f(CW\*(C`(?[...])\*(C'\fR)
1539 .Sp
1540 In a bracketed character class in a regular expression pattern, you
1541 had a range which has exactly one end of it specified using \f(CW\*(C`\eN{}\*(C'\fR, and
1542 the other end is specified using a non-portable mechanism. Perl treats
1543 the range as a Unicode range, that is, all the characters in it are
1544 considered to be the Unicode characters, and which may be different code
1545 points on some platforms Perl runs on. For example, \f(CW\*(C`[\eN{U+06}\-\ex08]\*(C'\fR
1546 is treated as if you had instead said \f(CW\*(C`[\eN{U+06}\-\eN{U+08}]\*(C'\fR, that is it
1547 matches the characters whose code points in Unicode are 6, 7, and 8.
1548 But that \f(CW\*(C`\ex08\*(C'\fR might indicate that you meant something different, so
1549 the warning gets raised.
1550 .IP \(bu 4
1551 Can't do %s("%s") on non\-UTF\-8 locale; resolved to "%s".
1552 .Sp
1553 (W locale) You are 1) running under "\f(CW\*(C`use locale\*(C'\fR"; 2) the current
1554 locale is not a UTF\-8 one; 3) you tried to do the designated case-change
1555 operation on the specified Unicode character; and 4) the result of this
1556 operation would mix Unicode and locale rules, which likely conflict.
1557 .Sp
1558 The warnings category \f(CW\*(C`locale\*(C'\fR is new.
1559 .IP \(bu 4
1560 :const is experimental
1561 .Sp
1562 (S experimental::const_attr) The \f(CW\*(C`const\*(C'\fR attribute is experimental.
1563 If you want to use the feature, disable the warning with \f(CWno warnings
1564 \&\*(Aqexperimental::const_attr\*(Aq\fR, but know that in doing so you are taking
1565 the risk that your code may break in a future Perl version.
1566 .IP \(bu 4
1567 gmtime(%f) failed
1568 .Sp
1569 (W overflow) You called \f(CW\*(C`gmtime\*(C'\fR with a number that it could not handle:
1570 too large, too small, or NaN. The returned value is \f(CW\*(C`undef\*(C'\fR.
1571 .IP \(bu 4
1572 Hexadecimal float: exponent overflow
1573 .Sp
1574 (W overflow) The hexadecimal floating point has larger exponent
1575 than the floating point supports.
1576 .IP \(bu 4
1577 Hexadecimal float: exponent underflow
1578 .Sp
1579 (W overflow) The hexadecimal floating point has smaller exponent
1580 than the floating point supports.
1581 .IP \(bu 4
1582 Hexadecimal float: mantissa overflow
1583 .Sp
1584 (W overflow) The hexadecimal floating point literal had more bits in
1585 the mantissa (the part between the \f(CW\*(C`0x\*(C'\fR and the exponent, also known as
1586 the fraction or the significand) than the floating point supports.
1587 .IP \(bu 4
1588 Hexadecimal float: precision loss
1589 .Sp
1590 (W overflow) The hexadecimal floating point had internally more
1591 digits than could be output. This can be caused by unsupported
1592 long double formats, or by 64\-bit integers not being available
1593 (needed to retrieve the digits under some configurations).
1594 .IP \(bu 4
1595 Locale '%s' may not work well.%s
1596 .Sp
1597 (W locale) You are using the named locale, which is a non\-UTF\-8 one, and
1598 which perl has determined is not fully compatible with what it can
1599 handle. The second \f(CW%s\fR gives a reason.
1600 .Sp
1601 The warnings category \f(CW\*(C`locale\*(C'\fR is new.
1602 .IP \(bu 4
1603 localtime(%f) failed
1604 .Sp
1605 (W overflow) You called \f(CW\*(C`localtime\*(C'\fR with a number that it could not handle:
1606 too large, too small, or NaN. The returned value is \f(CW\*(C`undef\*(C'\fR.
1607 .IP \(bu 4
1608 Negative repeat count does nothing
1609 .Sp
1610 (W numeric) You tried to execute the
1611 \&\f(CW\*(C`x\*(C'\fR repetition operator fewer than 0
1612 times, which doesn't make sense.
1613 .IP \(bu 4
1614 NO-BREAK SPACE in a charnames alias definition is deprecated
1615 .Sp
1616 (D deprecated) You defined a character name which contained a no-break
1617 space character. Change it to a regular space. Usually these names are
1618 defined in the \f(CW\*(C`:alias\*(C'\fR import argument to \f(CW\*(C`use charnames\*(C'\fR, but they
1619 could be defined by a translator installed into \f(CW$^H{charnames}\fR. See
1620 "CUSTOM ALIASES" in charnames.
1621 .IP \(bu 4
1622 Non-finite repeat count does nothing
1623 .Sp
1624 (W numeric) You tried to execute the
1625 \&\f(CW\*(C`x\*(C'\fR repetition operator \f(CW\*(C`Inf\*(C'\fR (or
1626 \&\f(CW\*(C`\-Inf\*(C'\fR) or NaN times, which doesn't make sense.
1627 .IP \(bu 4
1628 PerlIO layer ':win32' is experimental
1629 .Sp
1630 (S experimental::win32_perlio) The \f(CW\*(C`:win32\*(C'\fR PerlIO layer is
1631 experimental. If you want to take the risk of using this layer,
1632 simply disable this warning:
1633 .Sp
1634 .Vb 1
1635 \& no warnings "experimental::win32_perlio";
1636 .Ve
1637 .IP \(bu 4
1638 Ranges of ASCII printables should be some subset of "0\-9", "A\-Z", or "a\-z" in regex; marked by <\-\- HERE in m/%s/
1639 .Sp
1640 (W regexp) (only under \f(CW\*(C`use\ re\ \*(Aqstrict\*(Aq\*(C'\fR or within \f(CW\*(C`(?[...])\*(C'\fR)
1641 .Sp
1642 Stricter rules help to find typos and other errors. Perhaps you didn't
1643 even intend a range here, if the \f(CW"\-"\fR was meant to be some other
1644 character, or should have been escaped (like \f(CW"\e\-"\fR). If you did
1645 intend a range, the one that was used is not portable between ASCII and
1646 EBCDIC platforms, and doesn't have an obvious meaning to a casual
1647 reader.
1648 .Sp
1649 .Vb 7
1650 \& [3\-7] # OK; Obvious and portable
1651 \& [d\-g] # OK; Obvious and portable
1652 \& [A\-Y] # OK; Obvious and portable
1653 \& [A\-z] # WRONG; Not portable; not clear what is meant
1654 \& [a\-Z] # WRONG; Not portable; not clear what is meant
1655 \& [%\-.] # WRONG; Not portable; not clear what is meant
1656 \& [\ex41\-Z] # WRONG; Not portable; not obvious to non\-geek
1657 .Ve
1658 .Sp
1659 (You can force portability by specifying a Unicode range, which means that
1660 the endpoints are specified by
1661 \&\f(CW\*(C`\eN{...}\*(C'\fR, but the meaning may
1662 still not be obvious.)
1663 The stricter rules require that ranges that start or stop with an ASCII
1664 character that is not a control have all their endpoints be a literal
1665 character, and not some escape sequence (like \f(CW"\ex41"\fR), and the ranges
1666 must be all digits, or all uppercase letters, or all lowercase letters.
1667 .IP \(bu 4
1668 Ranges of digits should be from the same group in regex; marked by <\-\- HERE in m/%s/
1669 .Sp
1670 (W regexp) (only under \f(CW\*(C`use\ re\ \*(Aqstrict\*(Aq\*(C'\fR or within \f(CW\*(C`(?[...])\*(C'\fR)
1671 .Sp
1672 Stricter rules help to find typos and other errors. You included a
1673 range, and at least one of the end points is a decimal digit. Under the
1674 stricter rules, when this happens, both end points should be digits in
1675 the same group of 10 consecutive digits.
1676 .IP \(bu 4
1677 Redundant argument in \f(CW%s\fR
1678 .Sp
1679 (W redundant) You called a function with more arguments than were
1680 needed, as indicated by information within other arguments you supplied
1681 (\fIe.g\fR. a printf format). Currently only emitted when a printf-type format
1682 required fewer arguments than were supplied, but might be used in the
1683 future for \fIe.g.\fR "pack" in perlfunc.
1684 .Sp
1685 The warnings category \f(CW\*(C`redundant\*(C'\fR is new. See also
1686 [GH #13534] <https://github.com/Perl/perl5/issues/13534>.
1687 .IP \(bu 4
1688 Replacement list is longer than search list
1689 .Sp
1690 This is not a new diagnostic, but in earlier releases was accidentally
1691 not displayed if the transliteration contained wide characters. This is
1692 now fixed, so that you may see this diagnostic in places where you
1693 previously didn't (but should have).
1694 .IP \(bu 4
1695 Use of \eb{} for non\-UTF\-8 locale is wrong. Assuming a UTF\-8 locale
1696 .Sp
1697 (W locale) You are matching a regular expression using locale rules,
1698 and a Unicode boundary is being matched, but the locale is not a Unicode
1699 one. This doesn't make sense. Perl will continue, assuming a Unicode
1700 (UTF\-8) locale, but the results could well be wrong except if the locale
1701 happens to be ISO\-8859\-1 (Latin1) where this message is spurious and can
1702 be ignored.
1703 .Sp
1704 The warnings category \f(CW\*(C`locale\*(C'\fR is new.
1705 .IP \(bu 4
1706 Using /u for '%s' instead of /%s in regex; marked by <\-\- HERE in m/%s/
1707 .Sp
1708 (W regexp) You used a Unicode boundary (\f(CW\*(C`\eb{...}\*(C'\fR or \f(CW\*(C`\eB{...}\*(C'\fR) in a
1709 portion of a regular expression where the character set modifiers \f(CW\*(C`/a\*(C'\fR
1710 or \f(CW\*(C`/aa\*(C'\fR are in effect. These two modifiers indicate an ASCII
1711 interpretation, and this doesn't make sense for a Unicode definition.
1712 The generated regular expression will compile so that the boundary uses
1713 all of Unicode. No other portion of the regular expression is affected.
1714 .IP \(bu 4
1715 The bitwise feature is experimental
1716 .Sp
1717 (S experimental::bitwise) This warning is emitted if you use bitwise
1718 operators (\f(CW\*(C`& | ^ ~ &. |. ^. ~.\*(C'\fR) with the "bitwise" feature enabled.
1719 Simply suppress the warning if you want to use the feature, but know
1720 that in doing so you are taking the risk of using an experimental
1721 feature which may change or be removed in a future Perl version:
1722 .Sp
1723 .Vb 3
1724 \& no warnings "experimental::bitwise";
1725 \& use feature "bitwise";
1726 \& $x |.= $y;
1727 .Ve
1728 .IP \(bu 4
1729 Unescaped left brace in regex is deprecated, passed through in regex; marked by <\-\- HERE in m/%s/
1730 .Sp
1731 (D deprecated, regexp) You used a literal \f(CW"{"\fR character in a regular
1732 expression pattern. You should change to use \f(CW"\e{"\fR instead, because a future
1733 version of Perl (tentatively v5.26) will consider this to be a syntax error. If
1734 the pattern delimiters are also braces, any matching right brace
1735 (\f(CW"}"\fR) should also be escaped to avoid confusing the parser, for
1736 example,
1737 .Sp
1738 .Vb 1
1739 \& qr{abc\e{def\e}ghi}
1740 .Ve
1741 .IP \(bu 4
1742 Use of literal non-graphic characters in variable names is deprecated
1743 .Sp
1744 (D deprecated) Using literal non-graphic (including control)
1745 characters in the source to refer to the \fI^FOO\fR variables, like \f(CW$^X\fR and
1746 \&\f(CW\*(C`${^GLOBAL_PHASE}\*(C'\fR is now deprecated.
1747 .IP \(bu 4
1748 Useless use of attribute "const"
1749 .Sp
1750 (W misc) The \f(CW\*(C`const\*(C'\fR attribute has no effect except
1751 on anonymous closure prototypes. You applied it to
1752 a subroutine via attributes.pm. This is only useful
1753 inside an attribute handler for an anonymous subroutine.
1754 .IP \(bu 4
1755 Useless use of /d modifier in transliteration operator
1756 .Sp
1757 This is not a new diagnostic, but in earlier releases was accidentally
1758 not displayed if the transliteration contained wide characters. This is
1759 now fixed, so that you may see this diagnostic in places where you
1760 previously didn't (but should have).
1761 .IP \(bu 4
1762 "use re 'strict'" is experimental
1763 .Sp
1764 (S experimental::re_strict) The things that are different when a regular
1765 expression pattern is compiled under \f(CW\*(Aqstrict\*(Aq\fR are subject to change
1766 in future Perl releases in incompatible ways; there are also proposals
1767 to change how to enable strict checking instead of using this subpragma.
1768 This means that a pattern that compiles today may not in a future Perl
1769 release. This warning is to alert you to that risk.
