Commit | Line | Data |
---|---|---|
44691e6f AB |
1 | =encoding utf8 |
2 | ||
3 | =head1 NAME | |
4 | ||
5076a392 | 5 | perldelta - what is new for perl v5.14.0 |
c71a852f | 6 | |
5076a392 FC |
7 | =head1 DESCRIPTION |
8 | ||
9 | This document describes differences between the 5.12.0 release and | |
10 | the 5.14.0 release. | |
11 | ||
12 | Some of the bug fixes in this release have been backported to subsequent | |
13 | releases of 5.12.x. Those are indicated with the 5.12.x version in | |
14 | parentheses. | |
15 | ||
16 | XXX Go through the perl512*delta files and do that. | |
17 | ||
18 | =head1 Notice | |
19 | ||
20 | XXX Any important notices here | |
21 | ||
22 | =head1 Core Enhancements | |
23 | ||
1f539a1a | 24 | =head2 Unicode |
5076a392 | 25 | |
1f539a1a | 26 | =head3 Unicode Version 6.0 is now supported (mostly) |
5076a392 | 27 | |
1f539a1a FC |
28 | Perl comes with the Unicode 6.0 data base updated with |
29 | L<Corrigendum #8|http://www.unicode.org/versions/corrigendum8.html>, | |
30 | with one exception noted below. | |
31 | See L<http://unicode.org/versions/Unicode6.0.0> for details on the new | |
32 | release. Perl does not support any Unicode provisional properties, | |
33 | including the new ones for this release, but their database files are | |
34 | packaged with Perl. | |
5076a392 | 35 | |
1f539a1a | 36 | Unicode 6.0 has chosen to use the name C<BELL> for the character at U+1F514, |
e1b1739f | 37 | which is a symbol that looks like a bell, and is used in Japanese cell |
1f539a1a FC |
38 | phones. This conflicts with the long-standing Perl usage of having |
39 | C<BELL> mean the ASCII C<BEL> character, U+0007. In Perl 5.14, | |
40 | C<\N{BELL}> will continue to mean U+0007, but its use will generate a | |
41 | deprecated warning message, unless such warnings are turned off. The | |
42 | new name for U+0007 in Perl will be C<ALERT>, which corresponds nicely | |
43 | with the existing shorthand sequence for it, C<"\a">. C<\N{BEL}> will | |
44 | mean U+0007, with no warning given. The character at U+1F514 will not | |
45 | have a name in 5.14, but can be referred to by C<\N{U+1F514}>. The plan | |
46 | is that in Perl 5.16, C<\N{BELL}> will refer to U+1F514, and so all code | |
47 | that uses C<\N{BELL}> should convert by then to using C<\N{ALERT}>, | |
48 | C<\N{BEL}>, or C<"\a"> instead. | |
5076a392 | 49 | |
1f539a1a | 50 | =head3 Full functionality for C<use feature 'unicode_strings'> |
5076a392 | 51 | |
1f539a1a FC |
52 | This release provides full functionality for C<use feature |
53 | 'unicode_strings'>. Under its scope, all string operations executed and | |
54 | regular expressions compiled (even if executed outside its scope) have | |
55 | Unicode semantics. See L<feature>. | |
5076a392 | 56 | |
1f539a1a FC |
57 | This feature avoids most forms of the "Unicode Bug" (See |
58 | L<perlunicode/The "Unicode Bug"> for details.) If there is a | |
59 | possibility that your code will process Unicode strings, you are | |
60 | B<strongly> encouraged to use this subpragma to avoid nasty surprises. | |
5076a392 | 61 | |
1f539a1a | 62 | =head3 C<\N{I<name>}> and C<charnames> enhancements |
5076a392 | 63 | |
1f539a1a | 64 | =over |
5076a392 | 65 | |
1f539a1a | 66 | =item * |
5076a392 FC |
67 | |
68 | C<\N{}> and C<charnames::vianame> now know about the abbreviated | |
e1b1739f FC |
69 | character names listed by Unicode, such as NBSP, SHY, LRO, ZWJ, etc., all |
70 | the customary abbreviations for the C0 and C1 control characters (such as | |
71 | ACK, BEL, CAN, etc.), and a few new variants of some C1 full names that | |
72 | are in common usage. | |
5076a392 | 73 | |
1f539a1a FC |
74 | =item * |
75 | ||
959ad7d5 FC |
76 | Unicode has a number of named character sequences, in which particular sequences |
77 | of code points are given names. C<\N{...}> now recognizes these. | |
78 | ||
1f539a1a FC |
79 | =item * |
80 | ||
959ad7d5 FC |
81 | C<\N{}>, C<charnames::vianame>, C<charnames::viacode> now know about every |
82 | character in Unicode. Previously, they didn't know about the Hangul syllables | |
83 | nor a number of CJK (Chinese/Japanese/Korean) characters. | |
84 | ||
1f539a1a FC |
85 | =item * |
86 | ||
5076a392 FC |
87 | In the past, it was ineffective to override one of Perl's abbreviations |
88 | with your own custom alias. Now it works. | |
89 | ||
1f539a1a FC |
90 | =item * |
91 | ||
e1b1739f | 92 | You can also create a custom alias of the ordinal of a |
5076a392 FC |
93 | character, known by C<\N{...}>, C<charnames::vianame()>, and |
94 | C<charnames::viacode()>. Previously, an alias had to be to an official | |
95 | Unicode character name. This made it impossible to create an alias for | |
e1b1739f FC |
96 | a code point that had no name, such as those reserved for private |
97 | use. | |
5076a392 | 98 | |
1f539a1a FC |
99 | =item * |
100 | ||
959ad7d5 FC |
101 | A new function, C<charnames::string_vianame()>, has been added. |
102 | This function is a run-time version of C<\N{...}>, returning the string | |
103 | of characters whose Unicode name is its parameter. It can handle | |
104 | Unicode named character sequences, whereas the pre-existing | |
105 | C<charnames::vianame()> cannot, as the latter returns a single code | |
106 | point. | |
107 | ||
1f539a1a FC |
108 | =back |
109 | ||
5076a392 FC |
110 | See L<charnames> for details on all these changes. |
111 | ||
1f539a1a | 112 | =head3 Any unsigned value can be encoded as a character |
5076a392 | 113 | |
1f539a1a FC |
114 | With this release, Perl is adopting a model that any unsigned value can |
115 | be treated as a code point and encoded internally (as utf8) without | |
54c7bb16 | 116 | warnings - not just the code points that are legal in Unicode. |
1f539a1a FC |
117 | However, unless utf8 warnings have been |
118 | explicitly lexically turned off, outputting or performing a | |
119 | Unicode-defined operation (such as upper-casing) on such a code point | |
120 | will generate a warning. Attempting to input these using strict rules | |
121 | (such as with the C<:encoding('UTF-8')> layer) will continue to fail. | |
122 | Prior to this release the handling was very inconsistent, and incorrect | |
123 | in places. Also, the Unicode non-characters, some of which previously were | |
124 | erroneously considered illegal in places by Perl, contrary to the Unicode | |
125 | standard, are now always legal internally. But inputting or outputting | |
126 | them will work the same as for the non-legal Unicode code points, as the | |
127 | Unicode standard says they are illegal for "open interchange". | |
5076a392 | 128 | |
1f539a1a | 129 | =head3 New warnings categories for problematic (non-)Unicode code points. |
5076a392 | 130 | |
e1b1739f FC |
131 | Three new warnings subcategories of "utf8" have been added. These |
132 | allow you to turn off some "utf8" warnings, while allowing | |
133 | others warnings to remain on. The three categories are: | |
1f539a1a FC |
134 | C<surrogate> when UTF-16 surrogates are encountered; |
135 | C<nonchar> when Unicode non-character code points are encountered; | |
136 | and C<non_unicode> when code points that are above the legal Unicode | |
137 | maximum of 0x10FFFF are encountered. | |
5076a392 | 138 | |
1f539a1a | 139 | =head2 Regular Expressions |
5076a392 | 140 | |
1f539a1a | 141 | =head3 C<(?^...)> construct to signify default modifiers |
5076a392 | 142 | |
e1b1739f FC |
143 | An ASCII caret (also called a "circumflex accent") C<"^"> |
144 | immediately following a C<"(?"> in a regular expression | |
145 | now means that the subexpression does not inherit the | |
146 | surrounding modifiers such as C</i>, but reverts to the | |
5076a392 FC |
147 | Perl defaults. Any modifiers following the caret override the defaults. |
148 | ||
149 | The stringification of regular expressions now uses this | |
150 | notation. E.g., before, C<qr/hlagh/i> would be stringified as | |
151 | C<(?i-xsm:hlagh)>, but now it's stringified as C<(?^i:hlagh)>. | |
152 | ||
153 | The main purpose of this is to allow tests that rely on the | |
e1b1739f | 154 | stringification not to have to change when new modifiers are added. |
5076a392 FC |
155 | See L<perlre/Extended Patterns>. |
156 | ||
1f539a1a | 157 | =head3 C</d>, C</l>, C</u>, C</a>, and C</aa> modifiers |
5076a392 | 158 | |
959ad7d5 FC |
159 | Four new regular expression modifiers have been added. These are mutually |
160 | exclusive; one only can be turned on at a time. | |
5076a392 | 161 | |
959ad7d5 | 162 | The C</l> modifier says to compile the regular expression as if it were |
5076a392 FC |
163 | in the scope of C<use locale>, even if it is not. |
164 | ||
959ad7d5 | 165 | The C</u> modifier says to compile the regular expression as if it were |
5076a392 FC |
166 | in the scope of a C<use feature "unicode_strings"> pragma. |
167 | ||
e1b1739f | 168 | The C</d> (default) modifier is used to override any C<use locale> and |
5076a392 FC |
169 | C<use feature "unicode_strings"> pragmas that are in effect at the time |
170 | of compiling the regular expression. | |
171 | ||
959ad7d5 FC |
172 | The C</a> regular expression modifier restricts C<\s>, C<\d> and C<\w> and |
173 | the Posix (C<[[:posix:]]>) character classes to the ASCII range. The | |
174 | complements and C<\b> and C<\B> are correspondingly | |
175 | affected. Otherwise, C</a> behaves like the C</u> modifier, in that | |
e1b1739f | 176 | case-insensitive matching uses Unicode semantics. |
5076a392 | 177 | |
959ad7d5 FC |
178 | The C</aa> modifier is like C</a>, except that, in case-insensitive matching, no ASCII character will match a |
179 | non-ASCII character. For example, | |
5076a392 | 180 | |
959ad7d5 | 181 | 'k' =~ /\N{KELVIN SIGN}/ai |
5076a392 | 182 | |
959ad7d5 | 183 | will match; it won't under C</aa>. |
5076a392 | 184 | |
959ad7d5 | 185 | See L<perlre/Modifiers> for more detail. |
5076a392 | 186 | |
1f539a1a | 187 | =head3 Non-destructive substitution |
5076a392 | 188 | |
1f539a1a FC |
189 | The substitution (C<s///>) and transliteration |
190 | (C<y///>) operators now support an C</r> option that | |
191 | copies the input variable, carries out the substitution on | |
192 | the copy and returns the result. The original remains unmodified. | |
5076a392 | 193 | |
1f539a1a FC |
194 | my $old = 'cat'; |
195 | my $new = $old =~ s/cat/dog/r; | |
196 | # $old is 'cat' and $new is 'dog' | |
5076a392 | 197 | |
1f539a1a | 198 | This is particularly useful with C<map>. See L<perlop> for more examples. |
5076a392 | 199 | |
1f539a1a | 200 | =head3 Reentrant regular expression engine |
5076a392 | 201 | |
1f539a1a FC |
202 | It is now safe to use regular expressions within C<(?{...})> and |
203 | C<(??{...})> code blocks inside regular expressions. | |
5076a392 | 204 | |
1f539a1a | 205 | These block are still experimental, however, and still have problems with |
e1b1739f | 206 | lexical (C<my>) variables and abnormal exiting. |
5076a392 | 207 | |
1f539a1a | 208 | =head3 C<use re '/flags';> |
5076a392 FC |
209 | |
210 | The C<re> pragma now has the ability to turn on regular expression flags | |
211 | till the end of the lexical scope: | |
212 | ||
213 | use re '/x'; | |
214 | "foo" =~ / (.+) /; # /x implied | |
215 | ||
216 | See L<re/"'/flags' mode"> for details. | |
217 | ||
1f539a1a | 218 | =head3 \o{...} for octals |
5076a392 | 219 | |
e1b1739f FC |
220 | There is a new octal escape sequence, C<"\o">, in double-quote-like |
221 | contexts. This construct allows large octal ordinals beyond the | |
1f539a1a FC |
222 | current max of 0777 to be represented. It also allows you to specify a |
223 | character in octal which can safely be concatenated with other regex | |
224 | snippets and which won't be confused with being a backreference to | |
225 | a regex capture group. See L<perlre/Capture groups>. | |
226 | ||
227 | =head3 Add C<\p{Titlecase}> as a synonym for C<\p{Title}> | |
228 | ||
229 | This synonym is added for symmetry with the Unicode property names | |
230 | C<\p{Uppercase}> and C<\p{Lowercase}>. | |
5076a392 | 231 | |
1f539a1a FC |
232 | =head3 Regular expression debugging output improvement |
233 | ||
234 | Regular expression debugging output (turned on by C<use re 'debug';>) now | |
235 | uses hexadecimal when escaping non-ASCII characters, instead of octal. | |
236 | ||
1e463951 FC |
237 | =head3 Return value of C<delete $+{...}> |
238 | ||
239 | Custom regular expression engines can now determine the return value of | |
240 | C<delete> on an entry of C<%+> or C<%->. | |
241 | ||
1f539a1a FC |
242 | =head2 Syntactical Enhancements |
243 | ||
244 | =head3 Array and hash container functions accept references | |
5076a392 FC |
245 | |
246 | All built-in functions that operate directly on array or hash | |
247 | containers now also accept hard references to arrays or hashes: | |
248 | ||
249 | |----------------------------+---------------------------| | |
250 | | Traditional syntax | Terse syntax | | |
251 | |----------------------------+---------------------------| | |
252 | | push @$arrayref, @stuff | push $arrayref, @stuff | | |
253 | | unshift @$arrayref, @stuff | unshift $arrayref, @stuff | | |
254 | | pop @$arrayref | pop $arrayref | | |
255 | | shift @$arrayref | shift $arrayref | | |
256 | | splice @$arrayref, 0, 2 | splice $arrayref, 0, 2 | | |
257 | | keys %$hashref | keys $hashref | | |
258 | | keys @$arrayref | keys $arrayref | | |
259 | | values %$hashref | values $hashref | | |
260 | | values @$arrayref | values $arrayref | | |
261 | | ($k,$v) = each %$hashref | ($k,$v) = each $hashref | | |
262 | | ($k,$v) = each @$arrayref | ($k,$v) = each $arrayref | | |
263 | |----------------------------+---------------------------| | |
264 | ||
265 | This allows these built-in functions to act on long dereferencing chains | |
266 | or on the return value of subroutines without needing to wrap them in | |
267 | C<@{}> or C<%{}>: | |
268 | ||
269 | push @{$obj->tags}, $new_tag; # old way | |
270 | push $obj->tags, $new_tag; # new way | |
271 | ||
272 | for ( keys %{$hoh->{genres}{artists}} ) {...} # old way | |
273 | for ( keys $hoh->{genres}{artists} ) {...} # new way | |
274 | ||
275 | For C<push>, C<unshift> and C<splice>, the reference will auto-vivify | |
276 | if it is not defined, just as if it were wrapped with C<@{}>. | |
277 | ||
5076a392 FC |
278 | For C<keys>, C<values>, C<each>, when overloaded dereferencing is |
279 | present, the overloaded dereference is used instead of dereferencing the | |
e1b1739f FC |
280 | underlying reftype. Warnings are issued about assumptions made in |
281 | ambiguous cases. | |
5076a392 | 282 | |
1f539a1a FC |
283 | =head3 Single term prototype |
284 | ||
285 | The C<+> prototype is a special alternative to C<$> that will act like | |
286 | C<\[@%]> when given a literal array or hash variable, but will otherwise | |
e1b1739f | 287 | force scalar context on the argument. See L<perlsub/Prototypes>. |
1f539a1a FC |
288 | |
289 | =head3 C<package> block syntax | |
290 | ||
291 | A package declaration can now contain a code block, in which case the | |
292 | declaration is in scope only inside that block. So C<package Foo { ... }> | |
293 | is precisely equivalent to C<{ package Foo; ... }>. It also works with | |
294 | a version number in the declaration, as in C<package Foo 1.2 { ... }>. | |
e1b1739f | 295 | See L<perlfunc>. |
1f539a1a FC |
296 | |
297 | =head3 Statement labels can appear in more places | |
298 | ||
299 | Statement labels can now occur before any type of statement or declaration, | |
300 | such as C<package>. | |
301 | ||
302 | =head3 Stacked labels | |
303 | ||
304 | Multiple statement labels can now appear before a single statement. | |
305 | ||
306 | =head3 Uppercase X/B allowed in hexadecimal/binary literals | |
307 | ||
308 | Literals may now use either upper case C<0X...> or C<0B...> prefixes, | |
309 | in addition to the already supported C<0x...> and C<0b...> | |
e1b1739f | 310 | syntax [perl #76296]. |
1f539a1a FC |
311 | |
312 | C, Ruby, Python and PHP already supported this syntax, and it makes | |
313 | Perl more internally consistent. A round-trip with C<eval sprintf | |
e1b1739f | 314 | "%#X", 0x10> now returns C<16>, the way C<eval sprintf "%#x", 0x10> does. |
1f539a1a | 315 | |
1e463951 FC |
316 | =head3 Overridable tie functions |
317 | ||
318 | C<tie>, C<tied> and C<untie> can now be overridden [perl #75902]. | |
319 | ||
1f539a1a FC |
320 | =head2 Exception Handling |
321 | ||
322 | Several changes have been made to the way C<die>, C<warn>, and C<$@> | |
323 | behave, in order to make them more reliable and consistent. | |
324 | ||
325 | When an exception is thrown inside an C<eval>, the exception is no | |
326 | longer at risk of being clobbered by code running during unwinding | |
327 | (e.g., destructors). Previously, the exception was written into C<$@> | |
328 | early in the throwing process, and would be overwritten if C<eval> was | |
329 | used internally in the destructor for an object that had to be freed | |
330 | while exiting from the outer C<eval>. Now the exception is written | |
331 | into C<$@> last thing before exiting the outer C<eval>, so the code | |
332 | running immediately thereafter can rely on the value in C<$@> correctly | |
333 | corresponding to that C<eval>. (C<$@> is still also set before exiting the | |
334 | C<eval>, for the sake of destructors that rely on this.) | |
335 | ||
336 | Likewise, a C<local $@> inside an C<eval> will no longer clobber any | |
337 | exception thrown in its scope. Previously, the restoration of C<$@> upon | |
338 | unwinding would overwrite any exception being thrown. Now the exception | |
339 | gets to the C<eval> anyway. So C<local $@> is safe before a C<die>. | |
340 | ||
341 | Exceptions thrown from object destructors no longer modify the C<$@> | |
342 | of the surrounding context. (If the surrounding context was exception | |
343 | unwinding, this used to be another way to clobber the exception being | |
344 | thrown.) Previously such an exception was | |
345 | sometimes emitted as a warning, and then either was | |
346 | string-appended to the surrounding C<$@> or completely replaced the | |
347 | surrounding C<$@>, depending on whether that exception and the surrounding | |
348 | C<$@> were strings or objects. Now, an exception in this situation is | |
349 | always emitted as a warning, leaving the surrounding C<$@> untouched. | |
350 | In addition to object destructors, this also affects any function call | |
351 | performed by XS code using the C<G_KEEPERR> flag. | |
352 | ||
353 | Warnings for C<warn> can now be objects, in the same way as exceptions | |
354 | for C<die>. If an object-based warning gets the default handling, | |
355 | of writing to standard error, it is stringified as | |
356 | before, with the file and line number appended. But | |
357 | a C<$SIG{__WARN__}> handler will now receive an | |
358 | object-based warning as an object, where previously it was passed the | |
359 | result of stringifying the object. | |
360 | ||
361 | =head2 Other Enhancements | |
362 | ||
363 | =head3 Assignment to C<$0> sets the legacy process name with C<prctl()> on Linux | |
364 | ||
365 | On Linux the legacy process name will be set with L<prctl(2)>, in | |
366 | addition to altering the POSIX name via C<argv[0]> as perl has done | |
367 | since version 4.000. Now system utilities that read the legacy process | |
368 | name such as ps, top and killall will recognize the name you set when | |
369 | assigning to C<$0>. The string you supply will be cut off at 16 bytes, | |
370 | this is a limitation imposed by Linux. | |
371 | ||
372 | =head3 C<srand()> now returns the seed | |
373 | ||
e1b1739f FC |
374 | This allows programs that need to have repeatable results not to have to come |
375 | up with their own seed-generating mechanism. Instead, they can use C<srand()> | |
376 | and stash the return value for future use. Typical is a test program which | |
1f539a1a | 377 | has too many combinations to test comprehensively in the time available to it |
e1b1739f | 378 | each run. It can test a random subset each time and, should there be a failure, |
1f539a1a | 379 | log the seed used for that run so that it can later be used to reproduce the |
e1b1739f | 380 | same results. |
1f539a1a FC |
381 | |
382 | =head3 printf-like functions understand post-1980 size modifiers | |
383 | ||
384 | Perl's printf and sprintf operators, and Perl's internal printf replacement | |
385 | function, now understand the C90 size modifiers "hh" (C<char>), "z" | |
