| 1 |
.\" -*- mode: troff; coding: utf-8 -*- |
| 2 |
.\" Automatically generated by Pod::Man 5.0102 (Pod::Simple 3.45) |
| 3 |
.\" |
| 4 |
.\" Standard preamble: |
| 5 |
.\" ======================================================================== |
| 6 |
.de Sp \" Vertical space (when we can't use .PP) |
| 7 |
.if t .sp .5v |
| 8 |
.if n .sp |
| 9 |
.. |
| 10 |
.de Vb \" Begin verbatim text |
| 11 |
.ft CW |
| 12 |
.nf |
| 13 |
.ne \\$1 |
| 14 |
.. |
| 15 |
.de Ve \" End verbatim text |
| 16 |
.ft R |
| 17 |
.fi |
| 18 |
.. |
| 19 |
.\" \*(C` and \*(C' are quotes in nroff, nothing in troff, for use with C<>. |
| 20 |
.ie n \{\ |
| 21 |
. ds C` "" |
| 22 |
. ds C' "" |
| 23 |
'br\} |
| 24 |
.el\{\ |
| 25 |
. ds C` |
| 26 |
. ds C' |
| 27 |
'br\} |
| 28 |
.\" |
| 29 |
.\" Escape single quotes in literal strings from groff's Unicode transform. |
| 30 |
.ie \n(.g .ds Aq \(aq |
| 31 |
.el .ds Aq ' |
| 32 |
.\" |
| 33 |
.\" If the F register is >0, we'll generate index entries on stderr for |
| 34 |
.\" titles (.TH), headers (.SH), subsections (.SS), items (.Ip), and index |
| 35 |
.\" entries marked with X<> in POD. Of course, you'll have to process the |
| 36 |
.\" output yourself in some meaningful fashion. |
| 37 |
.\" |
| 38 |
.\" Avoid warning from groff about undefined register 'F'. |
| 39 |
.de IX |
| 40 |
.. |
| 41 |
.nr rF 0 |
| 42 |
.if \n(.g .if rF .nr rF 1 |
| 43 |
.if (\n(rF:(\n(.g==0)) \{\ |
| 44 |
. if \nF \{\ |
| 45 |
. de IX |
| 46 |
. tm Index:\\$1\t\\n%\t"\\$2" |
| 47 |
.. |
| 48 |
. if !\nF==2 \{\ |
| 49 |
. nr % 0 |
| 50 |
. nr F 2 |
| 51 |
. \} |
| 52 |
. \} |
| 53 |
.\} |
| 54 |
.rr rF |
| 55 |
.\" ======================================================================== |
| 56 |
.\" |
| 57 |
.IX Title "PERL5160DELTA 1" |
| 58 |
.TH PERL5160DELTA 1 2025-11-16 "perl v5.40.1" "Perl Programmers Reference Guide" |
| 59 |
.\" For nroff, turn off justification. Always turn off hyphenation; it makes |
| 60 |
.\" way too many mistakes in technical documents. |
| 61 |
.if n .ad l |
| 62 |
.nh |
| 63 |
.SH NAME |
| 64 |
perl5160delta \- what is new for perl v5.16.0 |
| 65 |
.SH DESCRIPTION |
| 66 |
.IX Header "DESCRIPTION" |
| 67 |
This document describes differences between the 5.14.0 release and |
| 68 |
the 5.16.0 release. |
| 69 |
.PP |
| 70 |
If you are upgrading from an earlier release such as 5.12.0, first read |
| 71 |
perl5140delta, which describes differences between 5.12.0 and |
| 72 |
5.14.0. |
| 73 |
.PP |
| 74 |
Some bug fixes in this release have been backported to later |
| 75 |
releases of 5.14.x. Those are indicated with the 5.14.x version in |
| 76 |
parentheses. |
| 77 |
.SH Notice |
| 78 |
.IX Header "Notice" |
| 79 |
With the release of Perl 5.16.0, the 5.12.x series of releases is now out of |
| 80 |
its support period. There may be future 5.12.x releases, but only in the |
| 81 |
event of a critical security issue. Users of Perl 5.12 or earlier should |
| 82 |
consider upgrading to a more recent release of Perl. |
| 83 |
.PP |
| 84 |
This policy is described in greater detail in |
| 85 |
perlpolicy. |
| 86 |
.SH "Core Enhancements" |
| 87 |
.IX Header "Core Enhancements" |
| 88 |
.ie n .SS """use \fIVERSION\fP""" |
| 89 |
.el .SS "\f(CWuse \fP\f(CIVERSION\fP\f(CW\fP" |
| 90 |
.IX Subsection "use VERSION" |
| 91 |
As of this release, version declarations like \f(CW\*(C`use v5.16\*(C'\fR now disable |
| 92 |
all features before enabling the new feature bundle. This means that |
| 93 |
the following holds true: |
| 94 |
.PP |
| 95 |
.Vb 4 |
| 96 |
\& use 5.016; |
| 97 |
\& # only 5.16 features enabled here |
| 98 |
\& use 5.014; |
| 99 |
\& # only 5.14 features enabled here (not 5.16) |
| 100 |
.Ve |
| 101 |
.PP |
| 102 |
\&\f(CW\*(C`use v5.12\*(C'\fR and higher continue to enable strict, but explicit \f(CW\*(C`use |
| 103 |
strict\*(C'\fR and \f(CW\*(C`no strict\*(C'\fR now override the version declaration, even |
| 104 |
when they come first: |
| 105 |
.PP |
| 106 |
.Vb 3 |
| 107 |
\& no strict; |
| 108 |
\& use 5.012; |
| 109 |
\& # no strict here |
| 110 |
.Ve |
| 111 |
.PP |
| 112 |
There is a new ":default" feature bundle that represents the set of |
| 113 |
features enabled before any version declaration or \f(CW\*(C`use feature\*(C'\fR has |
| 114 |