1770 .IP \(bu 4
1771 Warning: unable to close filehandle properly: \f(CW%s\fR
1772 .Sp
1773 Warning: unable to close filehandle \f(CW%s\fR properly: \f(CW%s\fR
1774 .Sp
1775 (S io) Previously, perl silently ignored any errors when doing an implicit
1776 close of a filehandle, \fIi.e.\fR where the reference count of the filehandle
1777 reached zero and the user's code hadn't already called \f(CWclose()\fR; \fIe.g.\fR
1778 .Sp
1779 .Vb 4
1780 \& {
1781 \& open my $fh, \*(Aq>\*(Aq, $file or die "open: \*(Aq$file\*(Aq: $!\en";
1782 \& print $fh, $data or die;
1783 \& } # implicit close here
1784 .Ve
1785 .Sp
1786 In a situation such as disk full, due to buffering, the error may only be
1787 detected during the final close, so not checking the result of the close is
1788 dangerous.
1789 .Sp
1790 So perl now warns in such situations.
1791 .IP \(bu 4
1792 Wide character (U+%X) in \f(CW%s\fR
1793 .Sp
1794 (W locale) While in a single-byte locale (\fIi.e.\fR, a non\-UTF\-8
1795 one), a multi-byte character was encountered. Perl considers this
1796 character to be the specified Unicode code point. Combining non\-UTF\-8
1797 locales and Unicode is dangerous. Almost certainly some characters
1798 will have two different representations. For example, in the ISO 8859\-7
1799 (Greek) locale, the code point 0xC3 represents a Capital Gamma. But so
1800 also does 0x393. This will make string comparisons unreliable.
1801 .Sp
1802 You likely need to figure out how this multi-byte character got mixed up
1803 with your single-byte locale (or perhaps you thought you had a UTF\-8
1804 locale, but Perl disagrees).
1805 .Sp
1806 The warnings category \f(CW\*(C`locale\*(C'\fR is new.
1807 .SS "Changes to Existing Diagnostics"
1808 .IX Subsection "Changes to Existing Diagnostics"
1809 .IP \(bu 4
1810 <> should be quotes
1811 .Sp
1812 This warning has been changed to
1813 <> at require-statement should be quotes
1814 to make the issue more identifiable.
1815 .IP \(bu 4
1816 Argument "%s" isn't numeric%s
1817 .Sp
1818 The perldiag entry for this warning has added this clarifying note:
1819 .Sp
1820 .Vb 4
1821 \& Note that for the Inf and NaN (infinity and not\-a\-number) the
1822 \& definition of "numeric" is somewhat unusual: the strings themselves
1823 \& (like "Inf") are considered numeric, and anything following them is
1824 \& considered non\-numeric.
1825 .Ve
1826 .IP \(bu 4
1827 Global symbol "%s" requires explicit package name
1828 .Sp
1829 This message has had '(did you forget to declare "my \f(CW%s\fR"?)' appended to it, to
1830 make it more helpful to new Perl programmers.
1831 [GH #13732] <https://github.com/Perl/perl5/issues/13732>
1832 .IP \(bu 4
1833 \&'"my" variable &foo::bar can't be in a package' has been reworded to say
1834 \&'subroutine' instead of 'variable'.
1835 .IP \(bu 4
1836 \&\eN{} in character class restricted to one character in regex; marked by
1837 <\-\-\ HERE in m/%s/
1838 .Sp
1839 This message has had \fIcharacter class\fR changed to \fIinverted character
1840 class or as a range end-point is\fR to reflect improvements in
1841 \&\f(CW\*(C`qr/[\eN{named sequence}]/\*(C'\fR (see under "Selected Bug Fixes").
1842 .IP \(bu 4
1843 panic: frexp
1844 .Sp
1845 This message has had ': \f(CW%f\fR' appended to it, to show what the offending
1846 floating point number is.
1847 .IP \(bu 4
1848 \&\fIPossible precedence problem on bitwise \fR\f(CI%c\fR\fI operator\fR reworded as
1849 Possible precedence problem on bitwise \f(CW%s\fR operator.
1850 .IP \(bu 4
1851 Unsuccessful \f(CW%s\fR on filename containing newline
1852 .Sp
1853 This warning is now only produced when the newline is at the end of
1854 the filename.
1855 .IP \(bu 4
1856 "Variable \f(CW%s\fR will not stay shared" has been changed to say "Subroutine"
1857 when it is actually a lexical sub that will not stay shared.
1858 .IP \(bu 4
1859 Variable length lookbehind not implemented in regex m/%s/
1860 .Sp
1861 The perldiag entry for this warning has had information about Unicode
1862 behavior added.
1863 .SS "Diagnostic Removals"
1864 .IX Subsection "Diagnostic Removals"
1865 .IP \(bu 4
1866 "Ambiguous use of \-foo resolved as \-&\fBfoo()\fR"
1867 .Sp
1868 There is actually no ambiguity here, and this impedes the use of negated
1869 constants; \fIe.g.\fR, \f(CW\*(C`\-Inf\*(C'\fR.
1870 .IP \(bu 4
1871 "Constant is not a FOO reference"
1872 .Sp
1873 Compile-time checking of constant dereferencing (\fIe.g.\fR, \f(CWmy_constant\->()\fR)
1874 has been removed, since it was not taking overloading into account.
1875 [GH #9891] <https://github.com/Perl/perl5/issues/9891>
1876 [GH #14044] <https://github.com/Perl/perl5/issues/14044>
1877 .SH "Utility Changes"
1878 .IX Header "Utility Changes"
1879 .SS "\fIfind2perl\fP, \fIs2p\fP and \fIa2p\fP removal"
1880 .IX Subsection "find2perl, s2p and a2p removal"
1881 .IP \(bu 4
1882 The \fIx2p/\fR directory has been removed from the Perl core.
1883 .Sp
1884 This removes find2perl, s2p and a2p. They have all been released to CPAN as
1885 separate distributions (\f(CW\*(C`App::find2perl\*(C'\fR, \f(CW\*(C`App::s2p\*(C'\fR, \f(CW\*(C`App::a2p\*(C'\fR).
1886 .SS h2ph
1887 .IX Subsection "h2ph"
1888 .IP \(bu 4
1889 \&\fIh2ph\fR now handles hexadecimal constants in the compiler's predefined
1890 macro definitions, as visible in \f(CW$Config{cppsymbols}\fR.
1891 [GH #14491] <https://github.com/Perl/perl5/issues/14491>.
1892 .SS encguess
1893 .IX Subsection "encguess"
1894 .IP \(bu 4
1895 No longer depends on non-core modules.
1896 .SH "Configuration and Compilation"
1897 .IX Header "Configuration and Compilation"
1898 .IP \(bu 4
1899 \&\fIConfigure\fR now checks for \f(CWlrintl()\fR, \f(CWlroundl()\fR, \f(CWllrintl()\fR, and
1900 \&\f(CWllroundl()\fR.
1901 .IP \(bu 4
1902 \&\fIConfigure\fR with \f(CW\*(C`\-Dmksymlinks\*(C'\fR should now be faster.
1903 [GH #13890] <https://github.com/Perl/perl5/issues/13890>.
1904 .IP \(bu 4
1905 The \f(CW\*(C`pthreads\*(C'\fR and \f(CW\*(C`cl\*(C'\fR libraries will be linked by default if present.
1906 This allows XS modules that require threading to work on non-threaded
1907 perls. Note that you must still pass \f(CW\*(C`\-Dusethreads\*(C'\fR if you want a
1908 threaded perl.
1909 .IP \(bu 4
1910 To get more precision and range for floating point numbers one can now
1911 use the GCC quadmath library which implements the quadruple precision
1912 floating point numbers on x86 and IA\-64 platforms. See \fIINSTALL\fR for
1913 details.
1914 .IP \(bu 4
1915 MurmurHash64A and MurmurHash64B can now be configured as the internal hash
1916 function.
1917 .IP \(bu 4
1918 \&\f(CW\*(C`make test.valgrind\*(C'\fR now supports parallel testing.
1919 .Sp
1920 For example:
1921 .Sp
1922 .Vb 1
1923 \& TEST_JOBS=9 make test.valgrind
1924 .Ve
1925 .Sp
1926 See "valgrind" in perlhacktips for more information.
1927 .Sp
1928 [GH #13658] <https://github.com/Perl/perl5/issues/13658>
1929 .IP \(bu 4
1930 The MAD (Misc Attribute Decoration) build option has been removed
1931 .Sp
1932 This was an unmaintained attempt at preserving
1933 the Perl parse tree more faithfully so that automatic conversion of
1934 Perl 5 to Perl 6 would have been easier.
1935 .Sp
1936 This build-time configuration option had been unmaintained for years,
1937 and had probably seriously diverged on both Perl 5 and Perl 6 sides.
1938 .IP \(bu 4
1939 A new compilation flag, \f(CW\*(C`\-DPERL_OP_PARENT\*(C'\fR is available. For details,
1940 see the discussion below at "Internal Changes".
1941 .IP \(bu 4
1942 Pathtools no longer tries to load XS on miniperl. This speeds up building perl
1943 slightly.
1944 .SH Testing
1945 .IX Header "Testing"
1946 .IP \(bu 4
1947 \&\fIt/porting/re_context.t\fR has been added to test that utf8 and its
1948 dependencies only use the subset of the \f(CW\*(C`$1..$n\*(C'\fR capture vars that
1949 \&\f(CWPerl_save_re_context()\fR is hard-coded to localize, because that function
1950 has no efficient way of determining at runtime what vars to localize.
1951 .IP \(bu 4
1952 Tests for performance issues have been added in the file \fIt/perf/taint.t\fR.
1953 .IP \(bu 4
1954 Some regular expression tests are written in such a way that they will
1955 run very slowly if certain optimizations break. These tests have been
1956 moved into new files, \fIt/re/speed.t\fR and \fIt/re/speed_thr.t\fR,
1957 and are run with a \f(CWwatchdog()\fR.
1958 .IP \(bu 4
1959 \&\f(CW\*(C`test.pl\*(C'\fR now allows \f(CW\*(C`plan skip_all => $reason\*(C'\fR, to make it
1960 more compatible with \f(CW\*(C`Test::More\*(C'\fR.
1961 .IP \(bu 4
1962 A new test script, \fIop/infnan.t\fR, has been added to test if infinity and NaN are
1963 working correctly. See "Infinity and NaN (not-a-number) handling improved".
1964 .SH "Platform Support"
1965 .IX Header "Platform Support"
1966 .SS "Regained Platforms"
1967 .IX Subsection "Regained Platforms"
1968 .IP "IRIX and Tru64 platforms are working again." 4
1969 .IX Item "IRIX and Tru64 platforms are working again."
1970 Some \f(CW\*(C`make test\*(C'\fR failures remain:
1971 [GH #14557] <https://github.com/Perl/perl5/issues/14557>
1972 and [GH #14727] <https://github.com/Perl/perl5/issues/14727>
1973 for IRIX; [GH #14629] <https://github.com/Perl/perl5/issues/14629>,
1974 [cpan #99605] <https://rt.cpan.org/Public/Bug/Display.html?id=99605>, and
1975 [cpan #104836] <https://rt.cpan.org/Ticket/Display.html?id=104836> for Tru64.
1976 .IP "z/OS running EBCDIC Code Page 1047" 4
1977 .IX Item "z/OS running EBCDIC Code Page 1047"
1978 Core perl now works on this EBCDIC platform. Earlier perls also worked, but,
1979 even though support wasn't officially withdrawn, recent perls would not compile
1980 and run well. Perl 5.20 would work, but had many bugs which have now been
1981 fixed. Many CPAN modules that ship with Perl still fail tests, including
1982 \&\f(CW\*(C`Pod::Simple\*(C'\fR. However the version of \f(CW\*(C`Pod::Simple\*(C'\fR currently on CPAN should work;
1983 it was fixed too late to include in Perl 5.22. Work is under way to fix many
1984 of the still-broken CPAN modules, which likely will be installed on CPAN when
1985 completed, so that you may not have to wait until Perl 5.24 to get a working
1986 version.
1987 .SS "Discontinued Platforms"
1988 .IX Subsection "Discontinued Platforms"
1989 .IP NeXTSTEP/OPENSTEP 4
1990 .IX Item "NeXTSTEP/OPENSTEP"
1991 NeXTSTEP was a proprietary operating system bundled with NeXT's
1992 workstations in the early to mid 90s; OPENSTEP was an API specification
1993 that provided a NeXTSTEP-like environment on a non-NeXTSTEP system. Both
1994 are now long dead, so support for building Perl on them has been removed.