386 | (C<size_t>), and "t" (C<ptrdiff_t>). Also, when compiled with a C99 | |
387 | compiler, Perl now understands the size modifier "j" (C<intmax_t>). | |
388 | ||
389 | So, for example, on any modern machine, C<sprintf('%hhd', 257)> returns '1'. | |
390 | ||
391 | =head3 New global variable C<${^GLOBAL_PHASE}> | |
5076a392 FC |
392 | |
393 | A new global variable, C<${^GLOBAL_PHASE}>, has been added to allow | |
394 | introspection of the current phase of the perl interpreter. It's explained in | |
395 | detail in L<perlvar/"${^GLOBAL_PHASE}"> and | |
396 | L<perlmod/"BEGIN, UNITCHECK, CHECK, INIT and END">. | |
397 | ||
1f539a1a | 398 | =head3 C<-d:-foo> calls C<Devel::foo::unimport> |
5076a392 FC |
399 | |
400 | The syntax C<-dI<B<:>foo>> was extended in 5.6.1 to make C<-dI<:fooB<=bar>>> | |
401 | equivalent to C<-MDevel::foo=bar>, which expands | |
402 | internally to C<use Devel::foo 'bar';>. | |
403 | F<perl> now allows prefixing the module name with C<->, with the same | |
404 | semantics as C<-M>, I<i.e.> | |
405 | ||
406 | =over 4 | |
407 | ||
408 | =item C<-d:-foo> | |
409 | ||
410 | Equivalent to C<-M-Devel::foo>, expands to | |
411 | C<no Devel::foo;>, calls C<< Devel::foo->unimport() >> | |
412 | if the method exists. | |
413 | ||
414 | =item C<-d:-foo=bar> | |
415 | ||
416 | Equivalent to C<-M-Devel::foo=bar>, expands to C<no Devel::foo 'bar';>, | |
417 | calls C<< Devel::foo->unimport('bar') >> if the method exists. | |
418 | ||
419 | =back | |
420 | ||
e1b1739f | 421 | This is particularly useful for suppressing the default actions of a |
5076a392 FC |
422 | C<Devel::*> module's C<import> method whilst still loading it for debugging. |
423 | ||
1f539a1a | 424 | =head3 Filehandle method calls load L<IO::File> on demand |
5076a392 FC |
425 | |
426 | When a method call on a filehandle would die because the method cannot | |
427 | be resolved, and L<IO::File> has not been loaded, Perl now loads L<IO::File> | |
428 | via C<require> and attempts method resolution again: | |
429 | ||
430 | open my $fh, ">", $file; | |
431 | $fh->binmode(":raw"); # loads IO::File and succeeds | |
432 | ||
433 | This also works for globs like STDOUT, STDERR and STDIN: | |
434 | ||
435 | STDOUT->autoflush(1); | |
436 | ||
437 | Because this on-demand load only happens if method resolution fails, the | |
438 | legacy approach of manually loading an L<IO::File> parent class for partial | |
439 | method support still works as expected: | |
440 | ||
441 | use IO::Handle; | |
442 | open my $fh, ">", $file; | |
443 | $fh->autoflush(1); # IO::File not loaded | |
444 | ||
1984204c FC |
445 | =head3 IPv6 support |
446 | ||
447 | The C<Socket> module provides new affordances for IPv6, | |
448 | including implementations of the C<Socket::getaddrinfo()> and | |
449 | C<Socket::getnameinfo()> functions, along with related constants, and a | |
450 | handful of new functions. See L<Socket>. | |
451 | ||
1f539a1a | 452 | =head3 DTrace probes now include package name |
5076a392 FC |
453 | |
454 | The DTrace probes now include an additional argument (C<arg3>) which contains | |
455 | the package the subroutine being entered or left was compiled in. | |
456 | ||
457 | For example using the following DTrace script: | |
458 | ||
459 | perl$target:::sub-entry | |
460 | { | |
461 | printf("%s::%s\n", copyinstr(arg0), copyinstr(arg3)); | |
462 | } | |
463 | ||
464 | and then running: | |
465 | ||
466 | perl -e'sub test { }; test' | |
467 | ||
468 | DTrace will print: | |
469 | ||
470 | main::test | |
471 | ||
9378886b FC |
472 | =head2 New C APIs |
473 | ||
1e463951 | 474 | See L</Internal Changes>. |
9378886b | 475 | |
5076a392 FC |
476 | =head1 Security |
477 | ||
948b8455 | 478 | =head2 User-defined regular expression properties |
5076a392 FC |
479 | |
480 | In L<perlunicode/"User-Defined Character Properties">, it says you can | |
481 | create custom properties by defining subroutines whose names begin with | |
948b8455 FC |
482 | "In" or "Is". However, Perl did not actually enforce that naming |
483 | restriction, so \p{foo::bar} could call foo::bar() if it existed. Now this | |
484 | convention has been enforced. | |
5076a392 | 485 | |
948b8455 FC |
486 | Also, Perl no longer allows a tainted regular expression to invoke a |
487 | user-defined. It simply dies instead [perl #82616]. | |
5076a392 FC |
488 | |
489 | =head1 Incompatible Changes | |
490 | ||
61752d82 FC |
491 | Perl 5.14.0 is not binary-compatible with any previous stable release. |
492 | ||
1e463951 FC |
493 | In addition to the sections that follow, see L</C API Changes>. |
494 | ||
61752d82 FC |
495 | =head2 Regular Expressions and String Escapes |
496 | ||
497 | =head3 C<\cI<X>> | |
5076a392 FC |
498 | |
499 | The backslash-c construct was designed as a way of specifying | |
500 | non-printable characters, but there were no restrictions (on ASCII | |
501 | platforms) on what the character following the C<c> could be. Now, that | |
502 | character must be one of the ASCII characters. | |
503 | ||
54c7bb16 | 504 | =head3 \400-\777 |
5076a392 | 505 | |
54c7bb16 | 506 | Use of C<\400>-C<\777> in regexes in certain circumstances has given |
5076a392 FC |
507 | different, anomalous behavior than their use in all other |
508 | double-quote-like contexts. Since 5.10.1, a deprecated warning message | |
509 | has been raised when this happens. Now, all double-quote-like contexts | |
510 | have the same behavior, namely to be equivalent to C<\x{100}> - | |
511 | C<\x{1FF}>, with no deprecation warning. Use of these values in the | |
512 | command line option C<"-0"> retains the current meaning to slurp input | |
513 | files whole; previously, this was documented only for C<"-0777">. It is | |
514 | recommended, however, because of various ambiguities, to use the new | |
515 | C<\o{...}> construct to represent characters in octal. | |
5076a392 | 516 | |
61752d82 | 517 | =head3 Most C<\p{}> properties are now immune to case-insensitive matching |
5076a392 | 518 | |
61752d82 FC |
519 | For most Unicode properties, it doesn't make sense to have them match |
520 | differently under C</i> case-insensitive matching than not. And doing | |
521 | so leads to unexpected results and potential security holes. For | |
522 | example | |
5076a392 | 523 | |
61752d82 | 524 | m/\p{ASCII_Hex_Digit}+/i |
5076a392 | 525 | |
61752d82 FC |
526 | could previously match non-ASCII characters because of the Unicode |
527 | matching rules. There were a number of bugs in this feature until an | |
528 | earlier release in the 5.13 series. Now this release reverts, and | |
529 | removes the feature completely except for the few properties where | |
530 | people have come to expect it, namely the ones where casing is an | |
531 | integral part of their functionality, such as C<m/\p{Uppercase}/i> and | |
532 | C<m/\p{Lowercase}/i>, both of which match the exact same code points, | |
533 | namely those matched by C<m/\p{Cased}/i>. Details are in | |
534 | L<perlrecharclass/Unicode Properties>. | |
5076a392 | 535 | |
61752d82 FC |
536 | XXX The mention of ‘until an earlier release in the 5.13 series’ needs to |
537 | change, but I do not fully understand what happened here. | |
5076a392 | 538 | |
61752d82 FC |
539 | User-defined property handlers that need to match differently under |
540 | C</i> must change to read the new boolean parameter passed to it which is | |
541 | non-zero if case-insensitive matching is in effect or 0 otherwise. See | |
542 | L<perluniprops/User-Defined Character Properties>. | |
5076a392 | 543 | |
61752d82 | 544 | =head3 \p{} implies Unicode semantics |
5076a392 | 545 | |
61752d82 FC |
546 | Now, a Unicode property match specified in the pattern will indicate |
547 | that the pattern is meant for matching according to Unicode rules, the way | |
548 | C<\x{}> does. | |
5076a392 | 549 | |
61752d82 | 550 | =head3 Regular expressions retain their localeness when interpolated |
5076a392 | 551 | |
61752d82 FC |
552 | Regular expressions compiled under C<"use locale"> now retain this when |
553 | interpolated into a new regular expression compiled outside a | |
554 | C<"use locale">, and vice-versa. | |
5076a392 | 555 | |
61752d82 FC |
556 | Previously, a regular expression interpolated into another one inherited |
557 | the localeness of the surrounding one, losing whatever state it | |
558 | originally had. This is considered a bug fix, but may trip up code that | |
559 | has come to rely on the incorrect behavior. | |
5076a392 | 560 | |
61752d82 | 561 | =head3 Stringification of regexes has changed |
5076a392 FC |
562 | |
563 | Default regular expression modifiers are now notated by using | |
564 | C<(?^...)>. Code relying on the old stringification will fail. The | |
565 | purpose of this is so that when new modifiers are added, such code will | |
566 | not have to change (after this one time), as the stringification will | |
567 | automatically incorporate the new modifiers. | |
568 | ||
569 | Code that needs to work properly with both old- and new-style regexes | |
f318e2e6 | 570 | can avoid the whole issue by using (for Perls since 5.9.5; see L<re>): |
5076a392 FC |
571 | |
572 | use re qw(regexp_pattern); | |
573 | my ($pat, $mods) = regexp_pattern($re_ref); | |
574 | ||
5076a392 FC |
575 | If the actual stringification is important, or older Perls need to be |
576 | supported, you can use something like the following: | |
577 | ||
578 | # Accept both old and new-style stringification | |
579 | my $modifiers = (qr/foobar/ =~ /\Q(?^/) ? '^' : '-xism'; | |
580 | ||
581 | And then use C<$modifiers> instead of C<-xism>. | |
582 | ||
61752d82 | 583 | =head3 Run-time code blocks in regular expressions inherit pragmata |
5076a392 | 584 | |
61752d82 FC |
585 | Code blocks in regular expressions (C<(?{...})> and C<(??{...})>) used not |
586 | to inherit any pragmata (strict, warnings, etc.) if the regular expression | |
587 | was compiled at run time as happens in cases like these two: | |
5076a392 | 588 | |
61752d82 FC |
589 | use re 'eval'; |
590 | $foo =~ $bar; # when $bar contains (?{...}) | |
591 | $foo =~ /$bar(?{ $finished = 1 })/; | |
592 | ||
593 | This was a bug, which has now been fixed. But it has the potential to break | |
594 | any code that was relying on it. | |
5076a392 | 595 | |
61752d82 | 596 | =head2 Stashes and Package Variables |
5076a392 | 597 | |
61752d82 | 598 | =head3 Localised tied hashes and arrays are no longed tied |
5076a392 | 599 | |
61752d82 | 600 | In the following: |
5076a392 | 601 | |
61752d82 FC |
602 | tie @a, ...; |
603 | { | |
604 | local @a; | |
605 | # here, @a is a now a new, untied array | |
606 | } | |
607 | # here, @a refers again to the old, tied array | |
5076a392 | 608 | |
61752d82 FC |
609 | The new local array used to be made tied too, which was fairly pointless, |
610 | and has now been fixed. This fix could however potentially cause a change | |
611 | in behaviour of some code. | |
5076a392 | 612 | |
61752d82 | 613 | =head3 Stashes are now always defined |
5076a392 | 614 | |
61752d82 FC |
615 | C<defined %Foo::> now always returns true, even when no symbols have yet been |
616 | defined in that package. | |
5076a392 | 617 | |
61752d82 FC |
618 | This is a side effect of removing a special case kludge in the tokeniser, |
619 | added for 5.10.0, to hide side effects of changes to the internal storage of | |
620 | hashes that drastically reduce their memory usage overhead. | |
621 | ||
622 | Calling defined on a stash has been deprecated since 5.6.0, warned on | |
623 | lexicals since 5.6.0, and warned for stashes (and other package | |
624 | variables) since 5.12.0. C<defined %hash> has always exposed an | |
625 | implementation detail - emptying a hash by deleting all entries from it does | |
626 | not make C<defined %hash> false, hence C<defined %hash> is not valid code to | |
627 | determine whether an arbitrary hash is empty. Instead, use the behaviour | |
628 | that an empty C<%hash> always returns false in a scalar context. | |
629 | ||
630 | =head3 Dereferencing typeglobs | |
5076a392 FC |
631 | |
632 | If you assign a typeglob to a scalar variable: | |
633 | ||
634 | $glob = *foo; | |
635 | ||
636 | the glob that is copied to C<$glob> is marked with a special flag | |
637 | indicating that the glob is just a copy. This allows subsequent assignments | |
638 | to C<$glob> to overwrite the glob. The original glob, however, is | |
639 | immutable. | |
640 | ||
641 | Many Perl operators did not distinguish between these two types of globs. | |
642 | This would result in strange behaviour in edge cases: C<untie $scalar> | |
643 | would do nothing if the last thing assigned to the scalar was a glob | |
644 | (because it treated it as C<untie *$scalar>, which unties a handle). | |
f318e2e6 | 645 | Assignment to a glob slot (e.g., C<*$glob = \@some_array>) would simply |
5076a392 FC |
646 | assign C<\@some_array> to C<$glob>. |
647 | ||
648 | To fix this, the C<*{}> operator (including the C<*foo> and C<*$foo> forms) | |
649 | has been modified to make a new immutable glob if its operand is a glob | |
e262cb24 FC |
650 | copy. This allows operators that make a distinction between globs and |
651 | scalars to be modified to treat only immutable globs as globs. (C<tie>, | |
652 | C<tied> and C<untie> have been left as they are for compatibility's sake, | |
61752d82 | 653 | but will warn. See L</Deprecations>.) |
5076a392 FC |
654 | |
655 | This causes an incompatible change in code that assigns a glob to the | |
656 | return value of C<*{}> when that operator was passed a glob copy. Take the | |
657 | following code, for instance: | |
658 | ||
659 | $glob = *foo; | |
660 | *$glob = *bar; | |
661 | ||
662 | The C<*$glob> on the second line returns a new immutable glob. That new | |
663 | glob is made an alias to C<*bar>. Then it is discarded. So the second | |
664 | assignment has no effect. | |
665 | ||
5076a392 FC |
666 | See L<http://rt.perl.org/rt3/Public/Bug/Display.html?id=77810> for even |
667 | more detail. | |
668 | ||
61752d82 | 669 | =head3 Clearing stashes |
5076a392 FC |
670 | |
671 | Stash list assignment C<%foo:: = ()> used to make the stash anonymous | |
672 | temporarily while it was being emptied. Consequently, any of its | |
673 | subroutines referenced elsewhere would become anonymous (showing up as | |
674 | "(unknown)" in C<caller>). Now they retain their package names, such that | |
675 | C<caller> will return the original sub name if there is still a reference | |
676 | to its typeglob, or "foo::__ANON__" otherwise [perl #79208]. | |
677 | ||
61752d82 FC |
678 | =head3 Magic variables outside the main package |
679 | ||
680 | In previous versions of Perl, magic variables like C<$!>, C<%SIG>, etc. would | |
681 | 'leak' into other packages. So C<%foo::SIG> could be used to access signals, | |
682 | C<${"foo::!"}> (with strict mode off) to access C's C<errno>, etc. | |
683 | ||
684 | This was a bug, or an 'unintentional' feature, which caused various ill effects, | |
685 | such as signal handlers being wiped when modules were loaded, etc. | |
686 | ||
687 | This has been fixed (or the feature has been removed, depending on how you see | |
688 | it). | |
689 | ||
690 | =head2 Changes to Syntax or to Perl Operators | |
691 | ||
692 | =head3 C<given> return values | |
693 | ||
694 | C<given> blocks now return the last evaluated | |
695 | expression, or an empty list if the block was exited by C<break>. Thus you | |
696 | can now write: | |
697 | ||
698 | my $type = do { | |
699 | given ($num) { | |
700 | break when undef; | |
701 | 'integer' when /^[+-]?[0-9]+$/; | |
702 | 'float' when /^[+-]?[0-9]+(?:\.[0-9]+)?$/; | |
703 | 'unknown'; | |
704 | } | |
705 | }; | |
706 | ||
707 | See L<perlsyn/Return value> for details. | |
708 | ||
709 | =head3 Change in the parsing of certain prototypes | |
710 | ||
711 | Functions declared with the following prototypes now behave correctly as unary | |
712 | functions: | |
713 | ||
714 | * | |
715 | \$ \% \@ \* \& | |
716 | \[...] | |
717 | ;$ ;* | |
718 | ;\$ ;\% etc. | |
719 | ;\[...] | |
720 | ||
721 | Due to this bug fix [perl #75904], functions | |
722 | using the C<(*)>, C<(;$)> and C<(;*)> prototypes | |
723 | are parsed with higher precedence than before. So in the following example: | |
724 | ||
725 | sub foo($); | |
726 | foo $a < $b; | |
727 | ||
728 | the second line is now parsed correctly as C<< foo($a) < $b >>, rather than | |
729 | C<< foo($a < $b) >>. This happens when one of these operators is used in | |
730 | an unparenthesised argument: | |
731 | ||
732 | < > <= >= lt gt le ge | |
733 | == != <=> eq ne cmp ~~ | |
734 | & | |
735 | | ^ | |
736 | && | |
737 | || // | |
738 | .. ... | |
739 | ?: | |
740 | = += -= *= etc. | |
741 | ||
742 | =head3 Smart-matching against array slices | |
743 | ||
744 | Previously, the following code resulted in a successful match: | |
745 | ||
746 | my @a = qw(a y0 z); | |
747 | my @b = qw(a x0 z); | |
748 | @a[0 .. $#b] ~~ @b; | |
749 | ||
750 | This odd behaviour has now been fixed [perl #77468]. | |
751 | ||
752 | =head3 Negation treats strings differently from before | |
753 | ||
754 | The unary negation operator C<-> now treats strings that look like numbers | |
755 | as numbers [perl #57706]. | |
756 | ||
757 | =head3 Negative zero | |
758 | ||
759 | Negative zero (-0.0), when converted to a string, now becomes "0" on all | |
760 | platforms. It used to become "-0" on some, but "0" on others. | |
761 | ||
762 | If you still need to determine whether a zero is negative, use | |
763 | C<sprintf("%g", $zero) =~ /^-/> or the L<Data::Float> module on CPAN. | |
764 | ||
765 | =head3 C<:=> is now a syntax error | |
5076a392 FC |
766 | |
767 | Previously C<my $pi := 4;> was exactly equivalent to C<my $pi : = 4;>, | |
768 | with the C<:> being treated as the start of an attribute list, ending before | |
769 | the C<=>. The use of C<:=> to mean C<: => was deprecated in 5.12.0, and is now | |
770 | a syntax error. This will allow the future use of C<:=> as a new token. | |
771 | ||
772 | We find no Perl 5 code on CPAN using this construction, outside the core's | |
773 | tests for it, so we believe that this change will have very little impact on | |
774 | real-world codebases. | |
775 | ||
776 | If it is absolutely necessary to have empty attribute lists (for example, | |
777 | because of a code generator) then avoid the error by adding a space before | |
778 | the C<=>. | |
779 | ||
61752d82 | 780 | =head2 Threads and Processes |
5076a392 | 781 | |
61752d82 | 782 | =head3 Directory handles not copied to threads |
5076a392 | 783 | |
61752d82 FC |
784 | On systems other than Windows that do not have |
785 | a C<fchdir> function, newly-created threads no | |
786 | longer inherit directory handles from their parent threads. Such programs | |
787 | would usually have crashed anyway [perl #75154]. | |
5076a392 | 788 | |
61752d82 | 789 | =head3 C<close> on shared pipes |
5076a392 | 790 | |
61752d82 FC |
791 | The C<close> function no longer waits for the child process to exit if the |
792 | underlying file descriptor is still in use by another thread, to avoid | |
793 | deadlocks. It returns true in such cases. | |
5076a392 | 794 | |
144b6ea2 FC |
795 | =head3 fork() emulation will not wait for signalled children |
796 | ||
797 | On Windows parent processes would not terminate until all forked | |
798 | childred had terminated first. However, C<kill('KILL', ...)> is | |
799 | inherently unstable on pseudo-processes, and C<kill('TERM', ...)> | |
800 | might not get delivered if the child if blocked in a system call. | |
801 | ||
802 | To avoid the deadlock and still provide a safe mechanism to terminate | |
803 | the hosting process, Perl will now no longer wait for children that | |
804 | have been sent a SIGTERM signal. It is up to the parent process to | |
805 | waitpid() for these children if child clean-up processing must be | |
806 | allowed to finish. However, it is also the responsibility of the | |
807 | parent then to avoid the deadlock by making sure the child process | |
808 | can't be blocked on I/O either. | |
809 | ||
810 | See L<perlfork> for more information about the fork() emulation on | |
811 | Windows. | |
812 | ||
61752d82 | 813 | =head2 Configuration |
5076a392 | 814 | |
61752d82 | 815 | =head3 Naming fixes in Policy_sh.SH may invalidate Policy.sh |
5076a392 | 816 | |
61752d82 FC |
817 | Several long-standing typos and naming confusions in Policy_sh.SH have |
818 | been fixed, standardizing on the variable names used in config.sh. | |
5076a392 | 819 | |
61752d82 FC |
820 | This will change the behavior of Policy.sh if you happen to have been |
821 | accidentally relying on the Policy.sh incorrect behavior. | |
f318e2e6 | 822 | |
5a1f7719 FC |
823 | =head3 Perl source code is read in text mode on Windows |
824 | ||
825 | Perl scripts used to be read in binary mode on Windows for the benefit | |
826 | of the ByteLoader module (which is no longer part of core Perl). This | |
827 | had the side effect of breaking various operations on the DATA filehandle, | |
828 | including seek()/tell(), and even simply reading from DATA after file handles | |
829 | have been flushed by a call to system(), backticks, fork() etc. | |
830 | ||
831 | The default build options for Windows have been changed to read Perl source | |
832 | code on Windows in text mode now. Hopefully ByteLoader will be updated on | |
833 | CPAN to automatically handle this situation [perl #28106]. | |
834 | ||
5076a392 FC |
835 | =head1 Deprecations |
836 | ||
1e463951 FC |
837 | See also L</Deprecated C APIs>. |
838 | ||
5076a392 FC |
839 | =head2 Omitting a space between a regular expression and subsequent word |
840 | ||
18139a1b FC |
841 | Omitting a space between a regular expression operator or |
842 | its modifiers and the following word is deprecated. For | |
843 | example, C<< m/foo/sand $bar >> will still be parsed | |
844 | as C<< m/foo/s and $bar >> but will issue a warning. | |
5076a392 FC |
845 | |
846 | =head2 Deprecation warning added for deprecated-in-core .pl libs | |
847 | ||
848 | This is a mandatory warning, not obeying -X or lexical warning bits. | |
849 | The warning is modelled on that supplied by deprecate.pm for | |
850 | deprecated-in-core .pm libraries. It points to the specific CPAN | |
823d0e46 FC |
851 | distribution that contains the .pl libraries. The CPAN version, of |
852 | course, does not generate the warning. | |
5076a392 FC |
853 | |
854 | =head2 List assignment to C<$[> | |
855 | ||
823d0e46 FC |
856 | Assignment to C<$[> was deprecated and started to give warnings in |
857 | Perl version 5.12.0. This version of perl also starts to emit a warning when | |
858 | assigning to C<$[> in list context. This fixes an oversight in 5.12.0. | |
5076a392 FC |
859 | |
860 | =head2 Use of qw(...) as parentheses | |
861 | ||
862 | Historically the parser fooled itself into thinking that C<qw(...)> literals | |
863 | were always enclosed in parentheses, and as a result you could sometimes omit | |
864 | parentheses around them: | |
865 | ||
866 | for $x qw(a b c) { ... } | |
867 | ||
868 | The parser no longer lies to itself in this way. Wrap the list literal in | |
823d0e46 | 869 | parentheses, like this: |
5076a392 FC |
870 | |
871 | for $x (qw(a b c)) { ... } | |
872 | ||
873 | =head2 C<\N{BELL}> is deprecated | |
874 | ||
875 | This is because Unicode is using that name for a different character. | |
876 | See L</Unicode Version 6.0 is now supported (mostly)> for more | |
877 | explanation. | |
878 | ||
879 | =head2 C<?PATTERN?> is deprecated | |
880 | ||
881 | C<?PATTERN?> (without the initial m) has been deprecated and now produces | |
882 | a warning. This is to allow future use of C<?> in new operators. | |
883 | The match-once functionality is still available in the form of C<m?PATTERN?>. | |
884 | ||
5076a392 FC |
885 | =head2 Tie functions on scalars holding typeglobs |
886 | ||
887 | Calling a tie function (C<tie>, C<tied>, C<untie>) with a scalar argument | |
888 | acts on a file handle if the scalar happens to hold a typeglob. | |
889 | ||
890 | This is a long-standing bug that will be removed in Perl 5.16, as | |
891 | there is currently no way to tie the scalar itself when it holds | |
892 | a typeglob, and no way to untie a scalar that has had a typeglob | |
893 | assigned to it. | |
894 | ||
823d0e46 | 895 | Now there is a deprecation warning whenever a tie |
5076a392 FC |
896 | function is used on a handle without an explicit C<*>. |
897 | ||
18139a1b | 898 | =head2 User-defined case-mapping |
5076a392 | 899 | |
18139a1b FC |
900 | This feature is being deprecated due to its many issues, as documented in |
901 | L<perlunicode/User-Defined Case Mappings (for serious hackers only)>. | |
902 | It is planned to remove this feature in Perl 5.16. A CPAN module | |
903 | providing improved functionality is being prepared for release by the | |
904 | time 5.14 is. | |
905 | ||
823d0e46 FC |
906 | XXX What module is that? |
907 | ||
18139a1b | 908 | =head2 Deprecated modules |
5076a392 FC |
909 | |
910 | The following modules will be removed from the core distribution in a | |
911 | future release, and should be installed from CPAN instead. Distributions | |
912 | on CPAN which require these should add them to their prerequisites. The | |
823d0e46 | 913 | core versions of these modules will issue a deprecation warning. |
5076a392 FC |
914 | |
915 | If you ship a packaged version of Perl, either alone or as part of a | |
916 | larger system, then you should carefully consider the repercussions of | |
823d0e46 | 917 | core module deprecations. You may want to consider shipping your default |
5076a392 | 918 | build of Perl with packages for some or all deprecated modules which |
823d0e46 | 919 | install into C<vendor> or C<site> perl library directories. This will |
5076a392 FC |
920 | inhibit the deprecation warnings. |
921 | ||
922 | Alternatively, you may want to consider patching F<lib/deprecate.pm> | |
923 | to provide deprecation warnings specific to your packaging system | |
924 | or distribution of Perl, consistent with how your packaging system | |
925 | or distribution manages a staged transition from a release where the | |
926 | installation of a single package provides the given functionality, to | |
927 | a later release where the system administrator needs to know to install | |
928 | multiple packages to get that same functionality. | |
929 | ||
930 | You can silence these deprecation warnings by installing the modules | |
931 | in question from CPAN. To install the latest version of all of them, | |
932 | just install C<Task::Deprecations::5_14>. | |
933 | ||
934 | =over | |
935 | ||
936 | =item L<Devel::DProf> | |
937 | ||
938 | We strongly recommend that you install and used L<Devel::NYTProf> in | |
939 | preference, as it offers significantly improved profiling and reporting. | |
940 | ||
941 | =back | |
942 | ||
5076a392 FC |
943 | =head1 Performance Enhancements |
944 | ||
54c7bb16 | 945 | =head2 "Safe signals" optimisation |
df91d470 FC |
946 | |
947 | Signal dispatch has been moved from the runloop into control ops. This | |
948 | should give a few percent speed increase, and eliminates almost all of | |
949 | the speed penalty caused by the introduction of "safe signals" in | |
950 | 5.8.0. Signals should still be dispatched within the same statement as | |
951 | they were previously - if this is not the case, or it is possible to | |
952 | create uninterruptible loops, this is a bug, and reports are encouraged | |
953 | of how to recreate such issues. | |
954 | ||
54c7bb16 | 955 | =head2 Optimisation of shift; and pop; calls without arguments |
df91d470 | 956 | |
111b6aa7 FC |
957 | Two fewer OPs are used for shift and pop calls with no argument (with |
958 | implicit C<@_>). This change makes C<shift;> 5% faster than C<shift @_;> | |
959 | on non-threaded perls and 25% faster on threaded. | |
df91d470 | 960 | |
fa232254 | 961 | =head2 Optimisation of regexp engine string comparison work |
df91d470 | 962 | |
fa232254 FC |
963 | The foldEQ_utf8 API function for case-insensitive comparison of strings (which |
964 | is used heavily by the regexp engine) was substantially refactored and | |
965 | optimised - and its documentation much improved as a free bonus gift. | |
df91d470 | 966 | |
fa232254 | 967 | =head2 Regular expression compilation speed-up |
df91d470 | 968 | |
fa232254 FC |
969 | Compiling regular expressions has been made faster for the case where upgrading |
970 | the regex to utf8 is necessary but that isn't known when the compilation begins. | |
df91d470 | 971 | |
fa232254 | 972 | =head2 String appending is 100 times faster |
df91d470 | 973 | |
fa232254 FC |
974 | When doing a lot of string appending, perl could end up allocating a lot more |
975 | memory than needed in a very inefficient way, if perl was configured to use the | |
976 | system's C<malloc> implementation instead of its own. | |
977 | ||
978 | C<sv_grow>, which is what's being used to allocate more memory if necessary | |
979 | when appending to a string, has now been taught how to round up the memory | |
980 | it requests to a certain geometric progression, making it much faster on | |
981 | certain platforms and configurations. On Win32, it's now about 100 times | |
982 | faster. | |
983 | ||
984 | =head2 Eliminate C<PL_*> accessor functions under ithreads | |
985 | ||
986 | When C<MULTIPLICITY> was first developed, and interpreter state moved into | |
987 | an interpreter struct, thread and interpreter local C<PL_*> variables were | |
988 | defined as macros that called accessor functions, returning the address of | |
989 | the value, outside of the perl core. The intent was to allow members | |
990 | within the interpreter struct to change size without breaking binary | |
991 | compatibility, so that bug fixes could be merged to a maintenance branch | |
992 | that necessitated such a size change. | |
993 | ||
994 | However, some non-core code defines C<PERL_CORE>, sometimes intentionally | |
995 | to bypass this mechanism for speed reasons, sometimes for other reasons but | |
996 | with the inadvertent side effect of bypassing this mechanism. As some of | |
997 | this code is widespread in production use, the result is that the core | |
998 | I<can't> change the size of members of the interpreter struct, as it will | |
999 | break such modules compiled against a previous release on that maintenance | |
1000 | branch. The upshot is that this mechanism is redundant, and well-behaved | |
1001 | code is penalised by it. Hence it can and should be removed (and has | |
1002 | been). | |
1003 | ||
1004 | =head2 Freeing weak references | |
1005 | ||
1006 | When an object has many weak references to it, freeing that object | |
1007 | can under some some circumstances take O(N^2) time to free (where N is the | |
1008 | number of references). The number of circumstances has been reduced | |
1009 | [perl #75254] | |
1010 | ||
1011 | =head2 Lexical array and hash assignments | |
1012 | ||
1013 | An earlier optimisation to speed up C<my @array = ...> and | |
1014 | C<my %hash = ...> assignments caused a bug and was disabled in Perl 5.12.0. | |
1015 | ||
1016 | Now we have found another way to speed up these assignments [perl #82110]. | |
df91d470 | 1017 | |
111b6aa7 | 1018 | =head2 C<@_> uses less memory |
df91d470 | 1019 | |
111b6aa7 FC |
1020 | Previously, C<@_> was allocated for every subroutine at compile time with |
1021 | enough space for four entries. Now this allocation is done on demand when | |
1022 | the subroutine is called [perl #72416]. | |
5076a392 | 1023 | |
5076a392 FC |
1024 | =head2 Size optimisations to SV and HV structures |
1025 | ||
1026 | xhv_fill has been eliminated from struct xpvhv, saving 1 IV per hash and | |
111b6aa7 | 1027 | on some systems will cause struct xpvhv to become cache-aligned. To avoid |
5076a392 FC |
1028 | this memory saving causing a slowdown elsewhere, boolean use of HvFILL |
1029 | now calls HvTOTALKEYS instead (which is equivalent) - so while the fill | |
111b6aa7 FC |
1030 | data when actually required are now calculated on demand, the cases when |
1031 | this needs to be done should be few and far between. | |
5076a392 | 1032 | |
111b6aa7 FC |
1033 | The order of structure elements in SV bodies has changed. Effectively, |
1034 | the NV slot has swapped location with STASH and MAGIC. As all access to | |
1035 | SV members is via macros, this should be completely transparent. This | |
5076a392 FC |
1036 | change allows the space saving for PVHVs documented above, and may reduce |
1037 | the memory allocation needed for PVIVs on some architectures. | |
1038 | ||
fa232254 FC |
1039 | C<XPV>, C<XPVIV>, and C<XPVNV> now only allocate the parts of the C<SV> body |
1040 | they actually use, saving some space. | |
5076a392 | 1041 | |
fa232254 FC |
1042 | Scalars containing regular expressions now only allocate the part of the C<SV> |
1043 | body they actually use, saving some space. | |
5076a392 FC |
1044 | |
1045 | =head2 Memory consumption improvements to Exporter | |
1046 | ||
1047 | The @EXPORT_FAIL AV is no longer created unless required, hence neither is | |
111b6aa7 FC |
1048 | the typeglob backing it. This saves about 200 bytes for every package that |
1049 | uses Exporter but doesn't use this functionality. | |
5076a392 | 1050 | |
111b6aa7 | 1051 | =head2 Memory savings for weak references |
5076a392 FC |
1052 | |
1053 | For weak references, the common case of just a single weak reference per | |
1054 | referent has been optimised to reduce the storage required. In this case it | |
111b6aa7 | 1055 | saves the equivalent of one small Perl array per referent. |
5076a392 | 1056 | |
111b6aa7 | 1057 | =head2 C<%+> and C<%-> use less memory |
5076a392 FC |
1058 | |
1059 | The bulk of the C<Tie::Hash::NamedCapture> module used to be in the perl | |
111b6aa7 | 1060 | core. It has now been moved to an XS module, to reduce the overhead for |
5076a392 FC |
1061 | programs that do not use C<%+> or C<%->. |
1062 | ||
fa232254 | 1063 | =head2 Multiple small improvements to threads |
5076a392 | 1064 | |
fa232254 FC |
1065 | The internal structures of threading now make fewer API calls and fewer |
1066 | allocations, resulting in noticeably smaller object code. Additionally, | |
1067 | many thread context checks have been deferred so that they're only done | |
1068 | when required (although this is only possible for non-debugging builds). | |
5076a392 | 1069 | |
fa232254 | 1070 | =head2 Adjacent pairs of nextstate opcodes are now optimized away |
5076a392 | 1071 | |
fa232254 | 1072 | Previously, in code such as |
5076a392 | 1073 | |
fa232254 | 1074 | use constant DEBUG => 0; |
5076a392 | 1075 | |
fa232254 FC |
1076 | sub GAK { |
1077 | warn if DEBUG; | |
1078 | print "stuff\n"; | |
1079 | } | |
5076a392 | 1080 | |
fa232254 FC |
1081 | the ops for C<warn if DEBUG;> would be folded to a C<null> op (C<ex-const>), but |
1082 | the C<nextstate> op would remain, resulting in a runtime op dispatch of | |
1083 | C<nextstate>, C<nextstate>, .... | |
5076a392 | 1084 | |
fa232254 FC |
1085 | The execution of a sequence of C<nextstate> ops is indistinguishable from just |
1086 | the last C<nextstate> op so the peephole optimizer now eliminates the first of | |
1087 | a pair of C<nextstate> ops, except where the first carries a label, since labels | |
1088 | must not be eliminated by the optimizer and label usage isn't conclusively known | |
1089 | at compile time. | |
5076a392 FC |
1090 | |
1091 | =head1 Modules and Pragmata | |
1092 | ||
1093 | =head2 New Modules and Pragmata | |
1094 | ||
1095 | =over 4 | |
1096 | ||
1097 | =item * | |
1098 | ||
1099 | C<CPAN::Meta::YAML> 0.003 has been added as a dual-life module. It supports a | |
1100 | subset of YAML sufficient for reading and writing META.yml and MYMETA.yml files | |
1101 | included with CPAN distributions or generated by the module installation | |
1102 | toolchain. It should not be used for any other general YAML parsing or | |
1103 | generation task. | |
1104 | ||
1105 | =item * | |
1106 | ||
1107 | C<CPAN::Meta> version 2.110440 has been added as a dual-life module. It | |
1108 | provides a standard library to read, interpret and write CPAN distribution | |
1109 | metadata files (e.g. META.json and META.yml) which describes a | |
1110 | distribution, its contents, and the requirements for building it and | |
1111 | installing it. The latest CPAN distribution metadata specification is | |
1112 | included as C<CPAN::Meta::Spec> and notes on changes in the specification | |
1113 | over time are given in C<CPAN::Meta::History>. | |
1114 | ||
1115 | =item * | |
1116 | ||
1117 | C<HTTP::Tiny> 0.010 has been added as a dual-life module. It is a very | |
1118 | small, simple HTTP/1.1 client designed for simple GET requests and file | |
1119 | mirroring. It has has been added to enable CPAN.pm and CPANPLUS to | |
1120 | "bootstrap" HTTP access to CPAN using pure Perl without relying on external | |
1121 | binaries like F<curl> or F<wget>. | |
1122 | ||
1123 | =item * | |
1124 | ||
1125 | C<JSON::PP> 2.27105 has been added as a dual-life module, for the sake of | |
1126 | reading F<META.json> files in CPAN distributions. | |
1127 | ||
1128 | =item * | |
1129 | ||
1130 | C<Module::Metadata> 1.000003 has been added as a dual-life module. It gathers | |
1131 | package and POD information from Perl module files. It is a standalone module | |
1132 | based on Module::Build::ModuleInfo for use by other module installation | |
1133 | toolchain components. Module::Build::ModuleInfo has been deprecated in | |
1134 | favor of this module instead. | |
1135 | ||
1136 | =item * | |
1137 | ||
1138 | C<Perl::OSType> 1.002 has been added as a dual-life module. It maps Perl | |
1139 | operating system names (e.g. 'dragonfly' or 'MSWin32') to more generic types | |
1140 | with standardized names (e.g. "Unix" or "Windows"). It has been refactored | |
1141 | out of Module::Build and ExtUtils::CBuilder and consolidates such mappings into | |
1142 | a single location for easier maintenance. | |
1143 | ||
1144 | =item * | |
1145 | ||
1146 | The following modules were added by the C<Unicode::Collate> | |
1147 | upgrade. See below for details. | |
1148 | ||
1149 | C<Unicode::Collate::CJK::Big5> | |
1150 | ||
1151 | C<Unicode::Collate::CJK::GB2312> | |
1152 | ||
1153 | C<Unicode::Collate::CJK::JISX0208> | |
1154 | ||
1155 | C<Unicode::Collate::CJK::Korean> | |
1156 | ||
1157 | C<Unicode::Collate::CJK::Pinyin> | |
1158 | ||
1159 | C<Unicode::Collate::CJK::Stroke> | |
1160 | ||
1161 | =item * | |
1162 | ||
1163 | C<Version::Requirements> version 0.101020 has been added as a dual-life | |
1164 | module. It provides a standard library to model and manipulates module | |
1165 | prerequisites and version constraints as defined in the L<CPAN::Meta::Spec>. | |
1166 | ||
1167 | =back | |
1168 | ||
d622514a | 1169 | =head2 Selected Module and Pragma Updates |
5076a392 FC |
1170 | |
1171 | =over 4 | |
1172 | ||
1173 | =item * | |
1174 | ||
5076a392 FC |
1175 | C<Archive::Extract> has been upgraded from version 0.38 to 0.48. |
1176 | ||
1177 | Updates since 0.38 include: a safe print method that guards | |
1178 | Archive::Extract from changes to $\; a fix to the tests when run in core | |
1984204c FC |
1179 | perl; support for TZ files; a modification for the lzma |
1180 | logic to favour IO::Uncompress::Unlzma; and a fix | |
1181 | for an issue with NetBSD-current and its new unzip | |
5076a392 FC |
1182 | executable. |
1183 | ||
1184 | =item * | |
1185 | ||
1186 | C<Archive::Tar> has been upgraded from version 1.54 to 1.76. | |
1187 | ||
1984204c FC |
1188 | Important changes since 1.54 include the following: |
1189 | ||
1190 | =over | |
1191 | ||
1192 | =item * | |
1193 | ||
1194 | Compatibility with busybox implementations of tar | |
1195 | ||
1196 | =item * | |
1197 | ||
1198 | A fix so that C<write()> and C<create_archive()> | |
1199 | close only handles they opened | |
1200 | ||
1201 | =item * | |
1202 | ||
1203 | A bug was fixed regarding the exit code of extract_archive. | |
5076a392 | 1204 | |
1984204c FC |
1205 | =item * |
1206 | ||
1207 | C<ptar> has a new option to allow safe | |
5076a392 FC |
1208 | creation of tarballs without world-writable files on Windows, allowing those |
1209 | archives to be uploaded to CPAN. | |
1210 | ||
1984204c FC |
1211 | =item * |
1212 | ||
1213 | A new ptargrep utility for using regular expressions against | |
5076a392 FC |
1214 | the contents of files in a tar archive. |
1215 | ||
1984204c FC |
1216 | =item * |
1217 | ||
1218 | Pax extended headers are now skipped. | |
1219 | ||
1220 | =back | |
5076a392 FC |
1221 | |
1222 | =item * | |
1223 | ||
5076a392 FC |
1224 | C<B> has been upgraded from version 1.23 to 1.27. |
1225 | ||
1226 | It no longer crashes when taking apart a C<y///> containing characters | |
1227 | outside the octet range or compiled in a C<use utf8> scope. | |
1228 | ||
1229 | The size of the shared object has been reduced by about 40%, with no | |
1230 | reduction in functionality. | |
1231 | ||
1232 | =item * | |
1233 | ||
1234 | C<B::Concise> has been upgraded from version 0.78 to 0.82. | |