1995 .SS "Platform-Specific Notes"
1996 .IX Subsection "Platform-Specific Notes"
1997 .IP EBCDIC 4
1998 .IX Item "EBCDIC"
1999 Special handling is required of the perl interpreter on EBCDIC platforms
2000 to get \f(CW\*(C`qr/[i\-j]/\*(C'\fR to match only \f(CW"i"\fR and \f(CW"j"\fR, since there are 7
2001 characters between the
2002 code points for \f(CW"i"\fR and \f(CW"j"\fR. This special handling had only been
2003 invoked when both ends of the range are literals. Now it is also
2004 invoked if any of the \f(CW\*(C`\eN{...}\*(C'\fR forms for specifying a character by
2005 name or Unicode code point is used instead of a literal. See
2006 "Character Ranges" in perlrecharclass.
2007 .IP HP-UX 4
2008 .IX Item "HP-UX"
2009 The archname now distinguishes use64bitint from use64bitall.
2010 .IP Android 4
2011 .IX Item "Android"
2012 Build support has been improved for cross-compiling in general and for
2013 Android in particular.
2014 .IP VMS 4
2015 .IX Item "VMS"
2016 .RS 4
2017 .PD 0
2018 .IP \(bu 4
2019 .PD
2020 When spawning a subprocess without waiting, the return value is now
2021 the correct PID.
2022 .IP \(bu 4
2023 Fix a prototype so linking doesn't fail under the VMS C++ compiler.
2024 .IP \(bu 4
2025 \&\f(CW\*(C`finite\*(C'\fR, \f(CW\*(C`finitel\*(C'\fR, and \f(CW\*(C`isfinite\*(C'\fR detection has been added to
2026 \&\f(CW\*(C`configure.com\*(C'\fR, environment handling has had some minor changes, and
2027 a fix for legacy feature checking status.
2028 .RE
2029 .RS 4
2030 .RE
2031 .IP Win32 4
2032 .IX Item "Win32"
2033 .RS 4
2034 .PD 0
2035 .IP \(bu 4
2036 .PD
2037 \&\fIminiperl.exe\fR is now built with \f(CW\*(C`\-fno\-strict\-aliasing\*(C'\fR, allowing 64\-bit
2038 builds to complete on GCC 4.8.
2039 [GH #14556] <https://github.com/Perl/perl5/issues/14556>
2040 .IP \(bu 4
2041 \&\f(CW\*(C`nmake minitest\*(C'\fR now works on Win32. Due to dependency issues you
2042 need to build \f(CW\*(C`nmake test\-prep\*(C'\fR first, and a small number of the
2043 tests fail.
2044 [GH #14318] <https://github.com/Perl/perl5/issues/14318>
2045 .IP \(bu 4
2046 Perl can now be built in C++ mode on Windows by setting the makefile macro
2047 \&\f(CW\*(C`USE_CPLUSPLUS\*(C'\fR to the value "define".
2048 .IP \(bu 4
2049 The list form of piped open has been implemented for Win32. Note: unlike
2050 \&\f(CW\*(C`system LIST\*(C'\fR this does not fall back to the shell.
2051 [GH #13574] <https://github.com/Perl/perl5/issues/13574>
2052 .IP \(bu 4
2053 New \f(CW\*(C`DebugSymbols\*(C'\fR and \f(CW\*(C`DebugFull\*(C'\fR configuration options added to
2054 Windows makefiles.
2055 .IP \(bu 4
2056 Previously, compiling XS modules (including CPAN ones) using Visual C++ for
2057 Win64 resulted in around a dozen warnings per file from \fIhv_func.h\fR. These
2058 warnings have been silenced.
2059 .IP \(bu 4
2060 Support for building without PerlIO has been removed from the Windows
2061 makefiles. Non-PerlIO builds were all but deprecated in Perl 5.18.0 and are
2062 already not supported by \fIConfigure\fR on POSIX systems.
2063 .IP \(bu 4
2064 Between 2 and 6 milliseconds and seven I/O calls have been saved per attempt
2065 to open a perl module for each path in \f(CW@INC\fR.
2066 .IP \(bu 4
2067 Intel C builds are now always built with C99 mode on.
2068 .IP \(bu 4
2069 \&\f(CW%I64d\fR is now being used instead of \f(CW%lld\fR for MinGW.
2070 .IP \(bu 4
2071 In the experimental \f(CW\*(C`:win32\*(C'\fR layer, a crash in \f(CW\*(C`open\*(C'\fR was fixed. Also
2072 opening \fI/dev/null\fR (which works under Win32 Perl's default \f(CW\*(C`:unix\*(C'\fR
2073 layer) was implemented for \f(CW\*(C`:win32\*(C'\fR.
2074 [GH #13968] <https://github.com/Perl/perl5/issues/13968>
2075 .IP \(bu 4
2076 A new makefile option, \f(CW\*(C`USE_LONG_DOUBLE\*(C'\fR, has been added to the Windows
2077 dmake makefile for gcc builds only. Set this to "define" if you want perl to
2078 use long doubles to give more accuracy and range for floating point numbers.
2079 .RE
2080 .RS 4
2081 .RE
2082 .IP OpenBSD 4
2083 .IX Item "OpenBSD"
2084 On OpenBSD, Perl will now default to using the system \f(CW\*(C`malloc\*(C'\fR due to the
2085 security features it provides. Perl's own malloc wrapper has been in use
2086 since v5.14 due to performance reasons, but the OpenBSD project believes
2087 the tradeoff is worth it and would prefer that users who need the speed
2088 specifically ask for it.
2089 .Sp
2090 [GH #13888] <https://github.com/Perl/perl5/issues/13888>.
2091 .IP Solaris 4
2092 .IX Item "Solaris"
2093 .RS 4
2094 .PD 0
2095 .IP \(bu 4
2096 .PD
2097 We now look for the Sun Studio compiler in both \fI/opt/solstudio*\fR and
2098 \&\fI/opt/solarisstudio*\fR.
2099 .IP \(bu 4
2100 Builds on Solaris 10 with \f(CW\*(C`\-Dusedtrace\*(C'\fR would fail early since make
2101 didn't follow implied dependencies to build \f(CW\*(C`perldtrace.h\*(C'\fR. Added an
2102 explicit dependency to \f(CW\*(C`depend\*(C'\fR.
2103 [GH #13334] <https://github.com/Perl/perl5/issues/13334>
2104 .IP \(bu 4
2105 C99 options have been cleaned up; hints look for \f(CW\*(C`solstudio\*(C'\fR
2106 as well as \f(CW\*(C`SUNWspro\*(C'\fR; and support for native \f(CW\*(C`setenv\*(C'\fR has been added.
2107 .RE
2108 .RS 4
2109 .RE
2110 .SH "Internal Changes"
2111 .IX Header "Internal Changes"
2112 .IP \(bu 4
2113 Experimental support has been added to allow ops in the optree to locate
2114 their parent, if any. This is enabled by the non-default build option
2115 \&\f(CW\*(C`\-DPERL_OP_PARENT\*(C'\fR. It is envisaged that this will eventually become
2116 enabled by default, so XS code which directly accesses the \f(CW\*(C`op_sibling\*(C'\fR
2117 field of ops should be updated to be future-proofed.
2118 .Sp
2119 On \f(CW\*(C`PERL_OP_PARENT\*(C'\fR builds, the \f(CW\*(C`op_sibling\*(C'\fR field has been renamed
2120 \&\f(CW\*(C`op_sibparent\*(C'\fR and a new flag, \f(CW\*(C`op_moresib\*(C'\fR, added. On the last op in a
2121 sibling chain, \f(CW\*(C`op_moresib\*(C'\fR is false and \f(CW\*(C`op_sibparent\*(C'\fR points to the
2122 parent (if any) rather than being \f(CW\*(C`NULL\*(C'\fR.
2123 .Sp
2124 To make existing code work transparently whether using \f(CW\*(C`PERL_OP_PARENT\*(C'\fR
2125 or not, a number of new macros and functions have been added that should
2126 be used, rather than directly manipulating \f(CW\*(C`op_sibling\*(C'\fR.
2127 .Sp
2128 For the case of just reading \f(CW\*(C`op_sibling\*(C'\fR to determine the next sibling,
2129 two new macros have been added. A simple scan through a sibling chain
2130 like this:
2131 .Sp
2132 .Vb 1
2133 \& for (; kid\->op_sibling; kid = kid\->op_sibling) { ... }
2134 .Ve
2135 .Sp
2136 should now be written as:
2137 .Sp
2138 .Vb 1
2139 \& for (; OpHAS_SIBLING(kid); kid = OpSIBLING(kid)) { ... }
2140 .Ve
2141 .Sp
2142 For altering optrees, a general-purpose function \f(CWop_sibling_splice()\fR
2143 has been added, which allows for manipulation of a chain of sibling ops.
2144 By analogy with the Perl function \f(CWsplice()\fR, it allows you to cut out
2145 zero or more ops from a sibling chain and replace them with zero or more
2146 new ops. It transparently handles all the updating of sibling, parent,
2147 op_last pointers etc.
2148 .Sp
2149 If you need to manipulate ops at a lower level, then three new macros,
2150 \&\f(CW\*(C`OpMORESIB_set\*(C'\fR, \f(CW\*(C`OpLASTSIB_set\*(C'\fR and \f(CW\*(C`OpMAYBESIB_set\*(C'\fR are intended to
2151 be a low-level portable way to set \f(CW\*(C`op_sibling\*(C'\fR / \f(CW\*(C`op_sibparent\*(C'\fR while
2152 also updating \f(CW\*(C`op_moresib\*(C'\fR. The first sets the sibling pointer to a new
2153 sibling, the second makes the op the last sibling, and the third
2154 conditionally does the first or second action. Note that unlike
2155 \&\f(CWop_sibling_splice()\fR these macros won't maintain consistency in the
2156 parent at the same time (\fIe.g.\fR by updating \f(CW\*(C`op_first\*(C'\fR and \f(CW\*(C`op_last\*(C'\fR where
2157 appropriate).
2158 .Sp
2159 A C\-level \f(CWPerl_op_parent()\fR function and a Perl-level \f(CWB::OP::parent()\fR
2160 method have been added. The C function only exists under
2161 \&\f(CW\*(C`PERL_OP_PARENT\*(C'\fR builds (using it is build-time error on vanilla
2162 perls). \f(CWB::OP::parent()\fR exists always, but on a vanilla build it
2163 always returns \f(CW\*(C`NULL\*(C'\fR. Under \f(CW\*(C`PERL_OP_PARENT\*(C'\fR, they return the parent
2164 of the current op, if any. The variable \f(CW$B::OP::does_parent\fR allows you
2165 to determine whether \f(CW\*(C`B\*(C'\fR supports retrieving an op's parent.
2166 .Sp
2167 \&\f(CW\*(C`PERL_OP_PARENT\*(C'\fR was introduced in 5.21.2, but the interface was
2168 changed considerably in 5.21.11. If you updated your code before the
2169 5.21.11 changes, it may require further revision. The main changes after
2170 5.21.2 were:
2171 .RS 4
2172 .IP \(bu 4
2173 The \f(CW\*(C`OP_SIBLING\*(C'\fR and \f(CW\*(C`OP_HAS_SIBLING\*(C'\fR macros have been renamed
2174 \&\f(CW\*(C`OpSIBLING\*(C'\fR and \f(CW\*(C`OpHAS_SIBLING\*(C'\fR for consistency with other
2175 op-manipulating macros.
2176 .IP \(bu 4
2177 The \f(CW\*(C`op_lastsib\*(C'\fR field has been renamed \f(CW\*(C`op_moresib\*(C'\fR, and its meaning
2178 inverted.
2179 .IP \(bu 4
2180 The macro \f(CW\*(C`OpSIBLING_set\*(C'\fR has been removed, and has been superseded by
2181 \&\f(CW\*(C`OpMORESIB_set\*(C'\fR \fIet al\fR.
2182 .IP \(bu 4
2183 The \f(CWop_sibling_splice()\fR function now accepts a null \f(CW\*(C`parent\*(C'\fR argument
2184 where the splicing doesn't affect the first or last ops in the sibling
2185 chain
2186 .RE
2187 .RS 4
2188 .RE
2189 .IP \(bu 4
2190 Macros have been created to allow XS code to better manipulate the POSIX locale
2191 category \f(CW\*(C`LC_NUMERIC\*(C'\fR. See "Locale-related functions and macros" in perlapi.
2192 .IP \(bu 4
2193 The previous \f(CW\*(C`atoi\*(C'\fR \fIet al\fR replacement function, \f(CW\*(C`grok_atou\*(C'\fR, has now been
2194 superseded by \f(CW\*(C`grok_atoUV\*(C'\fR. See perlclib for details.
2195 .IP \(bu 4
2196 A new function, \f(CWPerl_sv_get_backrefs()\fR, has been added which allows you
2197 retrieve the weak references, if any, which point at an SV.
2198 .IP \(bu 4
2199 The \f(CWscreaminstr()\fR function has been removed. Although marked as
2200 public API, it was undocumented and had no usage in CPAN modules. Calling
2201 it has been fatal since 5.17.0.
2202 .IP \(bu 4
2203 The \f(CWnewDEFSVOP()\fR, \f(CWblock_start()\fR, \f(CWblock_end()\fR and \f(CWintro_my()\fR
2204 functions have been added to the API.