1235 | ||
1236 | B::Concise marks rv2sv, rv2av and rv2hv ops with the new OPpDEREF flag | |
1237 | as "DREFed". | |
1238 | ||
1239 | It no longer produces mangled output with the C<-tree> option | |
1240 | [perl #80632]. | |
1241 | ||
1242 | =item * | |
1243 | ||
5076a392 FC |
1244 | C<B::Deparse> has been upgraded from version 0.96 to 1.02. |
1245 | ||
1984204c | 1246 | The deparsing of a nextstate op has changed when it has both a |
5076a392 FC |
1247 | change of package (relative to the previous nextstate), or a change of |
1248 | C<%^H> or other state, and a label. Previously the label was emitted | |
1984204c | 1249 | first, but now the label is emitted last. |
5076a392 | 1250 | |
1984204c | 1251 | The C<no 5.13.2> or similar form is now correctly handled by B::Deparse. |
5076a392 FC |
1252 | |
1253 | B::Deparse now properly handles the code that applies a conditional | |
1254 | pattern match against implicit C<$_> as it was fixed in [perl #20444]. | |
1255 | ||
1984204c FC |
1256 | Deparsing of C<our> followed by a variable with funny characters |
1257 | (as permitted under the C<utf8> pragma) has also been fixed [perl #33752]. | |
5076a392 FC |
1258 | |
1259 | =item * | |
1260 | ||
5076a392 FC |
1261 | C<Carp> has been upgraded from version 1.18 to 1.19. |
1262 | ||
1263 | L<Carp> now detects incomplete L<caller()|perlfunc/"caller EXPR"> overrides and | |
1264 | avoids using bogus C<@DB::args>. To provide backtraces, Carp relies on | |
1265 | particular behaviour of the caller built-in. Carp now detects if other code has | |
1266 | overridden this with an incomplete implementation, and modifies its backtrace | |
1267 | accordingly. Previously incomplete overrides would cause incorrect values in | |
1268 | backtraces (best case), or obscure fatal errors (worst case) | |
1269 | ||
1270 | This fixes certain cases of C<Bizarre copy of ARRAY> caused by modules | |
1271 | overriding C<caller()> incorrectly. | |
1272 | ||
1984204c | 1273 | It now also avoids using regular expressions that cause perl to |
5076a392 FC |
1274 | load its Unicode tables, in order to avoid the 'BEGIN not safe after |
1275 | errors' error that will ensue if there has been a syntax error | |
1276 | [perl #82854]. | |
1277 | ||
1278 | =item * | |
1279 | ||
1280 | C<CGI> has been upgraded from version 3.48 to 3.51. | |
1281 | ||
1282 | This provides the following security fixes: the MIME boundary in | |
1984204c FC |
1283 | multipart_init is now random and the handling of |
1284 | newlines embedded in header values has been improved. | |
5076a392 FC |
1285 | |
1286 | =item * | |
1287 | ||
5076a392 FC |
1288 | C<Compress::Raw::Bzip2> has been upgraded from version 2.024 to 2.033. |
1289 | ||
1984204c | 1290 | It has been updated to use bzip2 1.0.6. |
5076a392 FC |
1291 | |
1292 | =item * | |
1293 | ||
1984204c FC |
1294 | C<CPAN> has been upgraded from version 1.94_56 to 1.9600.- |
1295 | ||
1296 | Major highlights: | |
5076a392 FC |
1297 | |
1298 | =over 4 | |
1299 | ||
1984204c | 1300 | =item * much less configuration dialog hassle |
5076a392 | 1301 | |
1984204c | 1302 | =item * support for META/MYMETA.json |
5076a392 | 1303 | |
1984204c | 1304 | =item * support for local::lib |
5076a392 | 1305 | |
1984204c | 1306 | =item * support for HTTP::Tiny to reduce the dependency on ftp sites |
5076a392 | 1307 | |
1984204c | 1308 | =item * automatic mirror selection |
5076a392 | 1309 | |
1984204c | 1310 | =item * iron out all known bugs in configure_requires |
5076a392 | 1311 | |
1984204c FC |
1312 | =item * support for distributions compressed with bzip2 |
1313 | ||
1314 | =item * allow Foo/Bar.pm on the commandline to mean Foo::Bar | |
1315 | ||
1316 | =back | |
5076a392 FC |
1317 | |
1318 | =item * | |
1319 | ||
1320 | C<CPANPLUS> has been upgraded from version 0.90 to 0.9102. | |
1321 | ||
1984204c FC |
1322 | A dependency on Config was not recognised as a |
1323 | core module dependency. This has been fixed. | |
5076a392 | 1324 | |
1984204c | 1325 | CPANPLUS now includes support for META.json and MYMETA.json. |
5076a392 FC |
1326 | |
1327 | =item * | |
1328 | ||
5076a392 FC |
1329 | C<Data::Dumper> has been upgraded from version 2.125 to 2.130_02. |
1330 | ||
4ed2cea4 FC |
1331 | The indentation used to be off when C<$Data::Dumper::Terse> was set. This |
1332 | has been fixed [perl #73604]. | |
1333 | ||
1984204c FC |
1334 | This upgrade also fixes a crash when using custom sort functions that might |
1335 | cause the stack to change [perl #74170]. | |
5076a392 FC |
1336 | |
1337 | C<Dumpxs> no longer crashes with globs returned by C<*$io_ref> | |
1338 | [perl #72332]. | |
1339 | ||
1340 | =item * | |
1341 | ||
5076a392 FC |
1342 | C<Devel::DProf> has been upgraded from version 20080331.00 to 20110228.00. |
1343 | ||
1344 | Merely loading C<Devel::DProf> now no longer triggers profiling to start. | |
1345 | C<use Devel::DProf> and C<perl -d:DProf ...> still behave as before and start | |
1346 | the profiler. | |
1347 | ||
1348 | NOTE: C<Devel::DProf> is deprecated and will be removed from a future | |
1984204c | 1349 | version of Perl. We strongly recommend that you install and use |
5076a392 FC |
1350 | L<Devel::NYTProf> instead, as it offers significantly improved |
1351 | profiling and reporting. | |
1352 | ||
1353 | =item * | |
1354 | ||
5076a392 FC |
1355 | C<diagnostics> has been upgraded from version 1.19 to 1.22. |
1356 | ||
1357 | It now renders pod links slightly better, and has been taught to find | |
1358 | descriptions for messages that share their descriptions with other | |
1359 | messages. | |
1360 | ||
1361 | =item * | |
1362 | ||
1363 | C<Digest::MD5> has been upgraded from version 2.39 to 2.51. | |
1364 | ||
1365 | It is now safe to use this module in combination with threads. | |
1366 | ||
1367 | =item * | |
1368 | ||
1369 | C<Digest::SHA> has been upgraded from version 5.47 to 5.61. | |
1370 | ||
1371 | C<shasum> now more closely mimics C<sha1sum>/C<md5sum>. | |
1372 | ||
1373 | C<Addfile> accepts all POSIX filenames. | |
1374 | ||
1984204c FC |
1375 | New SHA-512/224 and SHA-512/256 transforms (ref. NIST Draft FIPS 180-4 |
1376 | [February 2011]) | |
5076a392 FC |
1377 | |
1378 | =item * | |
1379 | ||
5076a392 FC |
1380 | C<DynaLoader> has been upgraded from version 1.10 to 1.12. |
1381 | ||
1382 | It fixes a buffer overflow when passed a very long file name. | |
1383 | ||
1384 | It no longer inherits from AutoLoader; hence it no longer | |
1385 | produces weird error messages for unsuccessful method calls on classes that | |
1386 | inherit from DynaLoader [perl #84358]. | |
1387 | ||
1388 | =item * | |
1389 | ||
1390 | C<Encode> has been upgraded from version 2.39 to 2.42. | |
1391 | ||
1392 | Now, all 66 Unicode non-characters are treated the same way U+FFFF has | |
1984204c FC |
1393 | always been treated; in cases when it was disallowed, all 66 are |
1394 | disallowed; in those cases where it warned, all 66 warn. | |
5076a392 FC |
1395 | |
1396 | =item * | |
1397 | ||
5076a392 FC |
1398 | C<Errno> has been upgraded from version 1.11 to 1.13. |
1399 | ||
1400 | The implementation of C<Errno> has been refactored to use about 55% less memory. | |
5076a392 FC |
1401 | |
1402 | On some platforms with unusual header files, like Win32/gcc using mingw64 | |
1403 | headers, some constants which weren't actually error numbers have been exposed | |
1404 | by C<Errno>. This has been fixed [perl #77416]. | |
1405 | ||
1406 | =item * | |
1407 | ||
1408 | C<Exporter> has been upgraded from version 5.64_01 to 5.64_03. | |
1409 | ||
1410 | Exporter no longer overrides C<$SIG{__WARN__}> [perl #74472] | |
1411 | ||
1412 | =item * | |
1413 | ||
5076a392 FC |
1414 | C<ExtUtils::Constant> has been upgraded from 0.22 to 0.23. |
1415 | ||
1416 | The C<AUTOLOAD> helper code generated by C<ExtUtils::Constant::ProxySubs> | |
1417 | can now C<croak> for missing constants, or generate a complete C<AUTOLOAD> | |
1984204c FC |
1418 | subroutine in XS, allowing simplification of many modules that use it |
1419 | (C<Fcntl>, C<File::Glob>, C<GDBM_File>, C<I18N::Langinfo>, C<POSIX>, | |
1420 | C<Socket>). | |
5076a392 FC |
1421 | |
1422 | C<ExtUtils::Constant::ProxySubs> can now optionally push the names of all | |
1984204c | 1423 | constants onto the package's C<@EXPORT_OK>. |
5076a392 FC |
1424 | |
1425 | =item * | |
1426 | ||
5076a392 FC |
1427 | C<File::DosGlob> has been upgraded from version 1.01 to 1.03. |
1428 | ||
1429 | It allows patterns containing literal parentheses (they no longer need to | |
1984204c FC |
1430 | be escaped). On Windows, it no longer |
1431 | adds an extra F<./> to the file names | |
5076a392 FC |
1432 | returned when the pattern is a relative glob with a drive specification, |
1433 | like F<c:*.pl> [perl #71712]. | |
1434 | ||
1435 | =item * | |
1436 | ||
1437 | C<File::Fetch> has been upgraded from version 0.24 to 0.32. | |
1438 | ||
1439 | C<HTTP::Lite> is now supported for 'http' scheme. | |
1440 | ||
1441 | The C<fetch> utility is supported on FreeBSD, NetBSD and | |
1442 | Dragonfly BSD for the C<http> and C<ftp> schemes. | |
1443 | ||
1444 | =item * | |
1445 | ||
1446 | C<File::Find> has been upgraded from version 1.15 to 1.18. | |
1447 | ||
1984204c | 1448 | It improves handling of backslashes on Windows, so that paths like |
5076a392 FC |
1449 | F<c:\dir\/file> are no longer generated [perl #71710]. |
1450 | ||
1451 | =item * | |
1452 | ||
5076a392 FC |
1453 | C<File::stat> has been upgraded from 1.02 to 1.04. |
1454 | ||
1455 | The C<-x> and C<-X> file test operators now work correctly under the root | |
1456 | user. | |
1457 | ||
1458 | =item * | |
1459 | ||
5076a392 FC |
1460 | C<GDBM_File> has been upgraded from 1.10 to 1.13. |
1461 | ||
1462 | This fixes a memory leak when DBM filters are used. | |
1463 | ||
1464 | =item * | |
1465 | ||
1466 | C<Hash::Util> has been upgraded from 0.07 to 0.10. | |
1467 | ||
1984204c FC |
1468 | Hash::Util no longer emits spurious "uninitialized" warnings when |
1469 | recursively locking hashes that have undefined values [perl #74280]. | |
5076a392 FC |
1470 | |
1471 | =item * | |
1472 | ||
1473 | C<I18N::Langinfo> has been upgraded from version 0.03 to 0.07. | |
1474 | ||
1475 | C<langinfo()> now defaults to using C<$_> if there is no argument given, just | |
5334145a | 1476 | as the documentation has always claimed. |
5076a392 FC |
1477 | |
1478 | =item * | |
1479 | ||
1480 | C<IO::Select> has been upgraded from version 1.17 to 1.18. | |
1481 | ||
1482 | It now allows IO::Handle objects (and objects in derived classes) to be | |
1483 | removed from an IO::Select set even if the underlying file descriptor is | |
1484 | closed or invalid. | |
1485 | ||
1486 | =item * | |
1487 | ||
5076a392 FC |
1488 | C<IPC::Cmd> has been upgraded from version 0.54 to 0.68. |
1489 | ||
1984204c FC |
1490 | Resolves an issue with splitting Win32 command lines. An argument |
1491 | consisting of the single character "0" used to be omitted (CPAN RT #62961). | |
5076a392 FC |
1492 | |
1493 | =item * | |
1494 | ||
1495 | C<IPC::Open3> has been upgraded from 1.05 to 1.08. | |
1496 | ||
4ed2cea4 FC |
1497 | C<open3> now produces an error if the C<exec> call fails, allowing this |
1498 | condition to be distinguished from a child process that exited with a | |
1499 | non-zero status [perl #72016]. | |
1500 | ||
5076a392 FC |
1501 | The internal C<xclose> routine now knows how to handle file descriptors, as |
1502 | documented, so duplicating STDIN in a child process using its file | |
1503 | descriptor now works [perl #76474]. | |
1504 | ||
1505 | =item * | |
1506 | ||
5076a392 FC |
1507 | C<Locale::Maketext> has been upgraded from version 1.14 to 1.17. |
1508 | ||
1984204c | 1509 | Locale::Maketext now supports external caches. |
5076a392 | 1510 | |
1984204c FC |
1511 | This upgrade also fixes an infinite loop in |
1512 | C<Locale::Maketext::Guts::_compile()> when | |
5076a392 FC |
1513 | working with tainted values (CPAN RT #40727). |
1514 | ||
1984204c | 1515 | C<< ->maketext >> calls will now back up and restore C<$@> so that error |
5076a392 FC |
1516 | messages are not suppressed (CPAN RT #34182). |
1517 | ||
1518 | =item * | |
1519 | ||
5076a392 FC |
1520 | C<Math::BigInt> has been upgraded from version 1.89_01 to 1.994. |
1521 | ||
1522 | This fixes, among other things, incorrect results when computing binomial | |
1523 | coefficients [perl #77640]. | |
1524 | ||
1984204c | 1525 | It also prevents C<sqrt($int)> from crashing under C<use bigrat;> |
5076a392 FC |
1526 | [perl #73534]. |
1527 | ||
1528 | =item * | |
1529 | ||
5076a392 FC |
1530 | C<MIME::Base64> has been upgraded from 3.08 to 3.13. |
1531 | ||
1532 | Includes new functions to calculate the length of encoded and decoded | |
1533 | base64 strings. | |
1534 | ||
1535 | Now provides C<encode_base64url> and C<decode_base64url> functions to process | |
1536 | the base64 scheme for "URL applications". | |
1537 | ||
1538 | =item * | |
1539 | ||
1540 | C<Module::Build> has been upgraded from version 0.3603 to 0.3800. | |
1541 | ||
1542 | A notable change is the deprecation of several modules. | |
1543 | Module::Build::Version has been deprecated and Module::Build now relies | |
1544 | directly upon L<version>. Module::Build::ModuleInfo has been deprecated in | |
1545 | favor of a standalone copy of it called L<Module::Metadata>. | |
1546 | Module::Build::YAML has been deprecated in favor of L<CPAN::Meta::YAML>. | |
1547 | ||
1548 | Module::Build now also generates META.json and MYMETA.json files | |
1549 | in accordance with version 2 of the CPAN distribution metadata specification, | |
1550 | L<CPAN::Meta::Spec>. The older format META.yml and MYMETA.yml files are | |
1551 | still generated, as well. | |
1552 | ||
1553 | =item * | |
1554 | ||
1555 | C<Module::CoreList> has been upgraded from version 2.29 to XXX. | |
1556 | ||
1557 | Besides listing the updated core modules of this release, it also stops listing | |
1984204c | 1558 | the C<Filespec> module. That module never existed in core. The scripts |
5076a392 | 1559 | generating C<Module::CoreList> confused it with C<VMS::Filespec>, which actually |
1984204c | 1560 | is a core module as of perl 5.8.7. |
5076a392 FC |
1561 | |
1562 | =item * | |
1563 | ||
1984204c FC |
1564 | C<NDBM_File> and C<ODBM_File> have been upgraded from 1.08 to 1.11, and |
1565 | from 1.08 to 1.09, respectively. | |
5076a392 FC |
1566 | |
1567 | This fixes a memory leak when DBM filters are used. | |
1568 | ||
1569 | =item * | |
1570 | ||
5076a392 FC |
1571 | C<overload> has been upgraded from 1.11 to 1.12. |
1572 | ||
1573 | C<overload::Method> can now handle subroutines that are themselves blessed | |
1574 | into overloaded classes [perl #71998]. | |
1575 | ||
5076a392 FC |
1576 | The documentation has greatly improved. See L</Documentation> below. |
1577 | ||
1578 | =item * | |
1579 | ||
5076a392 FC |
1580 | C<Parse::CPAN::Meta> has been upgraded from version 1.40 to 1.4401. |
1581 | ||
1984204c | 1582 | The latest Parse::CPAN::Meta can now read YAML and JSON files using |
5076a392 FC |
1583 | L<CPAN::Meta::YAML> and L<JSON::PP>, which are now part of the Perl core. |
1584 | ||
1585 | =item * | |
1586 | ||
5076a392 FC |
1587 | C<PerlIO::scalar> has been upgraded from 0.07 to 0.11. |
1588 | ||
1589 | A C<read> after a C<seek> beyond the end of the string no longer thinks it | |
1590 | has data to read [perl #78716]. | |
1591 | ||
1592 | =item * | |
1593 | ||
5076a392 FC |
1594 | C<POSIX> has been upgraded from 1.19 to 1.23. |
1595 | ||
1596 | It now includes constants for POSIX signal constants. | |
1597 | ||
1598 | =item * | |
1599 | ||
1984204c | 1600 | C<re> has been upgraded from version 0.11 to 0.17. |
5076a392 FC |
1601 | |
1602 | New C<use re "/flags"> pragma | |
1603 | ||
5076a392 | 1604 | The C<regmust> function used to crash when called on a regular expression |
1984204c | 1605 | belonging to a pluggable engine. Now it croaks instead. |
5076a392 FC |
1606 | |
1607 | C<regmust> no longer leaks memory. | |
1608 | ||
1609 | =item * | |
1610 | ||
1611 | C<Safe> has been upgraded from version 2.25 to 2.29. | |
1612 | ||
1613 | This fixes a possible infinite loop when looking for coderefs. | |
1614 | ||
1615 | It adds C<&version::vxs::VCMP> to the default share. | |
1616 | ||
1617 | =item * | |
1618 | ||
5076a392 FC |
1619 | C<SelfLoader> has been upgraded from 1.17 to 1.18. |
1620 | ||
1621 | It now works in taint mode [perl #72062]. | |
1622 | ||
1623 | =item * | |
1624 | ||
1625 | C<sigtrap> has been upgraded from version 1.04 to 1.05. | |
1626 | ||
1627 | It no longer tries to modify read-only arguments when generating a | |
1628 | backtrace [perl #72340]. | |
1629 | ||
1630 | =item * | |
1631 | ||
1984204c | 1632 | C<Socket> has been upgraded from version 1.87 to 1.94. |
5076a392 | 1633 | |
1984204c | 1634 | See L</IPv6 support>, above. |
5076a392 FC |
1635 | |
1636 | =item * | |
1637 | ||
1638 | C<Storable> has been upgraded from version 2.22 to 2.27. | |
1639 | ||
1640 | Includes performance improvement for overloaded classes. | |
1641 | ||
5076a392 | 1642 | This adds support for serialising code references that contain UTF-8 strings |
1984204c FC |
1643 | correctly. The Storable minor version |
1644 | number changed as a result, meaning that | |
5076a392 FC |
1645 | Storable users who set C<$Storable::accept_future_minor> to a C<FALSE> value |
1646 | will see errors (see L<Storable/FORWARD COMPATIBILITY> for more details). | |
1647 | ||
1648 | Freezing no longer gets confused if the Perl stack gets reallocated | |
1649 | during freezing [perl #80074]. | |
1650 | ||
1651 | =item * | |
1652 | ||
5076a392 FC |
1653 | C<Test::Simple> has been upgraded from version 0.94 to 0.98. |
1654 | ||
1655 | Among many other things, subtests without a C<plan> or C<no_plan> now have an | |
1656 | implicit C<done_testing()> added to them. | |
1657 | ||
1658 | =item * | |
1659 | ||
5076a392 FC |
1660 | C<Thread::Semaphore> has been upgraded from version 2.09 to 2.12. |
1661 | ||
1984204c FC |
1662 | It provides two new methods that give more control over the decrementing of |
1663 | semaphores: C<down_nb> and C<down_force>. | |
5076a392 FC |
1664 | |
1665 | =item * | |
1666 | ||
5076a392 FC |
1667 | C<Tie::Hash> has been upgraded from version 1.03 to 1.04. |
1668 | ||
1669 | Calling C<< Tie::Hash-E<gt>TIEHASH() >> used to loop forever. Now it C<croak>s. | |
1670 | ||
1671 | =item * | |
1672 | ||
5076a392 FC |
1673 | C<Unicode::Collate> has been upgraded from version 0.52_01 to 0.73. |
1674 | ||
1984204c | 1675 | Unicode::Collate has been updated to use Unicode 6.0.0. |
5076a392 | 1676 | |
1984204c FC |
1677 | Unicode::Collate::Locale now supports a plethora of new locales: ar, be, |
1678 | bg, de__phonebook, hu, hy, kk, mk, nso, om, tn, vi, hr, ig, ja, ko, ru, sq, | |
1679 | se, sr, to, uk, zh, zh__big5han, zh__gb2312han, zh__pinyin and zh__stroke. | |
5076a392 FC |
1680 | |
1681 | The following modules have been added: | |
1682 | ||
1683 | C<Unicode::Collate::CJK::Big5> for C<zh__big5han> which makes | |
1684 | tailoring of CJK Unified Ideographs in the order of CLDR's big5han ordering. | |
1685 | ||
1686 | C<Unicode::Collate::CJK::GB2312> for C<zh__gb2312han> which makes | |
1687 | tailoring of CJK Unified Ideographs in the order of CLDR's gb2312han ordering. | |
1688 | ||
1689 | C<Unicode::Collate::CJK::JISX0208> which makes tailoring of 6355 kanji | |
1690 | (CJK Unified Ideographs) in the JIS X 0208 order. | |
1691 | ||
1692 | C<Unicode::Collate::CJK::Korean> which makes tailoring of CJK Unified Ideographs | |
1693 | in the order of CLDR's Korean ordering. | |
1694 | ||
1695 | C<Unicode::Collate::CJK::Pinyin> for C<zh__pinyin> which makes | |
1696 | tailoring of CJK Unified Ideographs in the order of CLDR's pinyin ordering. | |
1697 | ||
1698 | C<Unicode::Collate::CJK::Stroke> for C<zh__stroke> which makes | |
1699 | tailoring of CJK Unified Ideographs in the order of CLDR's stroke ordering. | |
1700 | ||
5076a392 FC |
1701 | This also sees the switch from using the pure-perl version of this |
1702 | module to the XS version. | |
1703 | ||
1704 | =item * | |
1705 | ||
5076a392 FC |
1706 | C<Unicode::UCD> has been upgraded from version 0.27 to 0.32. |
1707 | ||
1984204c FC |
1708 | A new function, C<Unicode::UCD::num()>, has been added. This function |
1709 | returns the numeric value of the string passed it or C<undef> if the string | |
1710 | in its entirety has no "safe" numeric value. (For more detail, and for the | |
1711 | definition of "safe", see L<Unicode::UCD/num>.) | |
5076a392 | 1712 | |