2205 .IP \(bu 4
2206 The internal \f(CW\*(C`convert\*(C'\fR function in \fIop.c\fR has been renamed
2207 \&\f(CW\*(C`op_convert_list\*(C'\fR and added to the API.
2208 .IP \(bu 4
2209 The \f(CWsv_magic()\fR function no longer forbids "ext" magic on read-only
2210 values. After all, perl can't know whether the custom magic will modify
2211 the SV or not.
2212 [GH #14202] <https://github.com/Perl/perl5/issues/14202>.
2213 .IP \(bu 4
2214 Accessing "CvPADLIST" in perlapi on an XSUB is now forbidden.
2215 .Sp
2216 The \f(CW\*(C`CvPADLIST\*(C'\fR field has been reused for a different internal purpose
2217 for XSUBs. So in particular, you can no longer rely on it being NULL as a
2218 test of whether a CV is an XSUB. Use \f(CWCvISXSUB()\fR instead.
2219 .IP \(bu 4
2220 SVs of type \f(CW\*(C`SVt_NV\*(C'\fR are now sometimes bodiless when the build
2221 configuration and platform allow it: specifically, when \f(CWsizeof(NV) <=
2222 sizeof(IV)\fR. "Bodiless" means that the NV value is stored directly in
2223 the head of an SV, without requiring a separate body to be allocated. This
2224 trick has already been used for IVs since 5.9.2 (though in the case of
2225 IVs, it is always used, regardless of platform and build configuration).
2226 .IP \(bu 4
2227 The \f(CW$DB::single\fR, \f(CW$DB::signal\fR and \f(CW$DB::trace\fR variables now have set\- and
2228 get-magic that stores their values as IVs, and those IVs are used when
2229 testing their values in \f(CWpp_dbstate()\fR. This prevents perl from
2230 recursing infinitely if an overloaded object is assigned to any of those
2231 variables.
2232 [GH #14013] <https://github.com/Perl/perl5/issues/14013>.
2233 .IP \(bu 4
2234 \&\f(CWPerl_tmps_grow()\fR, which is marked as public API but is undocumented, has
2235 been removed from the public API. This change does not affect XS code that
2236 uses the \f(CW\*(C`EXTEND_MORTAL\*(C'\fR macro to pre-extend the mortal stack.
2237 .IP \(bu 4
2238 Perl's internals no longer sets or uses the \f(CW\*(C`SVs_PADMY\*(C'\fR flag.
2239 \&\f(CWSvPADMY()\fR now returns a true value for anything not marked \f(CW\*(C`PADTMP\*(C'\fR
2240 and \f(CW\*(C`SVs_PADMY\*(C'\fR is now defined as 0.
2241 .IP \(bu 4
2242 The macros \f(CW\*(C`SETsv\*(C'\fR and \f(CW\*(C`SETsvUN\*(C'\fR have been removed. They were no longer used
2243 in the core since commit 6f1401dc2a five years ago, and have not been
2244 found present on CPAN.
2245 .IP \(bu 4
2246 The \f(CW\*(C`SvFAKE\*(C'\fR bit (unused on HVs) got informally reserved by
2247 David Mitchell for future work on vtables.
2248 .IP \(bu 4
2249 The \f(CWsv_catpvn_flags()\fR function accepts \f(CW\*(C`SV_CATBYTES\*(C'\fR and \f(CW\*(C`SV_CATUTF8\*(C'\fR
2250 flags, which specify whether the appended string is bytes or UTF\-8,
2251 respectively. (These flags have in fact been present since 5.16.0, but
2252 were formerly not regarded as part of the API.)
2253 .IP \(bu 4
2254 A new opcode class, \f(CW\*(C`METHOP\*(C'\fR, has been introduced. It holds
2255 information used at runtime to improve the performance
2256 of class/object method calls.
2257 .Sp
2258 \&\f(CW\*(C`OP_METHOD\*(C'\fR and \f(CW\*(C`OP_METHOD_NAMED\*(C'\fR have changed from being
2259 \&\f(CW\*(C`UNOP/SVOP\*(C'\fR to being \f(CW\*(C`METHOP\*(C'\fR.
2260 .IP \(bu 4
2261 \&\f(CWcv_name()\fR is a new API function that can be passed a CV or GV. It
2262 returns an SV containing the name of the subroutine, for use in
2263 diagnostics.
2264 .Sp
2265 [GH #12767] <https://github.com/Perl/perl5/issues/12767>
2266 [GH #13392] <https://github.com/Perl/perl5/issues/13392>
2267 .IP \(bu 4
2268 \&\f(CWcv_set_call_checker_flags()\fR is a new API function that works like
2269 \&\f(CWcv_set_call_checker()\fR, except that it allows the caller to specify
2270 whether the call checker requires a full GV for reporting the subroutine's
2271 name, or whether it could be passed a CV instead. Whatever value is
2272 passed will be acceptable to \f(CWcv_name()\fR. \f(CWcv_set_call_checker()\fR
2273 guarantees there will be a GV, but it may have to create one on the fly,
2274 which is inefficient.
2275 [GH #12767] <https://github.com/Perl/perl5/issues/12767>
2276 .IP \(bu 4
2277 \&\f(CW\*(C`CvGV\*(C'\fR (which is not part of the API) is now a more complex macro, which may
2278 call a function and reify a GV. For those cases where it has been used as a
2279 boolean, \f(CW\*(C`CvHASGV\*(C'\fR has been added, which will return true for CVs that
2280 notionally have GVs, but without reifying the GV. \f(CW\*(C`CvGV\*(C'\fR also returns a GV
2281 now for lexical subs.
2282 [GH #13392] <https://github.com/Perl/perl5/issues/13392>
2283 .IP \(bu 4
2284 The "sync_locale" in perlapi function has been added to the public API.
2285 Changing the program's locale should be avoided by XS code. Nevertheless,
2286 certain non-Perl libraries called from XS need to do so, such as \f(CW\*(C`Gtk\*(C'\fR.
2287 When this happens, Perl needs to be told that the locale has
2288 changed. Use this function to do so, before returning to Perl.
2289 .IP \(bu 4
2290 The defines and labels for the flags in the \f(CW\*(C`op_private\*(C'\fR field of OPs are now
2291 auto-generated from data in \fIregen/op_private\fR. The noticeable effect of this
2292 is that some of the flag output of \f(CW\*(C`Concise\*(C'\fR might differ slightly, and the
2293 flag output of \f(CW\*(C`perl\ \-Dx\*(C'\fR may differ considerably (they both use the same set
2294 of labels now). Also, debugging builds now have a new assertion in
2295 \&\f(CWop_free()\fR to ensure that the op doesn't have any unrecognized flags set in
2296 \&\f(CW\*(C`op_private\*(C'\fR.
2297 .IP \(bu 4
2298 The deprecated variable \f(CW\*(C`PL_sv_objcount\*(C'\fR has been removed.
2299 .IP \(bu 4
2300 Perl now tries to keep the locale category \f(CW\*(C`LC_NUMERIC\*(C'\fR set to "C"
2301 except around operations that need it to be set to the program's
2302 underlying locale. This protects the many XS modules that cannot cope
2303 with the decimal radix character not being a dot. Prior to this
2304 release, Perl initialized this category to "C", but a call to
2305 \&\f(CWPOSIX::setlocale()\fR would change it. Now such a call will change the
2306 underlying locale of the \f(CW\*(C`LC_NUMERIC\*(C'\fR category for the program, but the
2307 locale exposed to XS code will remain "C". There are new macros
2308 to manipulate the LC_NUMERIC locale, including
2309 \&\f(CW\*(C`STORE_LC_NUMERIC_SET_TO_NEEDED\*(C'\fR and
2310 \&\f(CW\*(C`STORE_LC_NUMERIC_FORCE_TO_UNDERLYING\*(C'\fR.
2311 See "Locale-related functions and macros" in perlapi.
2312 .IP \(bu 4
2313 A new macro \f(CW\*(C`isUTF8_CHAR\*(C'\fR has been written which
2314 efficiently determines if the string given by its parameters begins
2315 with a well-formed UTF\-8 encoded character.
2316 .IP \(bu 4
2317 The following private API functions had their context parameter removed:
2318 \&\f(CW\*(C`Perl_cast_ulong\*(C'\fR, \f(CW\*(C`Perl_cast_i32\*(C'\fR, \f(CW\*(C`Perl_cast_iv\*(C'\fR, \f(CW\*(C`Perl_cast_uv\*(C'\fR,
2319 \&\f(CW\*(C`Perl_cv_const_sv\*(C'\fR, \f(CW\*(C`Perl_mg_find\*(C'\fR, \f(CW\*(C`Perl_mg_findext\*(C'\fR, \f(CW\*(C`Perl_mg_magical\*(C'\fR,
2320 \&\f(CW\*(C`Perl_mini_mktime\*(C'\fR, \f(CW\*(C`Perl_my_dirfd\*(C'\fR, \f(CW\*(C`Perl_sv_backoff\*(C'\fR, \f(CW\*(C`Perl_utf8_hop\*(C'\fR.
2321 .Sp
2322 Note that the prefix-less versions of those functions that are part of the
2323 public API, such as \f(CWcast_i32()\fR, remain unaffected.
2324 .IP \(bu 4
2325 The \f(CW\*(C`PADNAME\*(C'\fR and \f(CW\*(C`PADNAMELIST\*(C'\fR types are now separate types, and no
2326 longer simply aliases for SV and AV.
2327 [GH #14250] <https://github.com/Perl/perl5/issues/14250>.
2328 .IP \(bu 4
2329 Pad names are now always UTF\-8. The \f(CW\*(C`PadnameUTF8\*(C'\fR macro always returns
2330 true. Previously, this was effectively the case already, but any support
2331 for two different internal representations of pad names has now been
2332 removed.
2333 .IP \(bu 4
2334 A new op class, \f(CW\*(C`UNOP_AUX\*(C'\fR, has been added. This is a subclass of
2335 \&\f(CW\*(C`UNOP\*(C'\fR with an \f(CW\*(C`op_aux\*(C'\fR field added, which points to an array of unions
2336 of UV, SV* etc. It is intended for where an op needs to store more data
2337 than a simple \f(CW\*(C`op_sv\*(C'\fR or whatever. Currently the only op of this type is
2338 \&\f(CW\*(C`OP_MULTIDEREF\*(C'\fR (see next item).
2339 .IP \(bu 4
2340 A new op has been added, \f(CW\*(C`OP_MULTIDEREF\*(C'\fR, which performs one or more
2341 nested array and hash lookups where the key is a constant or simple
2342 variable. For example the expression \f(CW\*(C`$a[0]{$k}[$i]\*(C'\fR, which previously
2343 involved ten \f(CW\*(C`rv2Xv\*(C'\fR, \f(CW\*(C`Xelem\*(C'\fR, \f(CW\*(C`gvsv\*(C'\fR and \f(CW\*(C`const\*(C'\fR ops is now performed
2344 by a single \f(CW\*(C`multideref\*(C'\fR op. It can also handle \f(CW\*(C`local\*(C'\fR, \f(CW\*(C`exists\*(C'\fR and
2345 \&\f(CW\*(C`delete\*(C'\fR. A non-simple index expression, such as \f(CW\*(C`[$i+1]\*(C'\fR is still done
2346 using \f(CW\*(C`aelem\*(C'\fR/\f(CW\*(C`helem\*(C'\fR, and single-level array lookup with a small constant
2347 index is still done using \f(CW\*(C`aelemfast\*(C'\fR.
2348 .SH "Selected Bug Fixes"
2349 .IX Header "Selected Bug Fixes"
2350 .IP \(bu 4
2351 \&\f(CW\*(C`close\*(C'\fR now sets \f(CW$!\fR
2352 .Sp
2353 When an I/O error occurs, the fact that there has been an error is recorded
2354 in the handle. \f(CW\*(C`close\*(C'\fR returns false for such a handle. Previously, the
2355 value of \f(CW$!\fR would be untouched by \f(CW\*(C`close\*(C'\fR, so the common convention of
2356 writing \f(CW\*(C`close\ $fh\ or\ die\ $!\*(C'\fR did not work reliably. Now the handle
2357 records the value of \f(CW$!\fR, too, and \f(CW\*(C`close\*(C'\fR restores it.
2358 .IP \(bu 4
2359 \&\f(CW\*(C`no re\*(C'\fR now can turn off everything that \f(CW\*(C`use re\*(C'\fR enables
2360 .Sp
2361 Previously, running \f(CW\*(C`no re\*(C'\fR would turn off only a few things. Now it
2362 can turn off all the enabled things. For example, the only way to
2363 stop debugging, once enabled, was to exit the enclosing block; that is
2364 now fixed.
2365 .IP \(bu 4
2366 \&\f(CW\*(C`pack("D", $x)\*(C'\fR and \f(CW\*(C`pack("F", $x)\*(C'\fR now zero the padding on x86 long
2367 double builds. Under some build options on GCC 4.8 and later, they used
2368 to either overwrite the zero-initialized padding, or bypass the
2369 initialized buffer entirely. This caused \fIop/pack.t\fR to fail.
2370 [GH #14554] <https://github.com/Perl/perl5/issues/14554>
2371 .IP \(bu 4
2372 Extending an array cloned from a parent thread could result in "Modification of
2373 a read-only value attempted" errors when attempting to modify the new elements.
2374 [GH #14605] <https://github.com/Perl/perl5/issues/14605>
2375 .IP \(bu 4
2376 An assertion failure and subsequent crash with \f(CW\*(C`*x=<y>\*(C'\fR has been fixed.
2377 [GH #14493] <https://github.com/Perl/perl5/issues/14493>
2378 .IP \(bu 4
2379 A possible crashing/looping bug related to compiling lexical subs has been
2380 fixed.
2381 [GH #14596] <https://github.com/Perl/perl5/issues/14596>
2382 .IP \(bu 4
2383 UTF\-8 now works correctly in function names, in unquoted HERE-document
2384 terminators, and in variable names used as array indexes.
2385 [GH #14601] <https://github.com/Perl/perl5/issues/14601>
2386 .IP \(bu 4
2387 Repeated global pattern matches in scalar context on large tainted strings were
2388 exponentially slow depending on the current match position in the string.
2389 [GH #14238] <https://github.com/Perl/perl5/issues/14238>
2390 .IP \(bu 4
2391 Various crashes due to the parser getting confused by syntax errors have been
2392 fixed.
2393 [GH #14496] <https://github.com/Perl/perl5/issues/14496>
2394 [GH #14497] <https://github.com/Perl/perl5/issues/14497>
2395 [GH #14548] <https://github.com/Perl/perl5/issues/14548>
2396 [GH #14564] <https://github.com/Perl/perl5/issues/14564>
2397 .IP \(bu 4
2398 \&\f(CW\*(C`split\*(C'\fR in the scope of lexical \f(CW$_\fR has been fixed not to fail assertions.
2399 [GH #14483] <https://github.com/Perl/perl5/issues/14483>
2400 .IP \(bu 4
2401 \&\f(CW\*(C`my $x : attr\*(C'\fR syntax inside various list operators no longer fails
2402 assertions.
2403 [GH #14500] <https://github.com/Perl/perl5/issues/14500>
2404 .IP \(bu 4
2405 An \f(CW\*(C`@\*(C'\fR sign in quotes followed by a non-ASCII digit (which is not a valid
2406 identifier) would cause the parser to crash, instead of simply trying the
2407 \&\f(CW\*(C`@\*(C'\fR as literal. This has been fixed.
2408 [GH #14553] <https://github.com/Perl/perl5/issues/14553>
2409 .IP \(bu 4
2410 \&\f(CW\*(C`*bar::=*foo::=*glob_with_hash\*(C'\fR has been crashing since Perl 5.14, but no
2411 longer does.
2412 [GH #14512] <https://github.com/Perl/perl5/issues/14512>
2413 .IP \(bu 4
2414 \&\f(CW\*(C`foreach\*(C'\fR in scalar context was not pushing an item on to the stack, resulting
2415 in bugs. (\f(CW\*(C`print\ 4,\ scalar\ do\ {\ foreach(@x){}\ }\ +\ 1\*(C'\fR would print 5.)
2416 It has been fixed to return \f(CW\*(C`undef\*(C'\fR.
2417 [GH #14569] <https://github.com/Perl/perl5/issues/14569>
2418 .IP \(bu 4
2419 Several cases of data used to store environment variable contents in core C
2420 code being potentially overwritten before being used have been fixed.
2421 [GH #14476] <https://github.com/Perl/perl5/issues/14476>
2422 .IP \(bu 4
2423 Some patterns starting with \f(CW\*(C`/.*..../\*(C'\fR matched against long strings have
2424 been slow since v5.8, and some of the form \f(CW\*(C`/.*..../i\*(C'\fR have been slow
2425 since v5.18. They are now all fast again.
2426 [GH #14475] <https://github.com/Perl/perl5/issues/14475>.
2427 .IP \(bu 4
2428 The original visible value of \f(CW$/\fR is now preserved when it is set to
2429 an invalid value. Previously if you set \f(CW$/\fR to a reference to an
2430 array, for example, perl would produce a runtime error and not set
2431 \&\f(CW\*(C`PL_rs\*(C'\fR, but Perl code that checked \f(CW$/\fR would see the array
2432 reference.
2433 [GH #14245] <https://github.com/Perl/perl5/issues/14245>.
2434 .IP \(bu 4
2435 In a regular expression pattern, a POSIX class, like \f(CW\*(C`[:ascii:]\*(C'\fR, must
2436 be inside a bracketed character class, like \f(CW\*(C`qr/[[:ascii:]]/\*(C'\fR. A
2437 warning is issued when something looking like a POSIX class is not
2438 inside a bracketed class. That warning wasn't getting generated when
2439 the POSIX class was negated: \f(CW\*(C`[:^ascii:]\*(C'\fR. This is now fixed.
2440 .IP \(bu 4
2441 Perl 5.14.0 introduced a bug whereby \f(CW\*(C`eval\ {\ LABEL:\ }\*(C'\fR would crash. This
2442 has been fixed.
2443 [GH #14438] <https://github.com/Perl/perl5/issues/14438>.
2444 .IP \(bu 4
2445 Various crashes due to the parser getting confused by syntax errors have
2446 been fixed.
2447 [GH #14421] <https://github.com/Perl/perl5/issues/14421>.
2448 [GH #14472] <https://github.com/Perl/perl5/issues/14472>.
2449 [GH #14480] <https://github.com/Perl/perl5/issues/14480>.
2450 [GH #14447] <https://github.com/Perl/perl5/issues/14447>.
2451 .IP \(bu 4
2452 Code like \f(CW\*(C`/$a[/\*(C'\fR used to read the next line of input and treat it as
2453 though it came immediately after the opening bracket. Some invalid code
2454 consequently would parse and run, but some code caused crashes, so this is
2455 now disallowed.
2456 [GH #14462] <https://github.com/Perl/perl5/issues/14462>.
2457 .IP \(bu 4
2458 Fix argument underflow for \f(CW\*(C`pack\*(C'\fR.
2459 [GH #14525] <https://github.com/Perl/perl5/issues/14525>.
2460 .IP \(bu 4
2461 Fix handling of non-strict \f(CW\*(C`\ex{}\*(C'\fR. Now \f(CW\*(C`\ex{}\*(C'\fR is equivalent to \f(CW\*(C`\ex{0}\*(C'\fR
2462 instead of faulting.
2463 .IP \(bu 4
2464 \&\f(CW\*(C`stat \-t\*(C'\fR is now no longer treated as stackable, just like \f(CW\*(C`\-t stat\*(C'\fR.
2465 [GH #14499] <https://github.com/Perl/perl5/issues/14499>.
2466 .IP \(bu 4
2467 The following no longer causes a SEGV: \f(CW\*(C`qr{x+(y(?0))*}\*(C'\fR.
2468 .IP \(bu 4
2469 Fixed infinite loop in parsing backrefs in regexp patterns.
2470 .IP \(bu 4
2471 Several minor bug fixes in behavior of Infinity and NaN, including
2472 warnings when stringifying Infinity-like or NaN-like strings. For example,
2473 "NaNcy" doesn't numify to NaN anymore.
2474 .IP \(bu 4
2475 A bug in regular expression patterns that could lead to segfaults and
2476 other crashes has been fixed. This occurred only in patterns compiled
2477 with \f(CW\*(C`/i\*(C'\fR while taking into account the current POSIX locale (which usually
2478 means they have to be compiled within the scope of \f(CW\*(C`use\ locale\*(C'\fR),
2479 and there must be a string of at least 128 consecutive bytes to match.
2480 [GH #14389] <https://github.com/Perl/perl5/issues/14389>.
2481 .IP \(bu 4
2482 \&\f(CW\*(C`s///g\*(C'\fR now works on very long strings (where there are more than 2
2483 billion iterations) instead of dying with 'Substitution loop'.
2484 [GH #11742] <https://github.com/Perl/perl5/issues/11742>.
2485 [GH #14190] <https://github.com/Perl/perl5/issues/14190>.
2486 .IP \(bu 4
2487 \&\f(CW\*(C`gmtime\*(C'\fR no longer crashes with not-a-number values.
2488 [GH #14365] <https://github.com/Perl/perl5/issues/14365>.
2489 .IP \(bu 4
2490 \&\f(CW\*(C`\e()\*(C'\fR (a reference to an empty list), and \f(CW\*(C`y///\*(C'\fR with lexical \f(CW$_\fR in
2491 scope, could both do a bad write past the end of the stack. They have
2492 both been fixed to extend the stack first.
2493 .IP \(bu 4
2494 \&\f(CWprototype()\fR with no arguments used to read the previous item on the
2495 stack, so \f(CW\*(C`print\ "foo",\ prototype()\*(C'\fR would print foo's prototype.
2496 It has been fixed to infer \f(CW$_\fR instead.
2497 [GH #14376] <https://github.com/Perl/perl5/issues/14376>.
2498 .IP \(bu 4
2499 Some cases of lexical state subs declared inside predeclared subs could
2500 crash, for example when evalling a string including the name of an outer
2501 variable, but no longer do.
2502 .IP \(bu 4
2503 Some cases of nested lexical state subs inside anonymous subs could cause
2504 \&'Bizarre copy' errors or possibly even crashes.
2505 .IP \(bu 4
2506 When trying to emit warnings, perl's default debugger (\fIperl5db.pl\fR) was
2507 sometimes giving 'Undefined subroutine &DB::db_warn called' instead. This
2508 bug, which started to occur in Perl 5.18, has been fixed.
2509 [GH #14400] <https://github.com/Perl/perl5/issues/14400>.
2510 .IP \(bu 4
2511 Certain syntax errors in substitutions, such as \f(CW\*(C`s/${<>{})//\*(C'\fR, would
2512 crash, and had done so since Perl 5.10. (In some cases the crash did not
2513 start happening till 5.16.) The crash has, of course, been fixed.
2514 [GH #14391] <https://github.com/Perl/perl5/issues/14391>.
2515 .IP \(bu 4
2516 Fix a couple of string grow size calculation overflows; in particular,
2517 a repeat expression like \f(CW\*(C`33\ x\ ~3\*(C'\fR could cause a large buffer
2518 overflow since the new output buffer size was not correctly handled by
2519 \&\f(CWSvGROW()\fR. An expression like this now properly produces a memory wrap
2520 panic.
2521 [GH #14401] <https://github.com/Perl/perl5/issues/14401>.
2522 .IP \(bu 4
2523 \&\f(CW\*(C`formline("@...", "a");\*(C'\fR would crash. The \f(CW\*(C`FF_CHECKNL\*(C'\fR case in
2524 \&\f(CWpp_formline()\fR didn't set the pointer used to mark the chop position,
2525 which led to the \f(CW\*(C`FF_MORE\*(C'\fR case crashing with a segmentation fault.
2526 This has been fixed.
2527 [GH #14388] <https://github.com/Perl/perl5/issues/14388>.
2528 .IP \(bu 4
2529 A possible buffer overrun and crash when parsing a literal pattern during
2530 regular expression compilation has been fixed.
2531 [GH #14416] <https://github.com/Perl/perl5/issues/14416>.
2532 .IP \(bu 4
2533 \&\f(CWfchmod()\fR and \f(CWfutimes()\fR now set \f(CW$!\fR when they fail due to being
2534 passed a closed file handle.
2535 [GH #14073] <https://github.com/Perl/perl5/issues/14073>.
2536 .IP \(bu 4
2537 \&\f(CWop_free()\fR and \f(CWscalarvoid()\fR no longer crash due to a stack overflow
2538 when freeing a deeply recursive op tree.
2539 [GH #11866] <https://github.com/Perl/perl5/issues/11866>.
2540 .IP \(bu 4
2541 In Perl 5.20.0, \f(CW$^N\fR accidentally had the internal UTF\-8 flag turned off
2542 if accessed from a code block within a regular expression, effectively
2543 UTF\-8\-encoding the value. This has been fixed.
2544 [GH #14211] <https://github.com/Perl/perl5/issues/14211>.
2545 .IP \(bu 4
2546 A failed \f(CW\*(C`semctl\*(C'\fR call no longer overwrites existing items on the stack,
2547 which means that \f(CW\*(C`(semctl(\-1,0,0,0))[0]\*(C'\fR no longer gives an
2548 "uninitialized" warning.
2549 .IP \(bu 4
2550 \&\f(CW\*(C`else{foo()}\*(C'\fR with no space before \f(CW\*(C`foo\*(C'\fR is now better at assigning the
2551 right line number to that statement.
2552 [GH #14070] <https://github.com/Perl/perl5/issues/14070>.