1984204c | 1713 | This upgrade also includes a number of bug fixes: |
5076a392 FC |
1714 | |
1715 | =over 4 | |
1716 | ||
1717 | =item charinfo() | |
1718 | ||
1719 | =over 4 | |
1720 | ||
1721 | =item * | |
1722 | ||
1723 | It is now updated to Unicode Version 6 with Corrigendum #8, except, | |
1724 | as with Perl 5.14, the code point at U+1F514 has no name. | |
1725 | ||
1726 | =item * | |
1727 | ||
1728 | The Hangul syllable code points have the correct names, and their | |
1729 | decompositions are always output without requiring L<Lingua::KO::Hangul::Util> | |
1730 | to be installed. | |
1731 | ||
1732 | =item * | |
1733 | ||
54c7bb16 FC |
1734 | The CJK (Chinese-Japanese-Korean) code points U+2A700 to U+2B734 |
1735 | and U+2B740 to U+2B81D are now properly handled. | |
5076a392 FC |
1736 | |
1737 | =item * | |
1738 | ||
1739 | The numeric values are now output for those CJK code points that have them. | |
1740 | ||
1741 | =item * | |
1742 | ||
1743 | The names that are output for code points with multiple aliases are now the | |
1744 | corrected ones. | |
1745 | ||
1746 | =back | |
1747 | ||
1748 | =item charscript() | |
1749 | ||
1750 | This now correctly returns "Unknown" instead of C<undef> for the script | |
1751 | of a code point that hasn't been assigned another one. | |
1752 | ||
1753 | =item charblock() | |
1754 | ||
1755 | This now correctly returns "No_Block" instead of C<undef> for the block | |
1756 | of a code point that hasn't been assigned to another one. | |
1757 | ||
1758 | =back | |
1759 | ||
5076a392 FC |
1760 | =item * |
1761 | ||
1762 | C<version> has been upgraded from 0.82 to 0.88. | |
1763 | ||
1984204c FC |
1764 | Due to a bug, now fixed, the C<is_strict> and C<is_lax> functions did not |
1765 | work when exported. | |
5076a392 FC |
1766 | |
1767 | =item * | |
1768 | ||
1769 | C<warnings> and C<warnings::register> have been upgraded from version 1.09 | |
1770 | to 1.11 and from version 1.01 to 1.02 respectively. | |
1771 | ||
1772 | Calling C<use warnings> without arguments is now significantly more efficient. | |
1773 | ||
1774 | It is now possible to register warning categories other than the names of | |
1775 | packages using C<warnings::register>. See L<perllexwarn> for more information. | |
1776 | ||
1777 | =item * | |
1778 | ||
1779 | C<VMS::DCLsym> has been upgraded from version 1.03 to 1.05. | |
1780 | ||
1781 | Two bugs have been fixed [perl #84086]: | |
1782 | ||
1783 | The symbol table name was lost when tying a hash, due to a thinko in | |
1784 | C<TIEHASH>. The result was that all tied hashes interacted with the | |
1785 | local symbol table. | |
1786 | ||
1787 | Unless a symbol table name had been explicitly specified in the call | |
1788 | to the constructor, querying the special key ':LOCAL' failed to | |
1789 | identify objects connected to the local symbol table. | |
1790 | ||
1791 | =item * | |
1792 | ||
1793 | C<Win32> has been upgraded from version 0.39 to 0.44. | |
1794 | ||
1984204c FC |
1795 | This release has several new functions: C<Win32::GetSystemMetrics>, |
1796 | C<Win32::GetProductInfo>, C<Win32::GetOSDisplayName>. | |
5076a392 | 1797 | |
1984204c FC |
1798 | The names returned by C<Win32::GetOSName> and C<Win32::GetOSDisplayName> |
1799 | have been corrected. | |
5076a392 | 1800 | |
5076a392 FC |
1801 | =back |
1802 | ||
5076a392 FC |
1803 | =head2 Removed Modules and Pragmata |
1804 | ||
1805 | The following modules have been removed from the core distribution, and if | |
1806 | needed should be installed from CPAN instead. | |
1807 | ||
1808 | =over | |
1809 | ||
1810 | =item C<Class::ISA> | |
1811 | ||
1812 | =item C<Pod::Plainer> | |
1813 | ||
1814 | =item C<Switch> | |
1815 | ||
1816 | =back | |
1817 | ||
1818 | The removal of C<Shell> has been deferred until after 5.14, as the | |
1819 | implementation of C<Shell> shipped with 5.12.0 did not correctly issue the | |
1820 | warning that it was to be removed from core. | |
1821 | ||
1822 | =head1 Documentation | |
1823 | ||
5076a392 FC |
1824 | =head2 New Documentation |
1825 | ||
41e29def | 1826 | =head3 L<perlgpl> |
5076a392 FC |
1827 | |
1828 | L<perlgpl> has been updated to contain GPL version 1, as is included in the | |
1829 | F<README> distributed with perl. | |
1830 | ||
41e29def | 1831 | =head3 Perl 5.12.x delta files |
5076a392 | 1832 | |
41e29def FC |
1833 | The perldelta files for Perl 5.12.1 to 5.12.3 have been added from the |
1834 | maintenance branch: L<perl5121delta>, L<perl5122delta>, L<perl5123delta>. | |
5076a392 FC |
1835 | |
1836 | =head3 L<perlpodstyle> | |
1837 | ||
1838 | New style guide for POD documentation, | |
1839 | split mostly from the NOTES section of the pod2man man page. | |
1840 | ||
41e29def FC |
1841 | =head3 L<perlsource>, L<perlinterp>, L<perlhacktut>, and L<perlhacktips> |
1842 | ||
1843 | See L</L<perlhack> and perlrepository revamp>, below. | |
5076a392 FC |
1844 | |
1845 | =head2 Changes to Existing Documentation | |
1846 | ||
41e29def | 1847 | =head3 L<perlmodlib> is now complete |
4ed2cea4 FC |
1848 | |
1849 | The perlmodlib page that came with Perl 5.12.0 was missing a lot of | |
1850 | modules, due to a bug in the script that generates the list. This has been | |
1851 | fixed [perl #74332]. | |
1852 | ||
41e29def | 1853 | =head3 Replace wrong tr/// table in L<perlebcdic> |
5076a392 | 1854 | |
41e29def | 1855 | L<perlebcdic> contains a helpful table to use in tr/// to convert |
5076a392 FC |
1856 | between EBCDIC and Latin1/ASCII. Unfortunately, the table was the |
1857 | inverse of the one it describes, though the code that used the table | |
1858 | worked correctly for the specific example given. | |
1859 | ||
1860 | The table has been changed to its inverse, and the sample code changed | |
1861 | to correspond, as this is easier for the person trying to follow the | |
1862 | instructions since deriving the old table is somewhat more complicated. | |
1863 | ||
1864 | The table has also been changed to hex from octal, as that is more the norm | |
1865 | these days, and the recipes in the pod altered to print out leading | |
41e29def | 1866 | zeros to make all the values the same length. |
5076a392 | 1867 | |
41e29def | 1868 | =head3 Tricks for user-defined casing |
5076a392 | 1869 | |
41e29def FC |
1870 | L<perlunicode> now contains an explanation of how to override, mangle |
1871 | and otherwise tweak the way perl handles upper-, lower- and other-case | |
1872 | conversions on Unicode data, and how to provide scoped changes to alter | |
1873 | one's own code's behaviour without stomping on anybody else. | |
5076a392 FC |
1874 | |
1875 | =head3 INSTALL explicitly states the requirement for C89 | |
1876 | ||
41e29def | 1877 | This was already true but it's now Officially Stated For The Record. |
5076a392 | 1878 | |
41e29def | 1879 | =head3 Explanation of C<\xI<HH>> and C<\oI<OOO>> escapes |
5076a392 | 1880 | |
41e29def FC |
1881 | L<perlop> has been updated with more detailed explanation of these two |
1882 | character escapes. | |
5076a392 | 1883 | |
41e29def | 1884 | =head3 C<-0I<NNN>> switch |
5076a392 | 1885 | |
41e29def FC |
1886 | In L<perlrun>, the behavior of the C<-0NNN> switch for C<-0400> or higher |
1887 | has been clarified. | |
5076a392 | 1888 | |
41e29def | 1889 | =head3 Deprecation policy |
5076a392 | 1890 | |
41e29def FC |
1891 | L<perlpolicy> now contains the policy on compatibility and deprecation |
1892 | along with definitions of terms like "deprecation" | |
5076a392 | 1893 | |
41e29def | 1894 | =head3 New descriptions in L<perldiag> |
5076a392 FC |
1895 | |
1896 | The following existing diagnostics are now documented: | |
1897 | ||
1898 | =over 4 | |
1899 | ||
1900 | =item * | |
1901 | ||
1902 | L<Ambiguous use of %c resolved as operator %c|perldiag/"Ambiguous use of %c resolved as operator %c"> | |
1903 | ||
1904 | =item * | |
1905 | ||
1906 | L<Ambiguous use of %c{%s} resolved to %c%s|perldiag/"Ambiguous use of %c{%s} resolved to %c%s"> | |
1907 | ||
1908 | =item * | |
1909 | ||
1910 | L<Ambiguous use of %c{%s%s} resolved to %c%s%s|perldiag/"Ambiguous use of %c{%s%s} resolved to %c%s%s"> | |
1911 | ||
1912 | =item * | |
1913 | ||
1914 | L<Ambiguous use of -%s resolved as -&%s()|perldiag/"Ambiguous use of -%s resolved as -&%s()"> | |
1915 | ||
1916 | =item * | |
1917 | ||
1918 | L<Invalid strict version format (%s)|perldiag/"Invalid strict version format (%s)"> | |
1919 | ||
1920 | =item * | |
1921 | ||
1922 | L<Invalid version format (%s)|perldiag/"Invalid version format (%s)"> | |
1923 | ||
1924 | =item * | |
1925 | ||
1926 | L<Invalid version object|perldiag/"Invalid version object"> | |
1927 | ||
1928 | =back | |
1929 | ||
5076a392 FC |
1930 | =head3 L<perlbook> |
1931 | ||
41e29def | 1932 | L<perlbook> has been expanded to cover many more popular books. |
5076a392 | 1933 | |
41e29def | 1934 | =head3 C<SvTRUE> macro |
5076a392 | 1935 | |
41e29def FC |
1936 | The documentation for the C<SvTRUE> macro in |
1937 | L<perlapi> was simply wrong in stating that | |
5076a392 FC |
1938 | get-magic is not processed. It has been corrected. |
1939 | ||
41e29def | 1940 | =head3 L<perlvar> revamp |
5076a392 | 1941 | |
41e29def | 1942 | L<perlvar> reorders the variables and groups them by topic. Each variable |
5076a392 | 1943 | introduced after Perl 5.000 notes the first version in which it is |
41e29def | 1944 | available. L<perlvar> also has a new section for deprecated variables to |
5076a392 FC |
1945 | note when they were removed. |
1946 | ||
41e29def | 1947 | =head3 Array and hash slices in scalar context |
5076a392 | 1948 | |
41e29def | 1949 | These are now documented in L<perldata>. |
5076a392 | 1950 | |
41e29def | 1951 | =head3 C<use locale> and formats |
5076a392 FC |
1952 | |
1953 | L<perlform> and L<perllocale> have been corrected to state that | |
1954 | C<use locale> affects formats. | |
1955 | ||
5076a392 FC |
1956 | =head3 L<overload> |
1957 | ||
5076a392 FC |
1958 | L<overload>'s documentation has practically undergone a rewrite. It |
1959 | is now much more straightforward and clear. | |
1960 | ||
41e29def | 1961 | =head3 L<perlhack> and perlrepository revamp |
5076a392 FC |
1962 | |
1963 | The L<perlhack> and perlrepository documents have been heavily edited and | |
1964 | split up into several new documents. | |
1965 | ||
1966 | The L<perlhack> document is now much shorter, and focuses on the Perl 5 | |
1967 | development process and submitting patches to Perl. The technical content has | |
1968 | been moved to several new documents, L<perlsource>, L<perlinterp>, | |
1969 | L<perlhacktut>, and L<perlhacktips>. This technical content has only been | |
1970 | lightly edited. | |
1971 | ||
1972 | The perlrepository document has been renamed to L<perlgit>. This new document | |
1973 | is just a how-to on using git with the Perl source code. Any other content | |
1974 | that used to be in perlrepository has been moved to perlhack. | |
1975 | ||
41e29def | 1976 | =head3 Time::Piece examples |
5076a392 FC |
1977 | |
1978 | Examples in L<perlfaq4> have been updated to show the use of | |
41e29def | 1979 | L<Time::Piece>. |
5076a392 FC |
1980 | |
1981 | =head1 Diagnostics | |
1982 | ||
1983 | The following additions or changes have been made to diagnostic output, | |
1984 | including warnings and fatal error messages. For the complete list of | |
1985 | diagnostic messages, see L<perldiag>. | |
1986 | ||
1987 | =head2 New Diagnostics | |
1988 | ||
a593b319 FC |
1989 | =head3 New Errors |
1990 | ||
5076a392 FC |
1991 | =over |
1992 | ||
a593b319 | 1993 | =item Closure prototype called |
5076a392 | 1994 | |
a593b319 FC |
1995 | This error occurs when a subroutine reference passed to an attribute |
1996 | handler is called, if the subroutine is a closure [perl #68560]. | |
5076a392 | 1997 | |
a593b319 | 1998 | =item Insecure user-defined property %s |
5076a392 | 1999 | |
a593b319 FC |
2000 | Perl detected tainted data when trying to compile a regular |
2001 | expression that contains a call to a user-defined character property | |
2002 | function, i.e. C<\p{IsFoo}> or C<\p{InFoo}>. | |
2003 | See L<perlunicode/User-Defined Character Properties> and L<perlsec>. | |
5076a392 | 2004 | |
a593b319 | 2005 | =item panic: gp_free failed to free glob pointer - something is repeatedly re-creating entries |
5076a392 | 2006 | |
a593b319 FC |
2007 | This new error is triggered if a destructor called on an object in a |
2008 | typeglob that is being freed creates a new typeglob entry containing an | |
2009 | object with a destructor that creates a new entry containing an object.... | |
5076a392 | 2010 | |
a593b319 | 2011 | =item Parsing code internal error (%s) |
5076a392 | 2012 | |
a593b319 FC |
2013 | This new fatal error is produced when parsing |
2014 | code supplied by an extension violates the | |
2015 | parser's API in a detectable way. | |
5076a392 | 2016 | |
a593b319 | 2017 | =item refcnt: fd %d%s |
5076a392 | 2018 | |
a593b319 FC |
2019 | This new error only occurs if a internal consistency check fails when a |
2020 | pipe is about to be closed. | |
5076a392 | 2021 | |
a593b319 | 2022 | =item Regexp modifier "/%c" may not appear twice |
5076a392 | 2023 | |
a593b319 FC |
2024 | The regular expression pattern has one of the |
2025 | mutually exclusive modifiers repeated. | |
5076a392 | 2026 | |
a593b319 | 2027 | =item Regexp modifiers "/%c" and "/%c" are mutually exclusive |
5076a392 | 2028 | |
a593b319 FC |
2029 | The regular expression pattern has more than one of the mutually |
2030 | exclusive modifiers. | |
5076a392 | 2031 | |
a593b319 | 2032 | =item Using !~ with %s doesn't make sense |
5076a392 | 2033 | |
a593b319 | 2034 | This error occurs when C<!~> is used with C<s///r> or C<y///r>. |
5076a392 | 2035 | |
a593b319 | 2036 | =back |
5076a392 | 2037 | |
a593b319 | 2038 | =head3 New Warnings |
5076a392 | 2039 | |
a593b319 | 2040 | =over |
5076a392 | 2041 | |
a593b319 | 2042 | =item "\b{" is deprecated; use "\b\{" instead |
5076a392 | 2043 | |
a593b319 | 2044 | =item "\B{" is deprecated; use "\B\{" instead |
5076a392 | 2045 | |
a593b319 FC |
2046 | Use of an unescaped "{" immediately following a C<\b> or C<\B> is now |
2047 | deprecated so as to reserve its use for Perl itself in a future release. | |
5076a392 | 2048 | |
a593b319 | 2049 | =item Operation "%s" returns its argument for ... |
5076a392 | 2050 | |
a593b319 FC |
2051 | Performing an operation requiring Unicode semantics (such as case-folding) |
2052 | on a Unicode surrogate or a non-Unicode character now triggers a warning: | |
2053 | 'Operation "%s" returns its argument for ...'. | |
5076a392 | 2054 | |
a593b319 FC |
2055 | =item Use of qw(...) as parentheses is deprecated |
2056 | ||
2057 | See L</"Use of qw(...) as parentheses">, above, for details. | |
5076a392 FC |
2058 | |
2059 | =back | |
2060 | ||
2061 | =head2 Changes to Existing Diagnostics | |
2062 | ||
2063 | =over 4 | |
2064 | ||
2065 | =item * | |
2066 | ||
4ed2cea4 FC |
2067 | The "Variable $foo is not imported" warning that precedes a |
2068 | C<strict 'vars'> error has now been assigned the "misc" category, so that | |
2069 | C<no warnings> will suppress it [perl #73712]. | |
2070 | ||
2071 | =item * | |
2072 | ||
5076a392 FC |
2073 | C<warn> and C<die> now produce 'Wide character' warnings when fed a |
2074 | character outside the byte range if STDERR is a byte-sized handle. | |
2075 | ||
2076 | =item * | |
2077 | ||
2078 | The 'Layer does not match this perl' error message has been replaced with | |
a593b319 | 2079 | these more helpful messages [perl #73754]: |
5076a392 FC |
2080 | |
2081 | =over 4 | |
2082 | ||
2083 | =item * | |
2084 | ||
2085 | PerlIO layer function table size (%d) does not match size expected by this | |
2086 | perl (%d) | |
2087 | ||
2088 | =item * | |
2089 | ||
2090 | PerlIO layer instance size (%d) does not match size expected by this perl | |
2091 | (%d) | |
2092 | ||
2093 | =back | |
2094 | ||
5076a392 FC |
2095 | =item * |
2096 | ||
2097 | The "Found = in conditional" warning that is emitted when a constant is | |
2098 | assigned to a variable in a condition is now withheld if the constant is | |
2099 | actually a subroutine or one generated by C<use constant>, since the value | |
2100 | of the constant may not be known at the time the program is written | |
2101 | [perl #77762]. | |
2102 | ||
2103 | =item * | |
2104 | ||
2105 | Previously, if none of the C<gethostbyaddr>, C<gethostbyname> and | |
2106 | C<gethostent> functions were implemented on a given platform, they would | |
2107 | all die with the message 'Unsupported socket function "gethostent" called', | |
2108 | with analogous messages for C<getnet*> and C<getserv*>. This has been | |
2109 | corrected. | |
2110 | ||
2111 | =item * | |
2112 | ||
a593b319 FC |
2113 | The warning message about unrecognized regular expression escapes passed |
2114 | through has been changed to include any literal '{' following the | |
2115 | two-character escape. E.g., "\q{" is now emitted instead of "\q". | |
5076a392 FC |
2116 | |
2117 | =back | |
2118 | ||
2119 | =head1 Utility Changes | |
2120 | ||
0b88cc74 | 2121 | =head3 L<perlbug> |
5076a392 FC |
2122 | |
2123 | =over 4 | |
2124 | ||
2125 | =item * | |
2126 | ||
0b88cc74 FC |
2127 | L<perlbug> now looks in the EMAIL environment variable for a return address |
2128 | if the REPLY-TO and REPLYTO variables are empty. | |
5076a392 FC |
2129 | |
2130 | =item * | |
2131 | ||
0b88cc74 FC |
2132 | L<perlbug> did not previously generate a From: header, potentially |
2133 | resulting in dropped mail. Now it does include that header. | |
5076a392 FC |
2134 | |
2135 | =item * | |
2136 | ||
0b88cc74 | 2137 | The user's address is now used as the return-path. |
4ed2cea4 | 2138 | |
0b88cc74 FC |
2139 | Many systems these days don't have a valid Internet domain name and |
2140 | perlbug@perl.org does not accept email with a return-path that does | |
2141 | not resolve. So the user's address is now passed to sendmail so it's | |
2142 | less likely to get stuck in a mail queue somewhere [perl #82996]. | |
5076a392 FC |
2143 | |
2144 | =back | |
2145 | ||
0b88cc74 | 2146 | =head3 L<perl5db.pl> |
5076a392 | 2147 | |
0b88cc74 | 2148 | =over |
5076a392 FC |
2149 | |
2150 | =item * | |
2151 | ||
0b88cc74 FC |
2152 | The remote terminal works after forking and spawns new sessions - one |
2153 | for each forked process. | |
5076a392 FC |
2154 | |
2155 | =back | |
2156 | ||
0b88cc74 | 2157 | =head3 L<ptargrep> |
5076a392 FC |
2158 | |
2159 | =over 4 | |
2160 | ||
2161 | =item * | |
2162 | ||
0b88cc74 FC |
2163 | L<ptargrep> is a new utility to apply pattern matching to the contents of |
2164 | files in a tar archive. It comes with C<Archive::Tar>. | |
5076a392 FC |
2165 | |
2166 | =back | |
2167 | ||
2168 | =head1 Configuration and Compilation | |
2169 | ||
61752d82 FC |
2170 | See also L</"Naming fixes in Policy_sh.SH may invalidate Policy.sh">, |
2171 | above. | |
2172 | ||
5076a392 FC |
2173 | =over 4 |
2174 | ||
2175 | =item * | |
2176 | ||
87595b22 FC |
2177 | CCINCDIR and CCLIBDIR for the mingw64 |
2178 | cross-compiler are now correctly under | |
5076a392 FC |
2179 | $(CCHOME)\mingw\include and \lib rather than immediately below $(CCHOME). |
2180 | ||
5076a392 FC |
2181 | This means the 'incpath', 'libpth', 'ldflags', 'lddlflags' and |
2182 | 'ldflags_nolargefiles' values in Config.pm and Config_heavy.pl are now | |
87595b22 | 2183 | set correctly. |
5076a392 FC |
2184 | |
2185 | =item * | |
2186 | ||
87595b22 FC |
2187 | 'make test.valgrind' has been adjusted to account for cpan/dist/ext |
2188 | separation. | |
5076a392 FC |
2189 | |
2190 | =item * | |
2191 | ||
2192 | On compilers that support it, C<-Wwrite-strings> is now added to cflags by | |
2193 | default. | |
2194 | ||
2195 | =item * | |
2196 | ||
2197 | The C<Encode> module can now (once again) be included in a static Perl | |
2198 | build. The special-case handling for this situation got broken in Perl | |
2199 | 5.11.0, and has now been repaired. | |
2200 | ||
1e463951 FC |
2201 | =item * |
2202 | ||
2203 | The previous default size of a PerlIO buffer (4096 bytes) has been increased | |
2204 | to the larger of 8192 bytes and your local BUFSIZ. Benchmarks show that doubling | |
2205 | this decade-old default increases read and write performance in the neighborhood | |
2206 | of 25% to 50% when using the default layers of perlio on top of unix. To choose | |
2207 | a non-default size, such as to get back the old value or to obtain an even | |
2208 | larger value, configure with: | |
2209 | ||
2210 | ./Configure -Accflags=-DPERLIOBUF_DEFAULT_BUFSIZ=N | |
2211 | ||