2553 .IP \(bu 4
2554 Sometimes the assignment in \f(CW\*(C`@array = split\*(C'\fR gets optimised so that \f(CW\*(C`split\*(C'\fR
2555 itself writes directly to the array. This caused a bug, preventing this
2556 assignment from being used in lvalue context. So
2557 \&\f(CW\*(C`(@a=split//,"foo")=bar()\*(C'\fR was an error. (This bug probably goes back to
2558 Perl 3, when the optimisation was added.) It has now been fixed.
2559 [GH #14183] <https://github.com/Perl/perl5/issues/14183>.
2560 .IP \(bu 4
2561 When an argument list fails the checks specified by a subroutine
2562 signature (which is still an experimental feature), the resulting error
2563 messages now give the file and line number of the caller, not of the
2564 called subroutine.
2565 [GH #13643] <https://github.com/Perl/perl5/issues/13643>.
2566 .IP \(bu 4
2567 The flip-flop operators (\f(CW\*(C`..\*(C'\fR and \f(CW\*(C`...\*(C'\fR in scalar context) used to maintain
2568 a separate state for each recursion level (the number of times the
2569 enclosing sub was called recursively), contrary to the documentation. Now
2570 each closure has one internal state for each flip-flop.
2571 [GH #14110] <https://github.com/Perl/perl5/issues/14110>.
2572 .IP \(bu 4
2573 The flip-flop operator (\f(CW\*(C`..\*(C'\fR in scalar context) would return the same
2574 scalar each time, unless the containing subroutine was called recursively.
2575 Now it always returns a new scalar.
2576 [GH #14110] <https://github.com/Perl/perl5/issues/14110>.
2577 .IP \(bu 4
2578 \&\f(CW\*(C`use\*(C'\fR, \f(CW\*(C`no\*(C'\fR, statement labels, special blocks (\f(CW\*(C`BEGIN\*(C'\fR) and pod are now
2579 permitted as the first thing in a \f(CW\*(C`map\*(C'\fR or \f(CW\*(C`grep\*(C'\fR block, the block after
2580 \&\f(CW\*(C`print\*(C'\fR or \f(CW\*(C`say\*(C'\fR (or other functions) returning a handle, and within
2581 \&\f(CW\*(C`${...}\*(C'\fR, \f(CW\*(C`@{...}\*(C'\fR, etc.
2582 [GH #14088] <https://github.com/Perl/perl5/issues/14088>.
2583 .IP \(bu 4
2584 The repetition operator \f(CW\*(C`x\*(C'\fR now propagates lvalue context to its left-hand
2585 argument when used in contexts like \f(CW\*(C`foreach\*(C'\fR. That allows
2586 \&\f(CW\*(C`for(($#that_array)x2)\ {\ ...\ }\*(C'\fR to work as expected if the loop modifies
2587 \&\f(CW$_\fR.
2588 .IP \(bu 4
2589 \&\f(CW\*(C`(...) x ...\*(C'\fR in scalar context used to corrupt the stack if one operand
2590 was an object with "x" overloading, causing erratic behavior.
2591 [GH #13811] <https://github.com/Perl/perl5/issues/13811>.
2592 .IP \(bu 4
2593 Assignment to a lexical scalar is often optimised away; for example in
2594 \&\f(CW\*(C`my $x; $x = $y + $z\*(C'\fR, the assign operator is optimised away and the add
2595 operator writes its result directly to \f(CW$x\fR. Various bugs related to
2596 this optimisation have been fixed. Certain operators on the right-hand
2597 side would sometimes fail to assign the value at all or assign the wrong
2598 value, or would call STORE twice or not at all on tied variables. The
2599 operators affected were \f(CW\*(C`$foo++\*(C'\fR, \f(CW\*(C`$foo\-\-\*(C'\fR, and \f(CW\*(C`\-$foo\*(C'\fR under \f(CW\*(C`use
2600 integer\*(C'\fR, \f(CW\*(C`chomp\*(C'\fR, \f(CW\*(C`chr\*(C'\fR and \f(CW\*(C`setpgrp\*(C'\fR.
2601 .IP \(bu 4
2602 List assignments were sometimes buggy if the same scalar ended up on both
2603 sides of the assignment due to use of \f(CW\*(C`tied\*(C'\fR, \f(CW\*(C`values\*(C'\fR or \f(CW\*(C`each\*(C'\fR. The
2604 result would be the wrong value getting assigned.
2605 .IP \(bu 4
2606 \&\f(CWsetpgrp($nonzero)\fR (with one argument) was accidentally changed in 5.16
2607 to mean \f(CWsetpgrp(0)\fR. This has been fixed.
2608 .IP \(bu 4
2609 \&\f(CW\*(C`_\|_SUB_\|_\*(C'\fR could return the wrong value or even corrupt memory under the
2610 debugger (the \f(CW\*(C`\-d\*(C'\fR switch) and in subs containing \f(CW\*(C`eval $string\*(C'\fR.
2611 .IP \(bu 4
2612 When \f(CW\*(C`sub\ ()\ {\ $var\ }\*(C'\fR becomes inlinable, it now returns a different
2613 scalar each time, just as a non-inlinable sub would, though Perl still
2614 optimises the copy away in cases where it would make no observable
2615 difference.
2616 .IP \(bu 4
2617 \&\f(CW\*(C`my\ sub\ f\ ()\ {\ $var\ }\*(C'\fR and \f(CW\*(C`sub\ ()\ :\ attr\ {\ $var\ }\*(C'\fR are no longer
2618 eligible for inlining. The former would crash; the latter would just
2619 throw the attributes away. An exception is made for the little-known
2620 \&\f(CW\*(C`:method\*(C'\fR attribute, which does nothing much.
2621 .IP \(bu 4
2622 Inlining of subs with an empty prototype is now more consistent than
2623 before. Previously, a sub with multiple statements, of which all but the last
2624 were optimised away, would be inlinable only if it were an anonymous sub
2625 containing a string \f(CW\*(C`eval\*(C'\fR or \f(CW\*(C`state\*(C'\fR declaration or closing over an
2626 outer lexical variable (or any anonymous sub under the debugger). Now any
2627 sub that gets folded to a single constant after statements have been
2628 optimised away is eligible for inlining. This applies to things like \f(CW\*(C`sub
2629 () { jabber() if DEBUG; 42 }\*(C'\fR.
2630 .Sp
2631 Some subroutines with an explicit \f(CW\*(C`return\*(C'\fR were being made inlinable,
2632 contrary to the documentation, Now \f(CW\*(C`return\*(C'\fR always prevents inlining.
2633 .IP \(bu 4
2634 On some systems, such as VMS, \f(CW\*(C`crypt\*(C'\fR can return a non-ASCII string. If a
2635 scalar assigned to had contained a UTF\-8 string previously, then \f(CW\*(C`crypt\*(C'\fR
2636 would not turn off the UTF\-8 flag, thus corrupting the return value. This
2637 would happen with \f(CW\*(C`$lexical\ =\ crypt\ ...\*(C'\fR.
2638 .IP \(bu 4
2639 \&\f(CW\*(C`crypt\*(C'\fR no longer calls \f(CW\*(C`FETCH\*(C'\fR twice on a tied first argument.
2640 .IP \(bu 4
2641 An unterminated here-doc on the last line of a quote-like operator
2642 (\f(CW\*(C`qq[${ <<END }]\*(C'\fR, \f(CW\*(C`/(?{ <<END })/\*(C'\fR) no longer causes a double free. It
2643 started doing so in 5.18.
2644 .IP \(bu 4
2645 \&\f(CWindex()\fR and \f(CWrindex()\fR no longer crash when used on strings over 2GB in
2646 size.
2647 [GH #13700] <https://github.com/Perl/perl5/issues/13700>.
2648 .IP \(bu 4
2649 A small, previously intentional, memory leak in
2650 \&\f(CW\*(C`PERL_SYS_INIT\*(C'\fR/\f(CW\*(C`PERL_SYS_INIT3\*(C'\fR on Win32 builds was fixed. This might
2651 affect embedders who repeatedly create and destroy perl engines within
2652 the same process.
2653 .IP \(bu 4
2654 \&\f(CWPOSIX::localeconv()\fR now returns the data for the program's underlying
2655 locale even when called from outside the scope of \f(CW\*(C`use\ locale\*(C'\fR.
2656 .IP \(bu 4
2657 \&\f(CWPOSIX::localeconv()\fR now works properly on platforms which don't have
2658 \&\f(CW\*(C`LC_NUMERIC\*(C'\fR and/or \f(CW\*(C`LC_MONETARY\*(C'\fR, or for which Perl has been compiled
2659 to disregard either or both of these locale categories. In such
2660 circumstances, there are now no entries for the corresponding values in
2661 the hash returned by \f(CWlocaleconv()\fR.
2662 .IP \(bu 4
2663 \&\f(CWPOSIX::localeconv()\fR now marks appropriately the values it returns as
2664 UTF\-8 or not. Previously they were always returned as bytes, even if
2665 they were supposed to be encoded as UTF\-8.
2666 .IP \(bu 4
2667 On Microsoft Windows, within the scope of \f(CW\*(C`use\ locale\*(C'\fR, the following
2668 POSIX character classes gave results for many locales that did not
2669 conform to the POSIX standard:
2670 \&\f(CW\*(C`[[:alnum:]]\*(C'\fR,
2671 \&\f(CW\*(C`[[:alpha:]]\*(C'\fR,
2672 \&\f(CW\*(C`[[:blank:]]\*(C'\fR,
2673 \&\f(CW\*(C`[[:digit:]]\*(C'\fR,
2674 \&\f(CW\*(C`[[:graph:]]\*(C'\fR,
2675 \&\f(CW\*(C`[[:lower:]]\*(C'\fR,
2676 \&\f(CW\*(C`[[:print:]]\*(C'\fR,
2677 \&\f(CW\*(C`[[:punct:]]\*(C'\fR,
2678 \&\f(CW\*(C`[[:upper:]]\*(C'\fR,
2679 \&\f(CW\*(C`[[:word:]]\*(C'\fR,
2680 and
2681 \&\f(CW\*(C`[[:xdigit:]]\*(C'\fR.
2682 This was because the underlying Microsoft implementation does not
2683 follow the standard. Perl now takes special precautions to correct for
2684 this.
2685 .IP \(bu 4
2686 Many issues have been detected by Coverity <http://www.coverity.com/> and
2687 fixed.
2688 .IP \(bu 4
2689 \&\f(CWsystem()\fR and friends should now work properly on more Android builds.
2690 .Sp
2691 Due to an oversight, the value specified through \f(CW\*(C`\-Dtargetsh\*(C'\fR to \fIConfigure\fR
2692 would end up being ignored by some of the build process. This caused perls
2693 cross-compiled for Android to end up with defective versions of \f(CWsystem()\fR,
2694 \&\f(CWexec()\fR and backticks: the commands would end up looking for \f(CW\*(C`/bin/sh\*(C'\fR
2695 instead of \f(CW\*(C`/system/bin/sh\*(C'\fR, and so would fail for the vast majority
2696 of devices, leaving \f(CW$!\fR as \f(CW\*(C`ENOENT\*(C'\fR.
2697 .IP \(bu 4
2698 \&\f(CW\*(C`qr(...\e(...\e)...)\*(C'\fR,
2699 \&\f(CW\*(C`qr[...\e[...\e]...]\*(C'\fR,
2700 and
2701 \&\f(CW\*(C`qr{...\e{...\e}...}\*(C'\fR
2702 now work. Previously it was impossible to escape these three
2703 left-characters with a backslash within a regular expression pattern
2704 where otherwise they would be considered metacharacters, and the pattern
2705 opening delimiter was the character, and the closing delimiter was its
2706 mirror character.
2707 .IP \(bu 4
2708 \&\f(CW\*(C`s///e\*(C'\fR on tainted UTF\-8 strings corrupted \f(CWpos()\fR. This bug,
2709 introduced in 5.20, is now fixed.
2710 [GH #13948] <https://github.com/Perl/perl5/issues/13948>.
2711 .IP \(bu 4
2712 A non-word boundary in a regular expression (\f(CW\*(C`\eB\*(C'\fR) did not always
2713 match the end of the string; in particular \f(CW\*(C`q{} =~ /\eB/\*(C'\fR did not
2714 match. This bug, introduced in perl 5.14, is now fixed.
2715 [GH #13917] <https://github.com/Perl/perl5/issues/13917>.
2716 .IP \(bu 4
2717 \&\f(CW\*(C`" P" =~ /(?=.*P)P/\*(C'\fR should match, but did not. This is now fixed.
2718 [GH #13954] <https://github.com/Perl/perl5/issues/13954>.
2719 .IP \(bu 4
2720 Failing to compile \f(CW\*(C`use Foo\*(C'\fR in an \f(CW\*(C`eval\*(C'\fR could leave a spurious
2721 \&\f(CW\*(C`BEGIN\*(C'\fR subroutine definition, which would produce a "Subroutine
2722 BEGIN redefined" warning on the next use of \f(CW\*(C`use\*(C'\fR, or other \f(CW\*(C`BEGIN\*(C'\fR
2723 block.