2212 | where N is the desired size in bytes; it should probably be a multiple of | |
2213 | your page size. | |
2214 | ||
5076a392 FC |
2215 | =back |
2216 | ||
2217 | =head1 Testing | |
2218 | ||
6eff4dda FC |
2219 | XXX This section has not been cleaned up yet. Do we really need this |
2220 | section? How many people are going to find a list of new test files useful? | |
2221 | Anyone hacking on the core can use | |
2222 | ‘git diff v5.12.0..v5.14.0 MANIFEST|grep '^[+-]t/'’. | |
2223 | ||
5076a392 FC |
2224 | XXX Any significant changes to the testing of a freshly built perl should be |
2225 | listed here. Changes which create B<new> files in F<t/> go here as do any | |
2226 | large changes to the testing harness (e.g. when parallel testing was added). | |
2227 | Changes to existing files in F<t/> aren't worth summarising, although the bugs | |
2228 | that they represent may be covered elsewhere. | |
2229 | ||
2230 | =over 4 | |
2231 | ||
2232 | =item * | |
2233 | ||
2234 | F<t/harness> clears PERL5LIB, PERLLIB, PERL5OPT as t/TEST does (a2d3de1) | |
2235 | ||
2236 | =item * | |
2237 | ||
2238 | Many common testing routines were refactored into t/lib/common.pl | |
2239 | ||
2240 | =item * | |
2241 | ||
2242 | Several test files have been modernized to use Test::More | |
2243 | ||
2244 | =item * | |
2245 | ||
2246 | F<t/op/print.t> has been added to test implicit printing of C<$_>. | |
2247 | ||
2248 | =item * | |
2249 | ||
2250 | F<t/io/errnosig.t> has been added to test for restoration of of C<$!> when | |
2251 | leaving signal handlers. | |
2252 | ||
2253 | =item * | |
2254 | ||
2255 | F<t/op/tie_fetch_count.t> has been added to see if C<FETCH> is only called once | |
2256 | on tied variables. | |
2257 | ||
2258 | =item * | |
2259 | ||
2260 | F<lib/Tie/ExtraHash.t> has been added to make sure the, previously untested, | |
2261 | L<Tie::ExtraHash> keeps working. | |
2262 | ||
2263 | =item * | |
2264 | ||
2265 | F<t/re/overload.t> has been added to test against string corruption in pattern | |
2266 | matches on overloaded objects. This is a TODO test. | |
2267 | ||
2268 | =item * | |
2269 | ||
2270 | The new F<t/lib/universal.t> script tests the Internal::* functions and other | |
2271 | things in F<universal.c>. | |
2272 | ||
2273 | =item * | |
2274 | ||
2275 | A rare race condition in F<t/op/while_readdir.t> has been fixed, stopping it | |
2276 | from failing randomly when running tests in parallel. | |
2277 | ||
2278 | =item * | |
2279 | ||
2280 | The new F<t/op/leaky-magic.t> script tests that magic applied to variables in | |
2281 | the main packages does not affect other packages. | |
2282 | ||
2283 | =item * | |
2284 | ||
2285 | The script F<t/op/threads-dirh.t> has been added, which tests interaction | |
2286 | of threads and directory handles. | |
2287 | ||
2288 | =item * | |
2289 | ||
2290 | The new F<t/mro/isa_aliases.t> has been added, which tests that | |
2291 | C<*Foo::ISA = *Bar::ISA> works properly. | |
2292 | ||
2293 | =item * | |
2294 | ||
2295 | F<t/mro/isarev.t> has been added, which tests that C<PL_isarev> (accessible | |
2296 | at the Perl level via C<mro::get_isarev>) is updated properly. | |
2297 | ||
2298 | =item * | |
2299 | ||
2300 | F<t/run/switchd-78586.t> has been added, which tests that [perl #78586] | |
2301 | has been fixed (related to line numbers in the debugger). | |
2302 | ||
2303 | =item * | |
2304 | ||
2305 | C<lib/File/DosGlob.t> has been modernized and now uses C<Test::More>. | |
2306 | ||
2307 | =item * | |
2308 | ||
2309 | A new test script, C<t/porting/filenames.t>, makes sure that filenames and | |
2310 | paths are reasonably portable. | |
2311 | ||
2312 | =item * | |
2313 | ||
2314 | C<t/porting/diag.t> is now several orders of magnitude faster. | |
2315 | ||
2316 | =item * | |
2317 | ||
2318 | C<t/porting/buildtoc.t> now tests that the documentation TOC file is current and well-formed. | |
2319 | ||
2320 | =item * | |
2321 | ||
2322 | C<t/base/while.t> now tests the basics of a while loop with minimal dependencies. | |
2323 | ||
2324 | =item * | |
2325 | ||
2326 | C<t/cmd/while.t> now uses F<test.pl> for better maintainability. | |
2327 | ||
2328 | =item * | |
2329 | ||
2330 | C<t/op/split.t> now tests calls to C<split> without any pattern specified. | |
2331 | ||
2332 | =item * | |
2333 | ||
2334 | F<porting/FindExt.t> now skips all tests on a static (-Uusedl) build | |
2335 | of perl. | |
2336 | ||
2337 | =item * | |
2338 | ||
2339 | F<porting/FindExt.t> now passes on non-Win32 platforms when some | |
2340 | extensions are built statically. | |
2341 | ||
2342 | =item * | |
2343 | ||
2344 | The tests for C<split /\s/> and Unicode have been moved from | |
2345 | F<t/op/split.t> to the new F<t/op/split_unicode.t>. | |
2346 | ||
2347 | =item * | |
2348 | ||
2349 | F<t/re/re.t> has been moved to F<ext/re/t/re_funcs_u.t>. | |
2350 | ||
2351 | =item * | |
2352 | ||
2353 | The tests for [perl #72922] have been moved from F<t/re/qr.t> to the new | |
2354 | F<t/re/qr-72922.t>. | |
2355 | ||
2356 | =item * | |
2357 | ||
2358 | F<t/re/reg_unsafe.t> has been deleted and its only test moved to | |
2359 | F<t/re/pat_advanced.t>. | |
2360 | ||
2361 | =back | |
2362 | ||
2363 | =head1 Platform Support | |
2364 | ||
5076a392 FC |
2365 | =head2 New Platforms |
2366 | ||
5076a392 FC |
2367 | =over 4 |
2368 | ||
2369 | =item AIX | |
2370 | ||
2371 | Perl now builds on AIX 4.2. | |
2372 | ||
2373 | =back | |
2374 | ||
2375 | =head2 Discontinued Platforms | |
2376 | ||
2377 | =over 4 | |
2378 | ||
5076a392 FC |
2379 | =item Apollo DomainOS |
2380 | ||
2381 | The last vestiges of support for this platform have been excised from the | |
2382 | Perl distribution. It was officially discontinued in version 5.12.0. It had | |
2383 | not worked for years before that. | |
2384 | ||
2385 | =item MacOS Classic | |
2386 | ||
2387 | The last vestiges of support for this platform have been excised from the | |
2388 | Perl distribution. It was officially discontinued in an earlier version. | |
2389 | ||
2390 | =back | |
2391 | ||
2392 | =head2 Platform-Specific Notes | |
2393 | ||
554003a2 | 2394 | =head3 Cygwin |
5076a392 FC |
2395 | |
2396 | =over 4 | |
2397 | ||
2398 | =item * | |
2399 | ||
554003a2 | 2400 | MakeMaker has been updated to build man pages on cygwin. |
5076a392 FC |
2401 | |
2402 | =item * | |
2403 | ||
554003a2 FC |
2404 | Improved rebase behaviour |
2405 | ||
2406 | If a dll is updated on cygwin the old imagebase address is reused. | |
2407 | This solves most rebase errors, especially when updating on core dll's. | |
2408 | See L<http://www.tishler.net/jason/software/rebase/rebase-2.4.2.README> for more information. | |
5076a392 FC |
2409 | |
2410 | =item * | |
2411 | ||
554003a2 | 2412 | Support for the standard cygwin dll prefix, which is e.g. needed for FFI's |
5076a392 FC |
2413 | |
2414 | =item * | |
2415 | ||
554003a2 | 2416 | Updated build hints file |
5076a392 FC |
2417 | |
2418 | =back | |
2419 | ||
554003a2 | 2420 | =head3 IRIX |
5076a392 FC |
2421 | |
2422 | Conversion of strings to floating-point numbers is now more accurate on | |
2423 | IRIX systems [perl #32380]. | |
2424 | ||
554003a2 | 2425 | =head3 Mac OS X |
5076a392 FC |
2426 | |
2427 | Early versions of Mac OS X (Darwin) had buggy implementations of the | |
2428 | C<setregid>, C<setreuid>, C<setrgid> and C<setruid> functions, so perl | |
2429 | would pretend they did not exist. | |
2430 | ||
2431 | These functions are now recognised on Mac OS 10.5 (Leopard; Darwin 9) and | |
2432 | higher, as they have been fixed [perl #72990]. | |
2433 | ||
554003a2 | 2434 | =head3 MirBSD |
5076a392 | 2435 | |
554003a2 FC |
2436 | Previously if you built perl with a shared libperl.so on MirBSD (the |
2437 | default config), it would work up to the installation; however, once | |
2438 | installed, it would be unable to find libperl. So path handling is now | |
2439 | treated as in the other BSD dialects. | |
5076a392 | 2440 | |
554003a2 | 2441 | =head3 NetBSD |
5076a392 | 2442 | |
554003a2 FC |
2443 | The NetBSD hints file has been changed to make the system's malloc the |
2444 | default. | |
5076a392 | 2445 | |
554003a2 | 2446 | =head3 Recent OpenBSDs now use perl's malloc |
5076a392 | 2447 | |
554003a2 FC |
2448 | OpenBSD E<gt> 3.7 has a new malloc implementation which is mmap-based |
2449 | and as such can release memory back to the OS; however, perl's use of | |
2450 | this malloc causes a substantial slowdown so we now default to using | |
2451 | perl's malloc instead [perl #75742]. | |
5076a392 | 2452 | |
554003a2 | 2453 | =head3 OpenVOS |
5076a392 | 2454 | |
554003a2 FC |
2455 | perl now builds again with OpenVOS (formerly known as Stratus VOS) |
2456 | [perl #78132]. | |
5076a392 | 2457 | |
554003a2 | 2458 | =head3 Solaris |
5076a392 | 2459 | |
554003a2 FC |
2460 | DTrace is now supported on Solaris. There used to be build failures, but |
2461 | these have been fixed [perl #73630]. | |
5076a392 | 2462 | |
554003a2 | 2463 | =head3 VMS |
5076a392 FC |
2464 | |
2465 | =over | |
2466 | ||
2467 | =item * | |
2468 | ||
554003a2 | 2469 | C<PerlIOUnix_open> now honours the default permissions on VMS. |
5076a392 | 2470 | |
554003a2 FC |
2471 | When C<perlio> became the default and C<unixio> became the default bottom layer, |
2472 | the most common path for creating files from Perl became C<PerlIOUnix_open>, | |
2473 | which has always explicitly used C<0666> as the permission mask. | |
5076a392 | 2474 | |
554003a2 FC |
2475 | To avoid this, C<0777> is now passed as the permissions to C<open()>. In the |
2476 | VMS CRTL, C<0777> has a special meaning over and above intersecting with the | |
2477 | current umask; specifically, it allows Unix syscalls to preserve native default | |
2478 | permissions. | |
5076a392 FC |
2479 | |
2480 | =item * | |
2481 | ||
554003a2 FC |
2482 | The shortening of symbols longer than 31 characters in the C sources is |
2483 | now done by the compiler rather than by xsubpp (which could only do so | |
2484 | for generated symbols in XS code). | |
5076a392 FC |
2485 | |
2486 | =item * | |
2487 | ||
554003a2 FC |
2488 | Record-oriented files (record format variable or variable with fixed control) |
2489 | opened for write by the perlio layer will now be line-buffered to prevent the | |
2490 | introduction of spurious line breaks whenever the perlio buffer fills up. | |
5076a392 FC |
2491 | |
2492 | =back | |
2493 | ||
554003a2 | 2494 | =head3 Windows |
5076a392 | 2495 | |
5a1f7719 FC |
2496 | See also L</"fork() emulation will not wait for signalled children"> and |
2497 | L</"Perl source code is read in text mode on Windows">, above. | |
2498 | ||
5076a392 FC |
2499 | =over 4 |
2500 | ||
2501 | =item * | |
2502 | ||
554003a2 | 2503 | Fixed build process for SDK2003SP1 compilers. |
5076a392 | 2504 | |
554003a2 | 2505 | =item * |
5076a392 | 2506 | |
554003a2 FC |
2507 | When using old 32-bit compilers, the define C<_USE_32BIT_TIME_T> will now |
2508 | be set in C<$Config{ccflags}>. This improves portability when compiling | |
2509 | XS extensions using new compilers, but for a perl compiled with old 32-bit | |
2510 | compilers. | |
5076a392 FC |
2511 | |
2512 | =item * | |
2513 | ||
554003a2 FC |
2514 | C<$Config{gccversion}> is now set correctly when perl is built using the |
2515 | mingw64 compiler from L<http://mingw64.org> [perl #73754]. | |
5076a392 | 2516 | |
554003a2 FC |
2517 | =item * |
2518 | ||
2519 | The build process proceeds more smoothly with mingw and dmake when | |
2520 | F<C:\MSYS\bin> is in the PATH, due to a C<Cwd> fix. | |
5076a392 FC |
2521 | |
2522 | =item * | |
2523 | ||
554003a2 FC |
2524 | Support for building with Visual C++ 2010 is now underway, but is not yet |
2525 | complete. See F<README.win32> or L<perlwin32> for more details. | |
5076a392 | 2526 | |
554003a2 FC |
2527 | =item * |
2528 | ||
2529 | The option to use an externally-supplied C<crypt()>, or to build with no | |
2530 | C<crypt()> at all, has been removed. Perl supplies its own C<crypt()> | |
2531 | implementation for Windows, and the political situation that required | |
2532 | this part of the distribution to sometimes be omitted is long gone. | |
5076a392 FC |
2533 | |
2534 | =back | |
2535 | ||
2536 | =head1 Internal Changes | |
2537 | ||
1e463951 | 2538 | =head2 New APIs |
5076a392 | 2539 | |
1e463951 | 2540 | =head3 CLONE_PARAMS structure added to ease correct thread creation |
5076a392 | 2541 | |
1e463951 FC |
2542 | Modules that create threads should now create C<CLONE_PARAMS> structures |
2543 | by calling the new function C<Perl_clone_params_new()>, and free them with | |
2544 | C<Perl_clone_params_del()>. This will ensure compatibility with any future | |
2545 | changes to the internals of the C<CLONE_PARAMS> structure layout, and that | |
2546 | it is correctly allocated and initialised. | |
5076a392 | 2547 | |
1e463951 | 2548 | =head3 New parsing functions |
5076a392 | 2549 | |
1e463951 FC |
2550 | Several functions have been added for parsing statements or multiple |
2551 | statements: | |
5076a392 | 2552 | |
1e463951 | 2553 | =over |
5076a392 FC |
2554 | |
2555 | =item * | |
2556 | ||
1e463951 | 2557 | C<parse_fullstmt> parses a complete Perl statement. |
5076a392 FC |
2558 | |
2559 | =item * | |
2560 | ||
1e463951 FC |
2561 | C<parse_stmtseq> parses a sequence of statements, up |
2562 | to closing brace or EOF. | |
5076a392 FC |
2563 | |
2564 | =item * | |
2565 | ||
1e463951 | 2566 | C<parse_block> parses a block [perl #78222]. |
5076a392 FC |
2567 | |
2568 | =item * | |
2569 | ||
1e463951 FC |
2570 | C<parse_barestmt> parses a statement |
2571 | without a label. | |
5076a392 FC |
2572 | |
2573 | =item * | |
2574 | ||
1e463951 | 2575 | C<parse_label> parses a statement label, separate from statements. |
5076a392 | 2576 | |
1e463951 | 2577 | =back |
5076a392 | 2578 | |
1e463951 FC |
2579 | The |
2580 | L<C<parse_fullexpr()>|perlapi/parse_fullexpr>, | |
2581 | L<C<parse_listexpr()>|perlapi/parse_listexpr>, | |
2582 | L<C<parse_termexpr()>|perlapi/parse_termexpr>, and | |
2583 | L<C<parse_arithexpr()>|perlapi/parse_arithexpr> | |
2584 | functions have been added to the API. They perform | |
2585 | recursive-descent parsing of expressions at various precedence levels. | |
2586 | They are expected to be used by syntax plugins. | |
5076a392 | 2587 | |
1e463951 | 2588 | See L<perlapi> for details. |
5076a392 | 2589 | |
1e463951 | 2590 | =head3 Hints hash API |
5076a392 | 2591 | |
1e463951 FC |
2592 | A new C API for introspecting the hinthash C<%^H> at runtime has been |
2593 | added. See C<cop_hints_2hv>, C<cop_hints_fetchpvn>, C<cop_hints_fetchpvs>, | |
2594 | C<cop_hints_fetchsv>, and C<hv_copy_hints_hv> in L<perlapi> for details. | |
5076a392 | 2595 | |
1e463951 FC |
2596 | A new, experimental API has been added for accessing the internal |
2597 | structure that Perl uses for C<%^H>. See the functions beginning with | |
2598 | C<cophh_> in L<perlapi>. | |
5076a392 | 2599 | |
1e463951 | 2600 | =head3 C interface to C<caller()> |
5076a392 | 2601 | |
1e463951 FC |
2602 | The C<caller_cx> function has been added as an XSUB-writer's equivalent of |
2603 | C<caller()>. See L<perlapi> for details. | |
5076a392 | 2604 | |
1e463951 | 2605 | =head3 Custom per-subroutine check hooks |
5076a392 | 2606 | |
1e463951 FC |
2607 | XS code in an extension module can now annotate a subroutine (whether |
2608 | implemented in XS or in Perl) so that nominated XS code will be called | |
2609 | at compile time (specifically as part of op checking) to change the op | |
2610 | tree of that subroutine. The compile-time check function (supplied by | |
2611 | the extension module) can implement argument processing that can't be | |
2612 | expressed as a prototype, generate customised compile-time warnings, | |
2613 | perform constant folding for a pure function, inline a subroutine | |
2614 | consisting of sufficiently simple ops, replace the whole call with a | |
2615 | custom op, and so on. This was previously all possible by hooking the | |
2616 | C<entersub> op checker, but the new mechanism makes it easy to tie the | |
2617 | hook to a specific subroutine. See L<perlapi/cv_set_call_checker>. | |
5076a392 | 2618 | |
1e463951 FC |
2619 | To help in writing custom check hooks, several subtasks within standard |
2620 | C<entersub> op checking have been separated out and exposed in the API. | |
5076a392 | 2621 | |
1e463951 | 2622 | =head3 Improved support for custom OPs |
5076a392 | 2623 | |
1e463951 FC |
2624 | Custom ops can now be registered with the new C<custom_op_register> C |
2625 | function and the C<XOP> structure. This will make it easier to add new | |
2626 | properties of custom ops in the future. Two new properties have been added | |
2627 | already, C<xop_class> and C<xop_peep>. | |
5076a392 | 2628 | |
1e463951 FC |
2629 | C<xop_class> is one of the OA_*OP constants, and allows L<B> and other |
2630 | introspection mechanisms to work with custom ops | |
2631 | that aren't BASEOPs. C<xop_peep> is a pointer to | |
2632 | a function that will be called for ops of this | |
2633 | type from C<Perl_rpeep>. | |
5076a392 | 2634 | |
1e463951 FC |
2635 | See L<perlguts/Custom Operators> and L<perlapi/Custom Operators> for more |
2636 | detail. | |
5076a392 | 2637 | |
1e463951 FC |
2638 | The old C<PL_custom_op_names>/C<PL_custom_op_descs> interface is still |
2639 | supported but discouraged. | |
5076a392 | 2640 | |
1e463951 | 2641 | =head3 Scope hooks |
5076a392 | 2642 | |
1e463951 FC |
2643 | It is now possible for XS code to hook into Perl's lexical scope |
2644 | mechanism at compile time, using the new C<Perl_blockhook_register> | |
2645 | function. See L<perlguts/"Compile-time scope hooks">. | |
5076a392 | 2646 | |
1e463951 | 2647 | =head3 The recursive part of the peephole optimizer is now hookable |
5076a392 FC |
2648 | |
2649 | In addition to C<PL_peepp>, for hooking into the toplevel peephole optimizer, a | |
2650 | C<PL_rpeepp> is now available to hook into the optimizer recursing into | |
2651 | side-chains of the optree. | |
2652 | ||
1e463951 | 2653 | =head3 New non-magical variants of existing functions |
5076a392 | 2654 | |
1e463951 FC |
2655 | The following functions/macros have been added to the API. The C<*_nomg> |
2656 | macros are equivalent to their non-_nomg variants, except that they ignore | |
2657 | get-magic. Those ending in C<_flags> allow one to specify whether | |
2658 | get-magic is processed. | |
5076a392 | 2659 | |
1e463951 FC |
2660 | sv_2bool_flags |
2661 | SvTRUE_nomg | |
2662 | sv_2nv_flags | |
2663 | SvNV_nomg | |
2664 | sv_cmp_flags | |
2665 | sv_cmp_locale_flags | |
2666 | sv_eq_flags | |
2667 | sv_collxfrm_flags | |
5076a392 | 2668 | |
1e463951 | 2669 | In some of these cases, the non-_flags functions have |
5076a392 FC |
2670 | been replaced with wrappers around the new functions. |
2671 | ||
1e463951 | 2672 | =head3 pv/pvs/sv versions of existing functions |
5076a392 | 2673 | |
1e463951 | 2674 | Many functions ending with pvn now have equivalent pv/pvs/sv versions. |
5076a392 | 2675 | |
1e463951 | 2676 | =head3 List op-building functions |
5076a392 | 2677 | |
1e463951 FC |
2678 | List op-building functions have been added to the |
2679 | API. See L<op_append_elem|perlapi/op_append_elem>, | |
2680 | L<op_append_list|perlapi/op_append_list>, and | |
2681 | L<op_prepend_elem|perlapi/op_prepend_elem> in L<perlapi>. | |
5076a392 | 2682 | |
1e463951 | 2683 | =head3 C<LINKLIST> |
5076a392 | 2684 | |
1e463951 FC |
2685 | The L<LINKLIST|perlapi/LINKLIST> macro, part of op building that |
2686 | constructs the execution-order op chain, has been added to the API. | |
5076a392 | 2687 | |
1e463951 | 2688 | =head3 Localisation functions |
5076a392 | 2689 | |
1e463951 FC |
2690 | The C<save_freeop>, C<save_op>, C<save_pushi32ptr> and C<save_pushptrptr> |
2691 | functions have been added to the API. | |
5076a392 | 2692 | |
1e463951 | 2693 | =head3 Stash names |
5076a392 | 2694 | |
1e463951 FC |
2695 | A stash can now have a list of effective names in addition to its usual |
2696 | name. The first effective name can be accessed via the C<HvENAME> macro, | |
2697 | which is now the recommended name to use in MRO linearisations (C<HvNAME> | |
2698 | being a fallback if there is no C<HvENAME>). | |
5076a392 | 2699 | |
1e463951 FC |
2700 | These names are added and deleted via C<hv_ename_add> and |
2701 | C<hv_ename_delete>. These two functions are I<not> part of the API. | |
5076a392 | 2702 | |
1e463951 | 2703 | =head3 New functions for finding and removing magic |
5076a392 | 2704 | |
1e463951 FC |
2705 | The L<C<mg_findext()>|perlapi/mg_findext> and |