2724 [GH #13926] <https://github.com/Perl/perl5/issues/13926>.
2725 .IP \(bu 4
2726 \&\f(CW\*(C`method { BLOCK } ARGS\*(C'\fR syntax now correctly parses the arguments if they
2727 begin with an opening brace.
2728 [GH #9085] <https://github.com/Perl/perl5/issues/9085>.
2729 .IP \(bu 4
2730 External libraries and Perl may have different ideas of what the locale is.
2731 This is problematic when parsing version strings if the locale's numeric
2732 separator has been changed. Version parsing has been patched to ensure
2733 it handles the locales correctly.
2734 [GH #13863] <https://github.com/Perl/perl5/issues/13863>.
2735 .IP \(bu 4
2736 A bug has been fixed where zero-length assertions and code blocks inside of a
2737 regex could cause \f(CW\*(C`pos\*(C'\fR to see an incorrect value.
2738 [GH #14016] <https://github.com/Perl/perl5/issues/14016>.
2739 .IP \(bu 4
2740 Dereferencing of constants now works correctly for typeglob constants. Previously
2741 the glob was stringified and its name looked up. Now the glob itself is used.
2742 [GH #9891] <https://github.com/Perl/perl5/issues/9891>
2743 .IP \(bu 4
2744 When parsing a sigil (\f(CW\*(C`$\*(C'\fR \f(CW\*(C`@\*(C'\fR \f(CW\*(C`%\*(C'\fR \f(CW\*(C`&)\*(C'\fR followed by braces,
2745 the parser no
2746 longer tries to guess whether it is a block or a hash constructor (causing a
2747 syntax error when it guesses the latter), since it can only be a block.
2748 .IP \(bu 4
2749 \&\f(CW\*(C`undef\ $reference\*(C'\fR now frees the referent immediately, instead of hanging on
2750 to it until the next statement.
2751 [GH #14032] <https://github.com/Perl/perl5/issues/14032>
2752 .IP \(bu 4
2753 Various cases where the name of a sub is used (autoload, overloading, error
2754 messages) used to crash for lexical subs, but have been fixed.
2755 .IP \(bu 4
2756 Bareword lookup now tries to avoid vivifying packages if it turns out the
2757 bareword is not going to be a subroutine name.
2758 .IP \(bu 4
2759 Compilation of anonymous constants (\fIe.g.\fR, \f(CW\*(C`sub () { 3 }\*(C'\fR) no longer deletes
2760 any subroutine named \f(CW\*(C`_\|_ANON_\|_\*(C'\fR in the current package. Not only was
2761 \&\f(CW\*(C`*_\|_ANON_\|_{CODE}\*(C'\fR cleared, but there was a memory leak, too. This bug goes
2762 back to Perl 5.8.0.
2763 .IP \(bu 4
2764 Stub declarations like \f(CW\*(C`sub f;\*(C'\fR and \f(CW\*(C`sub f ();\*(C'\fR no longer wipe out constants
2765 of the same name declared by \f(CW\*(C`use constant\*(C'\fR. This bug was introduced in Perl
2766 5.10.0.
2767 .IP \(bu 4
2768 \&\f(CW\*(C`qr/[\eN{named sequence}]/\*(C'\fR now works properly in many instances.
2769 .Sp
2770 Some names
2771 known to \f(CW\*(C`\eN{...}\*(C'\fR refer to a sequence of multiple characters, instead of the
2772 usual single character. Bracketed character classes generally only match
2773 single characters, but now special handling has been added so that they can
2774 match named sequences, but not if the class is inverted or the sequence is
2775 specified as the beginning or end of a range. In these cases, the only
2776 behavior change from before is a slight rewording of the fatal error message
2777 given when this class is part of a \f(CW\*(C`?[...])\*(C'\fR construct. When the \f(CW\*(C`[...]\*(C'\fR
2778 stands alone, the same non-fatal warning as before is raised, and only the
2779 first character in the sequence is used, again just as before.
2780 .IP \(bu 4
2781 Tainted constants evaluated at compile time no longer cause unrelated
2782 statements to become tainted.
2783 [GH #14059] <https://github.com/Perl/perl5/issues/14059>
2784 .IP \(bu 4
2785 \&\f(CW\*(C`open\ $$fh,\ ...\*(C'\fR, which vivifies a handle with a name like
2786 \&\f(CW"main::_GEN_0"\fR, was not giving the handle the right reference count, so
2787 a double free could happen.
2788 .IP \(bu 4
2789 When deciding that a bareword was a method name, the parser would get confused
2790 if an \f(CW\*(C`our\*(C'\fR sub with the same name existed, and look up the method in the
2791 package of the \f(CW\*(C`our\*(C'\fR sub, instead of the package of the invocant.
2792 .IP \(bu 4
2793 The parser no longer gets confused by \f(CW\*(C`\eU=\*(C'\fR within a double-quoted string. It
2794 used to produce a syntax error, but now compiles it correctly.
2795 [GH #10882] <https://github.com/Perl/perl5/issues/10882>
2796 .IP \(bu 4
2797 It has always been the intention for the \f(CW\*(C`\-B\*(C'\fR and \f(CW\*(C`\-T\*(C'\fR file test operators to
2798 treat UTF\-8 encoded files as text. (perlfunc has
2799 been updated to say this.) Previously, it was possible for some files to be
2800 considered UTF\-8 that actually weren't valid UTF\-8. This is now fixed. The
2801 operators now work on EBCDIC platforms as well.
2802 .IP \(bu 4
2803 Under some conditions warning messages raised during regular expression pattern
2804 compilation were being output more than once. This has now been fixed.
2805 .IP \(bu 4
2806 Perl 5.20.0 introduced a regression in which a UTF\-8 encoded regular
2807 expression pattern that contains a single ASCII lowercase letter did not
2808 match its uppercase counterpart. That has been fixed in both 5.20.1 and
2809 5.22.0.
2810 [GH #14051] <https://github.com/Perl/perl5/issues/14051>
2811 .IP \(bu 4
2812 Constant folding could incorrectly suppress warnings if lexical warnings
2813 (\f(CW\*(C`use warnings\*(C'\fR or \f(CW\*(C`no warnings\*(C'\fR) were not in effect and \f(CW$^W\fR were
2814 false at compile time and true at run time.
2815 .IP \(bu 4
2816 Loading Unicode tables during a regular expression match could cause assertion
2817 failures under debugging builds if the previous match used the very same
2818 regular expression.
2819 [GH #14081] <https://github.com/Perl/perl5/issues/14081>
2820 .IP \(bu 4
2821 Thread cloning used to work incorrectly for lexical subs, possibly causing
2822 crashes or double frees on exit.
2823 .IP \(bu 4
2824 Since Perl 5.14.0, deleting \f(CW$SomePackage::{_\|_ANON_\|_}\fR and then undefining an
2825 anonymous subroutine could corrupt things internally, resulting in
2826 Devel::Peek crashing or B.pm giving nonsensical data. This has been
2827 fixed.
2828 .IP \(bu 4
2829 \&\f(CW\*(C`(caller\ $n)[3]\*(C'\fR now reports names of lexical subs, instead of
2830 treating them as \f(CW"(unknown)"\fR.
2831 .IP \(bu 4
2832 \&\f(CW\*(C`sort subname LIST\*(C'\fR now supports using a lexical sub as the comparison
2833 routine.
2834 .IP \(bu 4
2835 Aliasing (\fIe.g.\fR, via \f(CW\*(C`*x\ =\ *y\*(C'\fR) could confuse list assignments that mention the
2836 two names for the same variable on either side, causing wrong values to be
2837 assigned.
2838 [GH #5788] <https://github.com/Perl/perl5/issues/5788>
2839 .IP \(bu 4
2840 Long here-doc terminators could cause a bad read on short lines of input. This
2841 has been fixed. It is doubtful that any crash could have occurred. This bug
2842 goes back to when here-docs were introduced in Perl 3.000 twenty-five years
2843 ago.
2844 .IP \(bu 4
2845 An optimization in \f(CW\*(C`split\*(C'\fR to treat \f(CW\*(C`split\ /^/\*(C'\fR like \f(CW\*(C`split\ /^/m\*(C'\fR had the
2846 unfortunate side-effect of also treating \f(CW\*(C`split\ /\eA/\*(C'\fR like \f(CW\*(C`split\ /^/m\*(C'\fR,
2847 which it should not. This has been fixed. (Note, however, that \f(CW\*(C`split\ /^x/\*(C'\fR
2848 does not behave like \f(CW\*(C`split\ /^x/m\*(C'\fR, which is also considered to be a bug and
2849 will be fixed in a future version.)
2850 [GH #14086] <https://github.com/Perl/perl5/issues/14086>
2851 .IP \(bu 4
2852 The little-known \f(CW\*(C`my\ Class\ $var\*(C'\fR syntax (see fields and attributes)
2853 could get confused in the scope of \f(CW\*(C`use utf8\*(C'\fR if \f(CW\*(C`Class\*(C'\fR were a constant
2854 whose value contained Latin\-1 characters.
2855 .IP \(bu 4
2856 Locking and unlocking values via Hash::Util or \f(CW\*(C`Internals::SvREADONLY\*(C'\fR
2857 no longer has any effect on values that were read-only to begin with.
2858 Previously, unlocking such values could result in crashes, hangs or
2859 other erratic behavior.
2860 .IP \(bu 4
2861 Some unterminated \f(CW\*(C`(?(...)...)\*(C'\fR constructs in regular expressions would
2862 either crash or give erroneous error messages. \f(CW\*(C`/(?(1)/\*(C'\fR is one such
2863 example.
2864 .IP \(bu 4
2865 \&\f(CW\*(C`pack\ "w",\ $tied\*(C'\fR no longer calls FETCH twice.
2866 .IP \(bu 4
2867 List assignments like \f(CW\*(C`($x,\ $z)\ =\ (1,\ $y)\*(C'\fR now work correctly if \f(CW$x\fR and
2868 \&\f(CW$y\fR have been aliased by \f(CW\*(C`foreach\*(C'\fR.
2869 .IP \(bu 4
2870 Some patterns including code blocks with syntax errors, such as
2871 \&\f(CW\*(C`/\ (?{(^{})/\*(C'\fR, would hang or fail assertions on debugging builds. Now
2872 they produce errors.
2873 .IP \(bu 4
2874 An assertion failure when parsing \f(CW\*(C`sort\*(C'\fR with debugging enabled has been
2875 fixed.
2876 [GH #14087] <https://github.com/Perl/perl5/issues/14087>.
2877 .IP \(bu 4
2878 \&\f(CW\*(C`*a\ =\ *b;\ @a\ =\ split\ //,\ $b[1]\*(C'\fR could do a bad read and produce junk
2879 results.
2880 .IP \(bu 4
2881 In \f(CW\*(C`()\ =\ @array\ =\ split\*(C'\fR, the \f(CW\*(C`()\ =\*(C'\fR at the beginning no longer confuses
2882 the optimizer into assuming a limit of 1.
2883 .IP \(bu 4
2884 Fatal warnings no longer prevent the output of syntax errors.
2885 [GH #14155] <https://github.com/Perl/perl5/issues/14155>.
2886 .IP \(bu 4
2887 Fixed a NaN double-to-long-double conversion error on VMS. For quiet NaNs
2888 (and only on Itanium, not Alpha) negative infinity instead of NaN was
2889 produced.
2890 .IP \(bu 4
2891 Fixed the issue that caused \f(CW\*(C`make distclean\*(C'\fR to incorrectly leave some
2892 files behind.
2893 [GH #14108] <https://github.com/Perl/perl5/issues/14108>.
2894 .IP \(bu 4
2895 AIX now sets the length in \f(CW\*(C`getsockopt\*(C'\fR correctly.
2896 [GH #13484] <https://github.com/Perl/perl5/issues/13484>.
2897 [cpan #91183] <https://rt.cpan.org/Ticket/Display.html?id=91183>.
2898 [cpan #85570] <https://rt.cpan.org/Ticket/Display.html?id=85570>.
2899 .IP \(bu 4
2900 The optimization phase of a regexp compilation could run "forever" and
2901 exhaust all memory under certain circumstances; now fixed.
2902 [GH #13984] <https://github.com/Perl/perl5/issues/13984>.
2903 .IP \(bu 4
2904 The test script \fIt/op/crypt.t\fR now uses the SHA\-256 algorithm if the
2905 default one is disabled, rather than giving failures.
2906 [GH #13715] <https://github.com/Perl/perl5/issues/13715>.
2907 .IP \(bu 4
2908 Fixed an off-by-one error when setting the size of a shared array.
2909 [GH #14151] <https://github.com/Perl/perl5/issues/14151>.
2910 .IP \(bu 4
2911 Fixed a bug that could cause perl to enter an infinite loop during
2912 compilation. In particular, a \f(CWwhile(1)\fR within a sublist, \fIe.g.\fR
2913 .Sp
2914 .Vb 1
2915 \& sub foo { () = ($a, my $b, ($c, do { while(1) {} })) }
2916 .Ve
2917 .Sp
2918 The bug was introduced in 5.20.0
2919 [GH #14165] <https://github.com/Perl/perl5/issues/14165>.
2920 .IP \(bu 4
2921 On Win32, if a variable was \f(CW\*(C`local\*(C'\fR\-ized in a pseudo-process that later
2922 forked, restoring the original value in the child pseudo-process caused
2923 memory corruption and a crash in the child pseudo-process (and therefore the
2924 OS process).
2925 [GH #8641] <https://github.com/Perl/perl5/issues/8641>.
2926 .IP \(bu 4
2927 Calling \f(CW\*(C`write\*(C'\fR on a format with a \f(CW\*(C`^**\*(C'\fR field could produce a panic
2928 in \f(CWsv_chop()\fR if there were insufficient arguments or if the variable
2929 used to fill the field was empty.
2930 [GH #14255] <https://github.com/Perl/perl5/issues/14255>.
2931 .IP \(bu 4
2932 Non-ASCII lexical sub names now appear without trailing junk when they
2933 appear in error messages.
2934 .IP \(bu 4
2935 The \f(CW\*(C`\e@\*(C'\fR subroutine prototype no longer flattens parenthesized arrays
2936 (taking a reference to each element), but takes a reference to the array
2937 itself.
2938 [GH #9111] <https://github.com/Perl/perl5/issues/9111>.
2939 .IP \(bu 4
2940 A block containing nothing except a C\-style \f(CW\*(C`for\*(C'\fR loop could corrupt the
2941 stack, causing lists outside the block to lose elements or have elements
2942 overwritten. This could happen with \f(CW\*(C`map { for(...){...} } ...\*(C'\fR and with
2943 lists containing \f(CW\*(C`do { for(...){...} }\*(C'\fR.
2944 [GH #14269] <https://github.com/Perl/perl5/issues/14269>.
2945 .IP \(bu 4
2946 \&\f(CWscalar()\fR now propagates lvalue context, so that
2947 \&\f(CW\*(C`for(scalar($#foo))\ {\ ...\ }\*(C'\fR can modify \f(CW$#foo\fR through \f(CW$_\fR.
2948 .IP \(bu 4
2949 \&\f(CW\*(C`qr/@array(?{block})/\*(C'\fR no longer dies with "Bizarre copy of ARRAY".
2950 [GH #14292] <https://github.com/Perl/perl5/issues/14292>.
2951 .IP \(bu 4
2952 \&\f(CW\*(C`eval\ \*(Aq$variable\*(Aq\*(C'\fR in nested named subroutines would sometimes look up a
2953 global variable even with a lexical variable in scope.
2954 .IP \(bu 4
2955 In perl 5.20.0, \f(CW\*(C`sort CORE::fake\*(C'\fR where 'fake' is anything other than a
2956 keyword, started chopping off the last 6 characters and treating the result
2957 as a sort sub name. The previous behavior of treating \f(CW\*(C`CORE::fake\*(C'\fR as a
2958 sort sub name has been restored.
2959 [GH #14323] <https://github.com/Perl/perl5/issues/14323>.
2960 .IP \(bu 4
2961 Outside of \f(CW\*(C`use utf8\*(C'\fR, a single-character Latin\-1 lexical variable is
2962 disallowed. The error message for it, "Can't use global \f(CW$foo\fR...", was
2963 giving garbage instead of the variable name.
2964 .IP \(bu 4
2965 \&\f(CW\*(C`readline\*(C'\fR on a nonexistent handle was causing \f(CW\*(C`${^LAST_FH}\*(C'\fR to produce a
2966 reference to an undefined scalar (or fail an assertion). Now
2967 \&\f(CW\*(C`${^LAST_FH}\*(C'\fR ends up undefined.
2968 .IP \(bu 4
2969 \&\f(CW\*(C`(...) x ...\*(C'\fR in void context now applies scalar context to the left-hand
2970 argument, instead of the context the current sub was called in.
2971 [GH #14174] <https://github.com/Perl/perl5/issues/14174>.
2972 .SH "Known Problems"
2973 .IX Header "Known Problems"
2974 .IP \(bu 4
2975 \&\f(CW\*(C`pack\*(C'\fR\-ing a NaN on a perl compiled with Visual C 6 does not behave properly,
2976 leading to a test failure in \fIt/op/infnan.t\fR.
2977 [GH #14705] <https://github.com/Perl/perl5/issues/14705>
2978 .IP \(bu 4
2979 A goal is for Perl to be able to be recompiled to work reasonably well on any
2980 Unicode version. In Perl 5.22, though, the earliest such version is Unicode
2981 5.1 (current is 7.0).
2982 .IP \(bu 4
2983 EBCDIC platforms
2984 .RS 4
2985 .IP \(bu 4
2986 The \f(CW\*(C`cmp\*(C'\fR (and hence \f(CW\*(C`sort\*(C'\fR) operators do not necessarily give the
2987 correct results when both operands are UTF-EBCDIC encoded strings and
2988 there is a mixture of ASCII and/or control characters, along with other
2989 characters.
2990 .IP \(bu 4
2991 Ranges containing \f(CW\*(C`\eN{...}\*(C'\fR in the \f(CW\*(C`tr///\*(C'\fR (and \f(CW\*(C`y///\*(C'\fR)
2992 transliteration operators are treated differently than the equivalent
2993 ranges in regular expression patterns. They should, but don't, cause
2994 the values in the ranges to all be treated as Unicode code points, and
2995 not native ones. ("Version 8 Regular Expressions" in perlre gives
2996 details as to how it should work.)
2997 .IP \(bu 4
2998 Encode and encoding are mostly broken.
2999 .IP \(bu 4
3000 Many CPAN modules that are shipped with core show failing tests.
3001 .IP \(bu 4
3002 \&\f(CW\*(C`pack\*(C'\fR/\f(CW\*(C`unpack\*(C'\fR with \f(CW"U0"\fR format may not work properly.
3003 .RE
3004 .RS 4
3005 .RE
3006 .IP \(bu 4
3007 The following modules are known to have test failures with this version of
3008 Perl. In many cases, patches have been submitted, so there will hopefully be
3009 new releases soon:
3010 .RS 4
3011 .IP \(bu 4
3012 B::Generate version 1.50
3013 .IP \(bu 4
3014 B::Utils version 0.25
3015 .IP \(bu 4
3016 Coro version 6.42
3017 .IP \(bu 4
3018 Dancer version 1.3130
3019 .IP \(bu 4
3020 Data::Alias version 1.18
3021 .IP \(bu 4
3022 Data::Dump::Streamer version 2.38
3023 .IP \(bu 4
3024 Data::Util version 0.63
3025 .IP \(bu 4
3026 Devel::Spy version 0.07
3027 .IP \(bu 4
3028 invoker version 0.34
3029 .IP \(bu 4
3030 Lexical::Var version 0.009
3031 .IP \(bu 4
3032 LWP::ConsoleLogger version 0.000018
3033 .IP \(bu 4
3034 Mason version 2.22
3035 .IP \(bu 4
3036 NgxQueue version 0.02
3037 .IP \(bu 4
3038 Padre version 1.00
3039 .IP \(bu 4
3040 Parse::Keyword 0.08
3041 .RE
3042 .RS 4
3043 .RE
3044 .SH Obituary
3045 .IX Header "Obituary"
3046 Brian McCauley died on May 8, 2015. He was a frequent poster to Usenet, Perl
3047 Monks, and other Perl forums, and made several CPAN contributions under the
3048 nick NOBULL, including to the Perl FAQ. He attended almost every
3049 YAPC::Europe, and indeed, helped organise YAPC::Europe 2006 and the QA
3050 Hackathon 2009. His wit and his delight in intricate systems were
3051 particularly apparent in his love of board games; many Perl mongers will
3052 have fond memories of playing Fluxx and other games with Brian. He will be
3053 missed.
3054 .SH Acknowledgements
3055 .IX Header "Acknowledgements"
3056 Perl 5.22.0 represents approximately 12 months of development since Perl 5.20.0
3057 and contains approximately 590,000 lines of changes across 2,400 files from 94
3058 authors.
3059 .PP
3060 Excluding auto-generated files, documentation and release tools, there were
3061 approximately 370,000 lines of changes to 1,500 .pm, .t, .c and .h files.
3062 .PP
3063 Perl continues to flourish into its third decade thanks to a vibrant community
3064 of users and developers. The following people are known to have contributed the
3065 improvements that became Perl 5.22.0:
3066 .PP
3067 Aaron Crane, Abhijit Menon-Sen, Abigail, Alberto Simões, Alex Solovey, Alex
3068 Vandiver, Alexandr Ciornii, Alexandre (Midnite) Jousset, Andreas König,
3069 Andreas Voegele, Andrew Fresh, Andy Dougherty, Anthony Heading, Aristotle
3070 Pagaltzis, brian d foy, Brian Fraser, Chad Granum, Chris 'BinGOs' Williams,
3071 Craig A. Berry, Dagfinn Ilmari Mannsåker, Daniel Dragan, Darin McBride, Dave
3072 Rolsky, David Golden, David Mitchell, David Wheeler, Dmitri Tikhonov, Doug
3073 Bell, E. Choroba, Ed J, Eric Herman, Father Chrysostomos, George Greer, Glenn
3074 D. Golden, Graham Knop, H.Merijn Brand, Herbert Breunung, Hugo van der Sanden,
3075 James E Keenan, James McCoy, James Raspass, Jan Dubois, Jarkko Hietaniemi,
3076 Jasmine Ngan, Jerry D. Hedden, Jim Cromie, John Goodyear, kafka, Karen
3077 Etheridge, Karl Williamson, Kent Fredric, kmx, Lajos Veres, Leon Timmermans,
3078 Lukas Mai, Mathieu Arnold, Matthew Horsfall, Max Maischein, Michael Bunk,
3079 Nicholas Clark, Niels Thykier, Niko Tyni, Norman Koch, Olivier Mengué, Peter
3080 John Acklam, Peter Martini, Petr Písař, Philippe Bruhat (BooK), Pierre
3081 Bogossian, Rafael Garcia-Suarez, Randy Stauner, Reini Urban, Ricardo Signes,
3082 Rob Hoelz, Rostislav Skudnov, Sawyer X, Shirakata Kentaro, Shlomi Fish,
3083 Sisyphus, Slaven Rezic, Smylers, Steffen Müller, Steve Hay, Sullivan Beck,
3084 syber, Tadeusz Sośnierz, Thomas Sibley, Todd Rinaldo, Tony Cook, Vincent Pit,
3085 Vladimir Marek, Yaroslav Kuzmin, Yves Orton, Ævar Arnfjörð Bjarmason.
3086 .PP
3087 The list above is almost certainly incomplete as it is automatically generated
3088 from version control history. In particular, it does not include the names of
3089 the (very much appreciated) contributors who reported issues to the Perl bug
3090 tracker.
3091 .PP
3092 Many of the changes included in this version originated in the CPAN modules
3093 included in Perl's core. We're grateful to the entire CPAN community for
3094 helping Perl to flourish.
3095 .PP
3096 For a more complete list of all of Perl's historical contributors, please see
3097 the \fIAUTHORS\fR file in the Perl source distribution.
3098 .SH "Reporting Bugs"
3099 .IX Header "Reporting Bugs"
3100 If you find what you think is a bug, you might check the articles recently
3101 posted to the comp.lang.perl.misc newsgroup and the perl bug database at
3102 <https://rt.perl.org/>. There may also be information at
3103 <http://www.perl.org/>, the Perl Home Page.
3104 .PP
3105 If you believe you have an unreported bug, please run the perlbug program
3106 included with your release. Be sure to trim your bug down to a tiny but
3107 sufficient test case. Your bug report, along with the output of \f(CW\*(C`perl \-V\*(C'\fR,
3108 will be sent off to [email protected] to be analysed by the Perl porting team.
3109 .PP
3110 If the bug you are reporting has security implications, which make it
3111 inappropriate to send to a publicly archived mailing list, then please send it
3112 to perl5\-security\[email protected]. This points to a closed subscription
3113 unarchived mailing list, which includes all the core committers, who will be
3114 able to help assess the impact of issues, figure out a resolution, and help
3115 co-ordinate the release of patches to mitigate or fix the problem across all
3116 platforms on which Perl is supported. Please only use this address for
3117 security issues in the Perl core, not for modules independently distributed on
3118 CPAN.
3119 .SH "SEE ALSO"
3120 .IX Header "SEE ALSO"
3121 The \fIChanges\fR file for an explanation of how to view exhaustive details on
3122 what changed.
3123 .PP
3124 The \fIINSTALL\fR file for how to build Perl.
3125 .PP
3126 The \fIREADME\fR file for general stuff.
3127 .PP
3128 The \fIArtistic\fR and \fICopying\fR files for copyright information.