2706 | L<C<sv_unmagicext()>|perlapi/sv_unmagicext> | |
2707 | functions have been added to the API. | |
2708 | They allow extension authors to find and remove magic attached to | |
2709 | scalars based on both the magic type and the magic virtual table, similar to how | |
2710 | C<sv_magicext()> attaches magic of a certain type and with a given virtual table | |
2711 | to a scalar. This eliminates the need for extensions to walk the list of | |
2712 | C<MAGIC> pointers of an C<SV> to find the magic that belongs to them. | |
5076a392 | 2713 | |
1e463951 | 2714 | =head3 C<find_rundefsv> |
5076a392 | 2715 | |
1e463951 FC |
2716 | This function returns the SV representing C<$_>, whether it's lexical |
2717 | or dynamic. | |
5076a392 | 2718 | |
1e463951 | 2719 | =head3 C<Perl_croak_no_modify> |
5076a392 | 2720 | |
1e463951 FC |
2721 | C<Perl_croak_no_modify()> is short-hand for |
2722 | C<Perl_croak("%s", PL_no_modify)>. | |
5076a392 | 2723 | |
1e463951 | 2724 | =head3 C<PERL_STATIC_INLINE> define |
5076a392 | 2725 | |
1e463951 FC |
2726 | The C<PERL_STATIC_INLINE> define has been added to provide the best-guess |
2727 | incantation to use for static inline functions, if the C compiler supports | |
2728 | C99-style static inline. If it doesn't, it'll give a plain C<static>. | |
5076a392 | 2729 | |
1e463951 FC |
2730 | C<HAS_STATIC_INLINE> can be used to check if the compiler actually supports |
2731 | inline functions. | |
5076a392 | 2732 | |
1e463951 | 2733 | =head3 New C<pv_escape> option for hexadecimal escapes |
5076a392 | 2734 | |
1e463951 FC |
2735 | A new option, C<PERL_PV_ESCAPE_NONASCII>, has been added to C<pv_escape> to |
2736 | dump all characters above ASCII in hexadecimal. Before, one could get all | |
2737 | characters as hexadecimal or the Latin1 non-ASCII as octal. | |
5076a392 | 2738 | |
1e463951 | 2739 | =head3 C<lex_start> |
5076a392 | 2740 | |
1e463951 | 2741 | C<lex_start> has been added to the API, but is considered experimental. |
5076a392 | 2742 | |
1e463951 | 2743 | =head3 C<op_scope()> and C<op_lvalue()> |
5076a392 | 2744 | |
1e463951 FC |
2745 | The C<op_scope()> and C<op_lvalue()> functions have been added to the API, |
2746 | but are considered experimental. | |
5076a392 | 2747 | |
1e463951 | 2748 | =head2 C API Changes |
5076a392 | 2749 | |
1e463951 | 2750 | =head3 C<PERL_POLLUTE> has been removed |
5076a392 | 2751 | |
1e463951 FC |
2752 | The option to define C<PERL_POLLUTE> to expose older 5.005 symbols for |
2753 | backwards compatibility has been removed. It's use was always discouraged, | |
2754 | and MakeMaker contains a more specific escape hatch: | |
5076a392 | 2755 | |
1e463951 | 2756 | perl Makefile.PL POLLUTE=1 |
5076a392 | 2757 | |
1e463951 FC |
2758 | This can be used for modules that have not been upgraded to 5.6 naming |
2759 | conventions (and really should be completely obsolete by now). | |
5076a392 | 2760 | |
1e463951 | 2761 | =head3 Check API compatibility when loading XS modules |
5076a392 | 2762 | |
1e463951 FC |
2763 | When perl's API changes in incompatible ways (which usually happens between |
2764 | major releases), XS modules compiled for previous versions of perl will not | |
2765 | work anymore. They will need to be recompiled against the new perl. | |
5076a392 | 2766 | |
1e463951 FC |
2767 | In order to ensure that modules are recompiled, and to prevent users from |
2768 | accidentally loading modules compiled for old perls into newer ones, the | |
2769 | C<XS_APIVERSION_BOOTCHECK> macro has been added. That macro, which is | |
2770 | called when loading every newly compiled extension, compares the API | |
2771 | version of the running perl with the version a module has been compiled for | |
2772 | and raises an exception if they don't match. | |
5076a392 | 2773 | |
1e463951 | 2774 | =head3 Perl_fetch_cop_label |
5076a392 | 2775 | |
1e463951 FC |
2776 | The first argument of the C API function C<Perl_fetch_cop_label> has changed |
2777 | from C<struct refcounted he *> to C<COP *>, to insulate the user from | |
2778 | implementation details. | |
5076a392 | 2779 | |
1e463951 FC |
2780 | This API function was marked as "may change", and likely isn't in use outside |
2781 | the core. (Neither an unpacked CPAN, nor Google's codesearch, finds any other | |
2782 | references to it.) | |
5076a392 | 2783 | |
1e463951 | 2784 | =head3 GvCV() and GvGP() are no longer lvalues |
5076a392 | 2785 | |
1e463951 FC |
2786 | The new GvCV_set() and GvGP_set() macros are now provided to replace |
2787 | assignment to those two macros. | |
5076a392 | 2788 | |
1e463951 FC |
2789 | This allows a future commit to eliminate some backref magic between GV |
2790 | and CVs, which will require complete control over assignment to the | |
2791 | gp_cv slot. | |
5076a392 | 2792 | |
1e463951 | 2793 | =head3 CvGV() is no longer an lvalue |
5076a392 | 2794 | |
1e463951 FC |
2795 | Under some circumstances, the C<CvGV()> field of a CV is now |
2796 | reference-counted. To ensure consistent behaviour, direct assignment to | |
2797 | it, for example C<CvGV(cv) = gv> is now a compile-time error. A new macro, | |
2798 | C<CvGV_set(cv,gv)> has been introduced to perform this operation | |
2799 | safely. Note that modification of this field is not part of the public | |
2800 | API, regardless of this new macro (and despite its being listed in this section). | |
5076a392 | 2801 | |
1e463951 | 2802 | =head3 CvSTASH() is no longer an lvalue |
5076a392 | 2803 | |
1e463951 FC |
2804 | The C<CvSTASH()> macro can now only be used as an rvalue. C<CvSTASH_set()> |
2805 | has been added to replace assignment to C<CvSTASH()>. This is to ensure | |
2806 | that backreferences are handled properly. These macros are not part of the | |
2807 | API. | |
5076a392 | 2808 | |
1e463951 | 2809 | =head3 Calling conventions for C<newFOROP> and C<newWHILEOP> |
5076a392 | 2810 | |
1e463951 FC |
2811 | The way the parser handles labels has been cleaned up and refactored. As a |
2812 | result, the C<newFOROP()> constructor function no longer takes a parameter | |
2813 | stating what label is to go in the state op. | |
5076a392 | 2814 | |
1e463951 FC |
2815 | The C<newWHILEOP()> and C<newFOROP()> functions no longer accept a line |
2816 | number as a parameter. | |
5076a392 | 2817 | |
1e463951 | 2818 | =head3 Flags passed to C<uvuni_to_utf8_flags> and C<utf8n_to_uvuni> |
5076a392 | 2819 | |
1e463951 FC |
2820 | Some of the flags parameters to uvuni_to_utf8_flags() and |
2821 | utf8n_to_uvuni() have changed. This is a result of Perl's now allowing | |
5076a392 FC |
2822 | internal storage and manipulation of code points that are problematic |
2823 | in some situations. Hence, the default actions for these functions has | |
2824 | been complemented to allow these code points. The new flags are | |
2825 | documented in L<perlapi>. Code that requires the problematic code | |
2826 | points to be rejected needs to change to use these flags. Some flag | |
2827 | names are retained for backward source compatibility, though they do | |
2828 | nothing, as they are now the default. However the flags | |
2829 | C<UNICODE_ALLOW_FDD0>, C<UNICODE_ALLOW_FFFF>, C<UNICODE_ILLEGAL>, and | |
2830 | C<UNICODE_IS_ILLEGAL> have been removed, as they stem from a | |
2831 | fundamentally broken model of how the Unicode non-character code points | |
2832 | should be handled, which is now described in | |
2833 | L<perlunicode/Non-character code points>. See also L</Selected Bug Fixes>. | |
2834 | ||
1e463951 FC |
2835 | XXX Which bugs in particular? Selected Bug Fixes is too long for this link |
2836 | to be meaningful right now. | |
5076a392 | 2837 | |
1e463951 | 2838 | =head2 Deprecated C APIs |
5076a392 | 2839 | |
1e463951 | 2840 | =over |
5076a392 | 2841 | |
1e463951 | 2842 | =item C<Perl_ptr_table_clear> |
5076a392 | 2843 | |
1e463951 FC |
2844 | C<Perl_ptr_table_clear> is no longer part of Perl's public API. Calling it |
2845 | now generates a deprecation warning, and it will be removed in a future | |
2846 | release. | |
5076a392 | 2847 | |
1e463951 | 2848 | =item C<sv_compile_2op> |
5076a392 | 2849 | |
1e463951 FC |
2850 | The C<sv_compile_2op()> API function is now deprecated. Searches suggest |
2851 | that nothing on CPAN is using it, so this should have zero impact. | |
5076a392 | 2852 | |
1e463951 FC |
2853 | It attempted to provide an API to compile code down to an optree, but failed |
2854 | to bind correctly to lexicals in the enclosing scope. It's not possible to | |
2855 | fix this problem within the constraints of its parameters and return value. | |
5076a392 | 2856 | |
1e463951 | 2857 | =item C<find_rundefsvoffset> |
5076a392 | 2858 | |
1e463951 FC |
2859 | The C<find_rundefsvoffset> function has been deprecated. It appeared that |
2860 | its design was insufficient for reliably getting the lexical C<$_> at | |
2861 | run-time. | |
5076a392 | 2862 | |
1e463951 FC |
2863 | Use the new C<find_rundefsv> function or the C<UNDERBAR> macro |
2864 | instead. They directly return the right SV representing C<$_>, whether it's | |
2865 | lexical or dynamic. | |
5076a392 | 2866 | |
1e463951 | 2867 | =item C<CALL_FPTR> and C<CPERLscope> |
5076a392 | 2868 | |
1e463951 FC |
2869 | Those are left from an old implementation of C<MULTIPLICITY> using C++ objects, |
2870 | which was removed in Perl 5.8. Nowadays these macros do exactly nothing, so | |
2871 | they shouldn't be used anymore. | |
5076a392 | 2872 | |
1e463951 FC |
2873 | For compatibility, they are still defined for external C<XS> code. Only |
2874 | extensions defining C<PERL_CORE> must be updated now. | |
5076a392 | 2875 | |
1e463951 | 2876 | =back |
5076a392 | 2877 | |
1e463951 | 2878 | =head2 Other Internal Changes |
5076a392 | 2879 | |
1e463951 | 2880 | =head3 Stack unwinding |
5076a392 | 2881 | |
1e463951 FC |
2882 | The protocol for unwinding the C stack at the last stage of a C<die> |
2883 | has changed how it identifies the target stack frame. This now uses | |
2884 | a separate variable C<PL_restartjmpenv>, where previously it relied on | |
2885 | the C<blk_eval.cur_top_env> pointer in the C<eval> context frame that | |
2886 | has nominally just been discarded. This change means that code running | |
2887 | during various stages of Perl-level unwinding no longer needs to take | |
2888 | care to avoid destroying the ghost frame. | |
5076a392 | 2889 | |
1e463951 | 2890 | =head3 Scope stack entries |
5076a392 | 2891 | |
1e463951 FC |
2892 | The format of entries on the scope stack has been changed, resulting in a |
2893 | reduction of memory usage of about 10%. In particular, the memory used by | |
2894 | the scope stack to record each active lexical variable has been halved. | |
5076a392 | 2895 | |
1e463951 | 2896 | =head3 Memory allocation for pointer tables |
5076a392 | 2897 | |
1e463951 FC |
2898 | Memory allocation for pointer tables has been changed. Previously |
2899 | C<Perl_ptr_table_store> allocated memory from the same arena system as | |
2900 | C<SV> bodies and C<HE>s, with freed memory remaining bound to those arenas | |
2901 | until interpreter exit. Now it allocates memory from arenas private to the | |
2902 | specific pointer table, and that memory is returned to the system when | |
2903 | C<Perl_ptr_table_free> is called. Additionally, allocation and release are | |
2904 | both less CPU intensive. | |
5076a392 | 2905 | |
1e463951 | 2906 | =head3 C<UNDERBAR> |
5076a392 | 2907 | |
1e463951 FC |
2908 | The C<UNDERBAR> macro now calls C<find_rundefsv>. C<dUNDERBAR> is now a |
2909 | noop but should still be used to ensure past and future compatibility. | |
5076a392 | 2910 | |
1e463951 | 2911 | =head3 String comparison routines renamed |
5076a392 | 2912 | |
1e463951 FC |
2913 | The ibcmp_* functions have been renamed and are now called foldEQ, |
2914 | foldEQ_locale and foldEQ_utf8. The old names are still available as | |
2915 | macros. | |
2916 | ||
2917 | =head3 C<chop> and C<chomp> implementations merged | |
2918 | ||
2919 | The opcode bodies for C<chop> and C<chomp> and for C<schop> and C<schomp> | |
2920 | have been merged. The implementation functions C<Perl_do_chop()> and | |
2921 | C<Perl_do_chomp()>, never part of the public API, have been merged and | |
2922 | moved to a static function in F<pp.c>. This shrinks the perl binary | |
2923 | slightly, and should not affect any code outside the core (unless it is | |
2924 | relying on the order of side effects when C<chomp> is passed a I<list> of | |
2925 | values). | |
5076a392 FC |
2926 | |
2927 | =head1 Selected Bug Fixes | |
2928 | ||
e8e35311 FC |
2929 | =head2 I/O |
2930 | ||
5076a392 FC |
2931 | =over 4 |
2932 | ||
2933 | =item * | |
2934 | ||
e8e35311 | 2935 | Perl no longer produces this warning: |
a593b319 FC |
2936 | |
2937 | $ perl -we 'open my $f, ">", \my $x; binmode $f, "scalar"' | |
2938 | Use of uninitialized value in binmode at -e line 1. | |
2939 | ||
2940 | =item * | |
2941 | ||
e8e35311 FC |
2942 | Opening a glob reference via C<< open $fh, "E<gt>", \*glob >> will no longer |
2943 | cause the glob to be corrupted when the filehandle is printed to. This would | |
2944 | cause perl to crash whenever the glob's contents were accessed | |
2945 | [perl #77492]. | |
4ed2cea4 FC |
2946 | |
2947 | =item * | |
2948 | ||
e8e35311 FC |
2949 | PerlIO no longer crashes when called recursively, e.g., from a signal |
2950 | handler. Now it just leaks memory [perl #75556]. | |
4ed2cea4 FC |
2951 | |
2952 | =item * | |
2953 | ||
e8e35311 FC |
2954 | Most I/O functions were not warning for unopened handles unless the |
2955 | 'closed' and 'unopened' warnings categories were both enabled. Now only | |
2956 | C<use warnings 'unopened'> is necessary to trigger these warnings (as was | |
438c239f | 2957 | always meant to be the case). |
4ed2cea4 FC |
2958 | |
2959 | =item * | |
2960 | ||
438c239f | 2961 | There have been several fixes to PerlIO layers: |
e8e35311 | 2962 | |
438c239f FC |
2963 | When C<binmode FH, ":crlf"> pushes the C<:crlf> layer on top of the stack, |
2964 | it no longer enables crlf layers lower in the stack, to avoid unexpected | |
2965 | results [perl #38456]. | |
4ed2cea4 | 2966 | |
438c239f FC |
2967 | Opening a file in C<:raw> mode now does what it advertises to do (first |
2968 | open the file, then binmode it), instead of simply leaving off the top | |
2969 | layer [perl #80764]. | |
5076a392 | 2970 | |
438c239f FC |
2971 | The three layers C<:pop>, C<:utf8> and C<:bytes> didn't allow stacking when |
2972 | opening a file. For example | |
e8e35311 | 2973 | this: |
5076a392 | 2974 | |
e8e35311 | 2975 | open FH, '>:pop:perlio', 'some.file' or die $!; |
5076a392 | 2976 | |
438c239f FC |
2977 | Would throw an error: "Invalid argument". This has been fixed in this |
2978 | release [perl #82484]. | |
5076a392 | 2979 | |
e8e35311 FC |
2980 | =back |
2981 | ||
2982 | =head2 Regular Expression Bug Fixes | |
2983 | ||
2984 | =over | |
5076a392 FC |
2985 | |
2986 | =item * | |
2987 | ||
e8e35311 FC |
2988 | The regular expression engine no longer loops when matching |
2989 | C<"\N{LATIN SMALL LIGATURE FF}" =~ /f+/i> and similar expressions | |
2990 | [perl #72998]. | |
5076a392 FC |
2991 | |
2992 | =item * | |
2993 | ||
e8e35311 FC |
2994 | The trie runtime code should no longer allocate massive amounts of memory, |
2995 | fixing #74484. | |
5076a392 FC |
2996 | |
2997 | =item * | |
2998 | ||
438c239f FC |
2999 | Syntax errors in C<< (?{...}) >> blocks no longer cause panic messages |
3000 | [perl #2353]. | |
5076a392 FC |
3001 | |
3002 | =item * | |
3003 | ||
438c239f FC |
3004 | A pattern like C<(?:(o){2})?> no longer causes a "panic" error |
3005 | [perl #39233]. | |
5076a392 FC |
3006 | |
3007 | =item * | |
3008 | ||
438c239f FC |
3009 | A fatal error in regular expressions containing C<(.*?)> when processing |
3010 | UTF-8 data has been fixed [perl #75680]. | |
5076a392 FC |
3011 | |
3012 | =item * | |
3013 | ||
e8e35311 | 3014 | An erroneous regular expression engine optimisation that caused regex verbs like |
438c239f | 3015 | C<*COMMIT> sometimes to be ignored has been removed. |
5076a392 FC |
3016 | |
3017 | =item * | |
3018 | ||
e8e35311 FC |
3019 | The regular expression bracketed character class C<[\8\9]> was effectively the |
3020 | same as C<[89\000]>, incorrectly matching a NULL character. It also gave | |
3021 | incorrect warnings that the C<8> and C<9> were ignored. Now C<[\8\9]> is the | |
3022 | same as C<[89]> and gives legitimate warnings that C<\8> and C<\9> are | |
3023 | unrecognized escape sequences, passed-through. | |
5076a392 FC |
3024 | |
3025 | =item * | |
3026 | ||
e8e35311 FC |
3027 | A regular expression match in the right-hand side of a global substitution |
3028 | (C<s///g>) that is in the same scope will no longer cause match variables | |
3029 | to have the wrong values on subsequent iterations. This can happen when an | |
3030 | array or hash subscript is interpolated in the right-hand side, as in | |
3031 | C<s|(.)|@a{ print($1), /./ }|g> [perl #19078]. | |
5076a392 FC |
3032 | |
3033 | =item * | |
3034 | ||
438c239f FC |
3035 | Several cases in which characters in the Latin-1 non-ASCII range (0x80 to |
3036 | 0xFF) used not to match themselves or used to match both a character class | |
3037 | and its complement have been fixed. For instance, U+00E2 could match both | |
3038 | C<\w> and C<\W> [perl #78464] [perl #18281] [perl #60156]. | |
5076a392 FC |
3039 | |
3040 | =item * | |
3041 | ||
e8e35311 FC |
3042 | Matching a Unicode character against an alternation containing characters |
3043 | that happened to match continuation bytes in the former's UTF8 | |
3044 | representation (C<qq{\x{30ab}} =~ /\xab|\xa9/>) would cause erroneous | |
3045 | warnings [perl #70998]. | |
5076a392 FC |
3046 | |
3047 | =item * | |
3048 | ||
e8e35311 FC |
3049 | The trie optimisation was not taking empty groups into account, preventing |
3050 | 'foo' from matching C</\A(?:(?:)foo|bar|zot)\z/> [perl #78356]. | |
5076a392 FC |
3051 | |
3052 | =item * | |
3053 | ||
e8e35311 FC |
3054 | A pattern containing a C<+> inside a lookahead would sometimes cause an |
3055 | incorrect match failure in a global match (e.g., C</(?=(\S+))/g>) | |
3056 | [perl #68564]. | |
5076a392 FC |
3057 | |
3058 | =item * | |
3059 | ||
e8e35311 FC |
3060 | A regular expression optimisation would sometimes cause a match with a |
3061 | C<{n,m}> quantifier to fail when it should match [perl #79152]. | |
5076a392 FC |
3062 | |
3063 | =item * | |
3064 | ||
e8e35311 | 3065 | Case insensitive matching in regular expressions compiled under C<use |
438c239f FC |
3066 | locale> now works much more sanely when the pattern or |
3067 | target string is encoded internally in | |
3068 | UTF8. Previously, under these conditions the localeness | |
e8e35311 FC |
3069 | was completely lost. Now, code points above 255 are treated as Unicode, |
3070 | but code points between 0 and 255 are treated using the current locale | |
438c239f FC |
3071 | rules, regardless of whether the pattern or the string is encoded in UTF8. |
3072 | The few case-insensitive matches that cross the 255/256 boundary are not | |
e8e35311 FC |
3073 | allowed. For example, 0xFF does not caselessly match the character at |
3074 | 0x178, LATIN CAPITAL LETTER Y WITH DIAERESIS, because 0xFF may not be | |
3075 | LATIN SMALL LETTER Y in the current locale, and Perl has no way of | |
3076 | knowing if that character even exists in the locale, much less what code | |
3077 | point it is. | |
5076a392 FC |
3078 | |
3079 | =item * | |
3080 | ||
e8e35311 FC |
3081 | The C<(?|...)> regular expression construct no longer crashes if the final |
3082 | branch has more sets of capturing parentheses than any other branch. This | |
3083 | was fixed in Perl 5.10.1 for the case of a single branch, but that fix did | |
3084 | not take multiple branches into account [perl #84746]. | |
5076a392 FC |
3085 | |
3086 | =item * | |
3087 | ||
e8e35311 FC |
3088 | A bug has been fixed in the implementation of C<{...}> quantifiers in |
3089 | regular expressions that prevented the code block in | |
3090 | C</((\w+)(?{ print $2 })){2}/> from seeing the C<$2> sometimes | |
3091 | [perl #84294]. | |
5076a392 | 3092 | |
e8e35311 | 3093 | =back |
5076a392 | 3094 | |
e8e35311 FC |
3095 | =head2 Syntax/Parsing Bugs |
3096 | ||
3097 | =over | |
5076a392 FC |
3098 | |
3099 | =item * | |
3100 | ||
e8e35311 FC |
3101 | C<when(scalar){...}> no longer crashes, but produces a syntax error |
3102 | [perl #74114]. | |
5076a392 FC |
3103 | |
3104 | =item * | |
3105 | ||
e8e35311 FC |
3106 | A label right before a string eval (C<foo: eval $string>) no longer causes |
3107 | the label to be associated also with the first statement inside the eval | |
3108 | [perl #74290] (5.12.1). | |
5076a392 FC |
3109 | |
3110 | =item * | |
3111 | ||
b2c076b5 FC |
3112 | The C<no 5.13.2;> form of C<no> no longer tries to turn on features or |
3113 | pragmata (i.e., strict) [perl #70075]. | |
5076a392 FC |
3114 | |
3115 | =item * | |
3116 | ||
b2c076b5 FC |
3117 | C<BEGIN {require 5.12.0}> now behaves as documented, rather than behaving |
3118 | identically to C<use 5.12.0;>. Previously, C<require> in a C<BEGIN> block | |
3119 | was erroneously executing the C<use feature ':5.12.0'> and | |
3120 | C<use strict;> behaviour, which only C<use> was documented to | |
3121 | provide [perl #69050]. | |
5076a392 FC |
3122 | |
3123 | =item * | |
3124 | ||
b2c076b5 FC |
3125 | Lvalue subroutines are again able to return copy-on-write scalars. This |
3126 | had been broken since version 5.10.0 [perl #75656]. | |
5076a392 FC |
3127 | |
3128 | =item * | |
3129 | ||
5076a392 FC |
3130 | A regression introduced in Perl 5.12.0, making |
3131 | C<< my $x = 3; $x = length(undef) >> result in C<$x> set to C<3> has been | |
b2c076b5 FC |
3132 | fixed. C<$x> will now be C<undef> [perl #85508]. |
3133 | ||
3134 | =item * | |
3135 | ||
3136 | When strict 'refs' mode is off, C<%{...}> in rvalue context returns | |
3137 | C<undef> if its argument is undefined. An optimisation introduced in perl | |
3138 | 5.12.0 to make C<keys %{...}> faster when used as a boolean did not take | |
3139 | this into account, causing C<keys %{+undef}> (and C<keys %$foo> when | |
3140 | C<$foo> is undefined) to be an error, which it should only be in strict | |
3141 | mode [perl #81750]. | |
5076a392 FC |
3142 | |
3143 | =item * | |
3144 | ||
e8e35311 | 3145 | Constant-folding used to cause |
5076a392 | 3146 | |
e8e35311 | 3147 | $text =~ ( 1 ? /phoo/ : /bear/) |
5076a392 | 3148 | |
e8e35311 FC |
3149 | to turn into |
3150 | ||
3151 | $text =~ /phoo/ | |
3152 | ||
3153 | at compile time. Now it correctly matches against C<$_> [perl #20444]. | |
5076a392 FC |
3154 | |
3155 | =item * | |
3156 | ||
e8e35311 FC |
3157 | Parsing Perl code (either with string C<eval> or by loading modules) from |
3158 | within a C<UNITCHECK> block no longer causes the interpreter to crash | |
3159 | [perl #70614]. | |
5076a392 FC |
3160 | |
3161 | =item * | |
3162 | ||
b2c076b5 FC |
3163 | String evals no longer fail after 2 billion scopes have been |
3164 | compiled [perl #83364]. | |
3165 | ||
3166 | =item * | |
3167 | ||
e8e35311 FC |
3168 | The parser no longer hangs when encountering certain Unicode characters, |
3169 | such as U+387 [perl #74022]. | |
5076a392 FC |
3170 | |
3171 | =item * | |
3172 | ||
b2c076b5 FC |
3173 | Several contexts no longer allow a Unicode character to begin a word |
3174 | that should never begin words, for an example an accent that must follow | |
3175 | another character previously could precede all other characters. | |
5076a392 FC |
3176 | |
3177 | =item * | |
3178 | ||
e8e35311 FC |
3179 | Defining a constant with the same name as one of perl's special blocks |
3180 | (e.g., INIT) stopped working in 5.12.0, but has now been fixed | |
3181 | [perl #78634]. | |
5076a392 FC |
3182 | |
3183 | =item * | |
3184 | ||
e8e35311 FC |
3185 | A reference to a literal value used as a hash key (C<$hash{\"foo"}>) used |
3186 | to be stringified, even if the hash was tied [perl #79178]. | |
5076a392 FC |
3187 | |
3188 | =item * | |
3189 | ||
e8e35311 FC |
3190 | A closure containing an C<if> statement followed by a constant or variable |
3191 | is no longer treated as a constant [perl #63540]. | |
5076a392 FC |
3192 | |
3193 | =item * | |
3194 | ||
e8e35311 FC |
3195 | C<state> can now be used with attributes. It used to mean the same thing as |
3196 | C<my> if attributes were present [perl #68658]. | |
5076a392 FC |
3197 | |
3198 | =item * | |
3199 | ||
e8e35311 FC |
3200 | Expressions like C<< @$a > 3 >> no longer cause C<$a> to be mentioned in |
3201 | the "Use of uninitialized value in numeric gt" warning when C<$a> is | |
3202 | undefined (since it is not part of the C<E<gt>> expression, but the operand | |
3203 | of the C<@>) [perl #72090]. | |
5076a392 FC |
3204 | |
3205 | =item * | |
3206 | ||
e8e35311 FC |
3207 | C<require> no longer causes C<caller> to return the wrong file name for |
3208 | the scope that called C<require> and other scopes higher up that had the | |
3209 | same file name [perl #68712]. | |
5076a392 FC |
3210 | |
3211 | =item * | |
3212 | ||
e8e35311 FC |
3213 | Accessing an element of a package array with a hard-coded number (as |
3214 | opposed to an arbitrary expression) would crash if the array did not exist. | |
3215 | Usually the array would be autovivified during compilation, but typeglob | |
3216 | manipulation could remove it, as in these two cases which used to crash: | |
5076a392 | 3217 | |
e8e35311 FC |
3218 | *d = *a; print $d[0]; |
3219 | undef *d; print $d[0]; | |
5076a392 | 3220 | |
e8e35311 | 3221 | =back |
5076a392 | 3222 | |
e262cb24 FC |
3223 | =head2 Stashes, Globs and Method Lookup |
3224 | ||
3225 | Perl 5.10.0 introduced a new internal mechanism for caching MROs (method | |
3226 | resolution orders, or lists of parent classes; aka "isa" caches) to make | |
3227 | method lookup faster (so @ISA arrays would not have to be searched | |
3228 | repeatedly). Unfortunately, this brought with it quite a few bugs. Almost | |
3229 | all of these have been fixed now, along with a few MRO-related bugs that | |
3230 | existed before 5.10.0: | |
3231 | ||
3232 | =over | |
3233 | ||
3234 | =item * | |
3235 | ||
3236 | The following used to have erratic effects on method resolution, because | |
3237 | the "isa" caches were not reset or otherwise ended up listing the wrong | |
3238 | classes. These have been fixed. | |
3239 | ||
3240 | =over | |
3241 | ||
3242 | =item Aliasing packages by assigning to globs [perl #77358] | |
3243 | ||
3244 | =item Deleting packages by deleting their containing stash elements | |
3245 | ||
3246 | =item Undefining the glob containing a package (C<undef *Foo::>) | |
3247 | ||
3248 | =item Undefining an ISA glob (C<undef *Foo::ISA>) | |
3249 | ||
3250 | =item Deleting an ISA stash element (C<delete $Foo::{ISA}>) | |
3251 | ||
3252 | =item Sharing @ISA arrays between classes (via C<*Foo::ISA = \@Bar::ISA> or | |
3253 | C<*Foo::ISA = *Bar::ISA>) [perl #77238] | |
3254 | ||
3255 | =back | |
3256 | ||
3257 | C<undef *Foo::ISA> would even stop a new C<@Foo::ISA> array from updating | |
3258 | caches. | |
3259 | ||
3260 | =item * | |
3261 | ||
3262 | Typeglob assignments would crash if the glob's stash no longer existed, so | |
3263 | long as the glob assigned to was named 'ISA' or the glob on either side of | |
3264 | the assignment contained a subroutine. | |
3265 | ||
3266 | =item * | |
3267 | ||
3268 | C<PL_isarev>, which is accessible to Perl via C<mro::get_isarev> is now | |
3269 | updated properly when packages are deleted or removed from the C<@ISA> of | |
3270 | other classes. This allows many packages to be created and deleted without | |
3271 | causing a memory leak [perl #75176]. | |
3272 | ||
3273 | =back | |
3274 | ||
3275 | In addition, various other bugs related to typeglobs and stashes have been | |
3276 | fixed: | |
5076a392 | 3277 | |
e8e35311 | 3278 | =over |
5076a392 FC |
3279 | |
3280 | =item * | |
3281 | ||
e8e35311 | 3282 | Some work has been done on the internal pointers that link between symbol |
e262cb24 | 3283 | tables (stashes), typeglobs and subroutines. This has the effect that |
e8e35311 FC |
3284 | various edge cases related to deleting stashes or stash entries (e.g. |
3285 | <%FOO:: = ()>), and complex typeglob or code reference aliasing, will no | |
3286 | longer crash the interpreter. | |
5076a392 FC |
3287 | |
3288 | =item * | |
3289 | ||
e262cb24 FC |
3290 | Assigning a reference to a glob copy now assigns to a glob slot instead of |
3291 | overwriting the glob with a scalar [perl #1804] [perl #77508]. | |
5076a392 FC |
3292 | |
3293 | =item * | |
3294 | ||
e8e35311 FC |
3295 | A bug when replacing the glob of a loop variable within the loop has been fixed |
3296 | [perl #21469]. This | |
3297 | means the following code will no longer crash: | |
5076a392 | 3298 | |
e8e35311 FC |
3299 | for $x (...) { |
3300 | *x = *y; | |
3301 | } | |
5076a392 FC |
3302 | |
3303 | =item * | |
3304 | ||
3305 | Assigning a glob to a PVLV used to convert it to a plain string. Now it | |
3306 | works correctly, and a PVLV can hold a glob. This would happen when a | |
3307 | nonexistent hash or array element was passed to a subroutine: | |
3308 | ||
3309 | sub { $_[0] = *foo }->($hash{key}); | |
3310 | # $_[0] would have been the string "*main::foo" | |
3311 | ||
3312 | It also happened when a glob was assigned to, or returned from, an element | |
3313 | of a tied array or hash [perl #36051]. | |
3314 | ||
3315 | =item * | |
44691e6f | 3316 | |
e8e35311 | 3317 | When trying to report C<Use of uninitialized value $Foo::BAR>, crashes could |
e262cb24 FC |
3318 | occur if the glob holding the global variable in question had been detached |
3319 | from its original stash by, for example, C<delete $::{'Foo::'}>. This has | |
3320 | been fixed by disabling the reporting of variable names in those | |
e8e35311 | 3321 | cases. |
44691e6f AB |
3322 | |
3323 | =item * | |
3324 | ||
e8e35311 FC |
3325 | During the restoration of a localised typeglob on scope exit, any |
3326 | destructors called as a result would be able to see the typeglob in an | |
3327 | inconsistent state, containing freed entries, which could result in a | |
3328 | crash. This would affect code like this: | |
3329 | ||
3330 | local *@; | |
3331 | eval { die bless [] }; # puts an object in $@ | |
3332 | sub DESTROY { | |
3333 | local $@; # boom | |
3334 | } | |
3335 | ||
3336 | Now the glob entries are cleared before any destructors are called. This | |
3337 | also means that destructors can vivify entries in the glob. So perl tries | |
3338 | again and, if the entries are re-created too many times, dies with a | |
3339 | 'panic: gp_free...' error message. | |
3340 | ||
3341 | =back | |
3342 | ||
3343 | =head2 Unicode | |
3344 | ||
3345 | =over | |
44691e6f | 3346 | |
5076a392 | 3347 | =item * |
44691e6f | 3348 | |
e8e35311 FC |
3349 | What has become known as the "Unicode Bug" is mostly resolved in this release. |
3350 | Under C<use feature 'unicode_strings'>, the internal storage format of a | |
3351 | string no longer affects the external semantics. There are two known | |
3352 | exceptions. User-defined case changing functions, which are planned to | |
3353 | be deprecated in 5.14, require utf8-encoded strings to function; and the | |
3354 | character C<LATIN SMALL LETTER SHARP S> in regular expression | |
3355 | case-insensitive matching has a somewhat different set of bugs depending | |
3356 | on the internal storage format. Case-insensitive matching of all | |
3357 | characters that have multi-character matches, as this one does, is | |
3358 | problematical in Perl [perl #58182]. | |
911a3729 | 3359 | |
5076a392 | 3360 | =item * |
c8c13991 | 3361 | |
e8e35311 FC |
3362 | The handling of Unicode non-characters has changed. |
3363 | Previously they were mostly considered illegal, except that only one of | |
3364 | the 66 of them was known about in places. The Unicode standard | |
3365 | considers them legal, but forbids the "open interchange" of them. | |
3366 | This is part of the change to allow the internal use of any code point | |
3367 | (see L</Core Enhancements>). Together, these changes resolve | |
9a5a48b7 FC |
3368 | [perl #38722], [perl #51918], [perl #51936], [perl #63446]. |
3369 | ||
3370 | =item * | |
3371 | ||
3372 | Naming a deprecated character in \N{...} no longer leaks memory. | |
3373 | ||
3374 | =item * | |
3375 | ||
3376 | C<chop> now correctly handles characters above "\x{7fffffff}" | |
3377 | [perl #73246]. | |
3378 | ||
3379 | =item * | |
3380 | ||
3381 | Passing to C<index> an offset beyond the end of the string when the string | |
3382 | is encoded internally in UTF8 no longer causes panics [perl #75898]. | |
3383 | ||
3384 | =item * | |
3385 | ||
3386 | C<warn()> and C<die()> now respect utf8-encoded scalars [perl #45549]. | |
c8c13991 | 3387 | |
5076a392 FC |
3388 | =item * |
3389 | ||
9a5a48b7 | 3390 | Sometimes the UTF8 length cache would not be reset on a value |
e8e35311 | 3391 | returned by substr, causing C<length(substr($uni_string,...))> to give |
9a5a48b7 FC |
3392 | wrong answers. With C<${^UTF8CACHE}> set to -1, it would produce a 'panic' |
3393 | error message, too [perl #77692]. | |
e8e35311 FC |
3394 | |
3395 | =back | |
3396 | ||
e9784f55 | 3397 | =head2 Ties, Overloading and Other Magic |
e8e35311 FC |
3398 | |
3399 | =over | |
35cdccfc | 3400 | |
5076a392 FC |
3401 | =item * |
3402 | ||
e8e35311 | 3403 | Overloading now works properly in conjunction with tied |
e9784f55 | 3404 | variables. What formerly happened was that most ops checked their |
e8e35311 FC |
3405 | arguments for overloading I<before> checking for magic, so for example |
3406 | an overloaded object returned by a tied array access would usually be | |
e9784f55 | 3407 | treated as not overloaded [RT #57012]. |
911a3729 | 3408 | |
5076a392 | 3409 | =item * |
58f55cb3 | 3410 | |
e9784f55 FC |
3411 | Various cases of magic (e.g., tie methods) being called on tied variables |
3412 | too many or too few times have been fixed: | |
3413 | ||
3414 | =over | |
58f55cb3 | 3415 | |
5076a392 | 3416 | =item * |
6b3df227 | 3417 | |
e9784f55 | 3418 | FETCH is no longer called on tied variables in void context. |
6b3df227 | 3419 | |
5076a392 | 3420 | =item * |
c9989a74 | 3421 | |
e9784f55 | 3422 | C<$tied-E<gt>()> did not always call FETCH [perl #8438]. |
c9989a74 | 3423 | |
5076a392 FC |
3424 | =item * |
3425 | ||
e9784f55 FC |
3426 | Filetest operators and C<y///> and C<tr///> were calling FETCH too |
3427 | many times. | |
b7188eb5 FC |
3428 | |
3429 | =item * | |
3430 | ||
e9784f55 FC |
3431 | The C<=> operator used to ignore magic on its right-hand side if the |
3432 | scalar happened to hold a typeglob (if a typeglob was the last thing | |
3433 | returned from or assigned to a tied scalar) [perl #77498]. | |
911a3729 | 3434 | |
5076a392 | 3435 | =item * |
c8c13991 | 3436 | |
e9784f55 FC |
3437 | Dereference operators used to ignore magic if the argument was a |
3438 | reference already (e.g., from a previous FETCH) [perl #72144]. | |
c8c13991 | 3439 | |
e8e35311 | 3440 | =item * |
f00d3350 | 3441 | |
e9784f55 FC |
3442 | C<splice> now calls set-magic (so changes made |
3443 | by C<splice @ISA> are respected by method calls) [perl #78400]. | |
3444 | ||
3445 | =back | |
c34a735e | 3446 | |
5076a392 | 3447 | =item * |
b7188eb5 | 3448 | |
e9784f55 FC |
3449 | String C<eval> now detects taintedness of overloaded or tied |
3450 | arguments [perl #75716]. | |
b7188eb5 | 3451 | |
5076a392 | 3452 | =item * |
b7188eb5 | 3453 | |
e9784f55 FC |
3454 | String C<eval> and regular expression matches against objects with string |
3455 | overloading no longer cause memory corruption or crashes [perl 77084]. | |
c34a735e | 3456 | |
05dbc6f8 KW |
3457 | =item * |
3458 | ||
e9784f55 FC |
3459 | L<readline|perlfunc/"readline EXPR"> now honors C<< <> >> overloading on tied |
3460 | arguments. | |
7b98b857 | 3461 | |
5076a392 | 3462 | =item * |
05dbc6f8 | 3463 | |
e9784f55 FC |
3464 | C<< E<lt>exprE<gt> >> always respects overloading now if the expression is |
3465 | overloaded. | |
3466 | ||
3467 | Due to the way that 'E<lt>E<gt> as glob' was parsed differently from | |
3468 | 'E<lt>E<gt> as filehandle' from 5.6 onwards, something like C<< E<lt>$foo[0]E<gt> >> did | |
3469 | not handle overloading, even if C<$foo[0]> was an overloaded object. This | |
3470 | was contrary to the documentation for overload, and meant that C<< E<lt>E<gt> >> | |
3471 | could not be used as a general overloaded iterator operator. | |
05dbc6f8 KW |
3472 | |
3473 | =item * | |
3474 | ||
e8e35311 FC |
3475 | The fallback behaviour of overloading on binary operators was asymmetric |
3476 | [perl #71286]. | |
05dbc6f8 | 3477 | |
5076a392 | 3478 | =item * |
05dbc6f8 | 3479 | |
e9784f55 FC |
3480 | Magic applied to variables in the main package no longer affects other packages. |
3481 | See L</Magic variables outside the main package> above [perl #76138]. | |
05dbc6f8 | 3482 | |
5076a392 | 3483 | =item * |
05dbc6f8 | 3484 | |
e9784f55 FC |
3485 | Sometimes magic (ties, taintedness, etc.) attached to variables could cause |
3486 | an object to last longer than it should, or cause a crash if a tied | |
3487 | variable were freed from within a tie method. These have been fixed | |
3488 | [perl #81230]. | |
05dbc6f8 KW |
3489 | |
3490 | =item * | |
3491 | ||
e8e35311 FC |
3492 | DESTROY methods of objects implementing ties are no longer able to crash by |
3493 | accessing the tied variable through a weak reference [perl #86328]. | |
3494 | ||
3495 | =back | |
3496 | ||
3497 | =head2 The Debugger | |
3498 | ||
3499 | =over | |
05dbc6f8 | 3500 | |
5076a392 | 3501 | =item * |
05dbc6f8 | 3502 | |
bdb0cce8 | 3503 | The Perl debugger now also works in taint mode [perl #76872]. |
4d56cd4f | 3504 | |
5076a392 FC |
3505 | =item * |
3506 | ||
bdb0cce8 | 3507 | Subroutine redefinition works once more in the debugger [perl #48332]. |
05dbc6f8 | 3508 | |
911a3729 FC |
3509 | =item * |
3510 | ||
e8e35311 | 3511 | When C<-d> is used on the shebang (C<#!>) line, the debugger now has access |
bdb0cce8 | 3512 | to the lines of the main program. In the past, this sometimes worked and |
e8e35311 FC |
3513 | sometimes did not, depending on what order things happened to be arranged |
3514 | in memory [perl #71806]. | |
911a3729 | 3515 | |
5076a392 | 3516 | =item * |
44691e6f | 3517 | |
bdb0cce8 FC |
3518 | A possible memory leak when using L<caller()|perlfunc/"caller EXPR"> to set |
3519 | C<@DB::args> has been fixed. | |
3520 | ||
3521 | =item * | |
3522 | ||
3523 | Perl no longer stomps on $DB::single, $DB::trace and $DB::signal if they | |
3524 | already have values when $^P is assigned to [perl #72422]. | |
44691e6f | 3525 | |
5076a392 | 3526 | =item * |
44691e6f | 3527 | |
e8e35311 FC |
3528 | C<#line> directives in string evals were not properly updating the arrays |
3529 | of lines of code (C<< @{"_<..."} >>) that the debugger (or any debugging or | |
3530 | profiling module) uses. In threaded builds, they were not being updated at | |
3531 | all. In non-threaded builds, the line number was ignored, so any change to | |
3532 | the existing line number would cause the lines to be misnumbered | |
3533 | [perl #79442]. | |
3534 | ||
3535 | =back | |
3536 | ||
3537 | =head2 Threads, Signals | |
3538 | ||
3539 | =over | |
9dc513c5 | 3540 | |
5076a392 FC |
3541 | =item * |
3542 | ||
e8e35311 FC |
3543 | Don't accidentally clone lexicals in scope within active stack frames in |
3544 | the parent when creating a child thread (RT #73086) (05d04d). | |
9dc513c5 DG |
3545 | |
3546 | =item * | |
3547 | ||
e8e35311 FC |
3548 | Defer signal handling when shared SV locks are held to avoid |
3549 | deadlocks (RT #74868) (65c742). | |
9dc513c5 | 3550 | |
5076a392 | 3551 | =item * |
9dc513c5 | 3552 | |
e8e35311 | 3553 | Within signal handlers, C<$!> is now implicitly localized. |
44691e6f | 3554 | |
5076a392 | 3555 | =item * |
6d96b0fe | 3556 | |
e8e35311 FC |
3557 | Several memory leaks in cloning and freeing threaded Perl interpreters have been |
3558 | fixed [perl #77352]. | |
6d96b0fe | 3559 | |
5076a392 | 3560 | =item * |
6d96b0fe | 3561 | |
e8e35311 FC |
3562 | Creating a new thread when directory handles were open used to cause a |
3563 | crash, because the handles were not cloned, but simply passed to the new | |
3564 | thread, resulting in a double free. | |
3565 | ||
3566 | Now directory handles are cloned properly, on systems that have a C<fchdir> | |
3567 | function. On other systems, new threads simply do not inherit directory | |
3568 | handles from their parent threads [perl #75154]. | |
6d96b0fe | 3569 | |
5076a392 | 3570 | =item * |
b7188eb5 | 3571 | |
e8e35311 FC |
3572 | The typeglob C<*,>, which holds the scalar variable C<$,> (output field |
3573 | separator), had the wrong reference count in child threads. | |
b7188eb5 | 3574 | |
5076a392 | 3575 | =item * |
b7188eb5 | 3576 | |
e8e35311 FC |
3577 | [perl #78494] When pipes are shared between threads, the C<close> function |
3578 | (and any implicit close, such as on thread exit) no longer blocks. |