been seen. Version declarations below 5.10 now enable the ":default" |
| 115 |
feature set. This does not actually change the behavior of \f(CW\*(C`use |
| 116 |
v5.8\*(C'\fR, because features added to the ":default" set are those that were |
| 117 |
traditionally enabled by default, before they could be turned off. |
| 118 |
.PP |
| 119 |
\&\f(CW\*(C`no feature\*(C'\fR now resets to the default feature set. To disable all |
| 120 |
features (which is likely to be a pretty special-purpose request, since |
| 121 |
it presumably won't match any named set of semantics) you can now |
| 122 |
write \f(CW\*(C`no feature \*(Aq:all\*(Aq\*(C'\fR. |
| 123 |
.PP |
| 124 |
\&\f(CW$[\fR is now disabled under \f(CW\*(C`use v5.16\*(C'\fR. It is part of the default |
| 125 |
feature set and can be turned on or off explicitly with \f(CWuse feature |
| 126 |
\&\*(Aqarray_base\*(Aq\fR. |
| 127 |
.ie n .SS """_\|_SUB_\|_""" |
| 128 |
.el .SS \f(CW_\|_SUB_\|_\fP |
| 129 |
.IX Subsection "__SUB__" |
| 130 |
The new \f(CW\*(C`_\|_SUB_\|_\*(C'\fR token, available under the \f(CW\*(C`current_sub\*(C'\fR feature |
| 131 |
(see feature) or \f(CW\*(C`use v5.16\*(C'\fR, returns a reference to the current |
| 132 |
subroutine, making it easier to write recursive closures. |
| 133 |
.SS "New and Improved Built-ins" |
| 134 |
.IX Subsection "New and Improved Built-ins" |
| 135 |
\fIMore consistent \fR\f(CI\*(C`eval\*(C'\fR |
| 136 |
.IX Subsection "More consistent eval" |
| 137 |
.PP |
| 138 |
The \f(CW\*(C`eval\*(C'\fR operator sometimes treats a string argument as a sequence of |
| 139 |
characters and sometimes as a sequence of bytes, depending on the |
| 140 |
internal encoding. The internal encoding is not supposed to make any |
| 141 |
difference, but there is code that relies on this inconsistency. |
| 142 |
.PP |
| 143 |
The new \f(CW\*(C`unicode_eval\*(C'\fR and \f(CW\*(C`evalbytes\*(C'\fR features (enabled under \f(CWuse |
| 144 |
5.16.0\fR) resolve this. The \f(CW\*(C`unicode_eval\*(C'\fR feature causes \f(CWeval |
| 145 |
$string\fR to treat the string always as Unicode. The \f(CW\*(C`evalbytes\*(C'\fR |
| 146 |
features provides a function, itself called \f(CW\*(C`evalbytes\*(C'\fR, which |
| 147 |
evaluates its argument always as a string of bytes. |
| 148 |
.PP |
| 149 |
These features also fix oddities with source filters leaking to outer |
| 150 |
dynamic scopes. |
| 151 |
.PP |
| 152 |
See feature for more detail. |
| 153 |
.PP |
| 154 |
\fR\f(CI\*(C`substr\*(C'\fR\fI lvalue revamp\fR |
| 155 |
.IX Subsection "substr lvalue revamp" |
| 156 |
.PP |
| 157 |
When \f(CW\*(C`substr\*(C'\fR is called in lvalue or potential lvalue context with two |
| 158 |
or three arguments, a special lvalue scalar is returned that modifies |
| 159 |
the original string (the first argument) when assigned to. |
| 160 |
.PP |
| 161 |
Previously, the offsets (the second and third arguments) passed to |
| 162 |
\&\f(CW\*(C`substr\*(C'\fR would be converted immediately to match the string, negative |
| 163 |
offsets being translated to positive and offsets beyond the end of the |
| 164 |
string being truncated. |
| 165 |
.PP |
| 166 |
Now, the offsets are recorded without modification in the special |
| 167 |
lvalue scalar that is returned, and the original string is not even |
| 168 |
looked at by \f(CW\*(C`substr\*(C'\fR itself, but only when the returned lvalue is |
| 169 |
read or modified. |
| 170 |
.PP |
| 171 |
These changes result in an incompatible change: |
| 172 |
.PP |
| 173 |
If the original string changes length after the call to \f(CW\*(C`substr\*(C'\fR but |
| 174 |
before assignment to its return value, negative offsets will remember |
| 175 |
their position from the end of the string, affecting code like this: |
| 176 |
.PP |
| 177 |
.Vb 5 |
| 178 |
\& my $string = "string"; |
| 179 |
\& my $lvalue = \esubstr $string, \-4, 2; |
| 180 |
\& print $$lvalue, "\en"; # prints "ri" |
| 181 |
\& $string = "bailing twine"; |
| 182 |
\& print $$lvalue, "\en"; # prints "wi"; used to print "il" |
| 183 |
.Ve |
| 184 |
.PP |
| 185 |
The same thing happens with an omitted third argument. The returned |
| 186 |
lvalue will always extend to the end of the string, even if the string |
| 187 |
becomes longer. |
| 188 |
.PP |
| 189 |
Since this change also allowed many bugs to be fixed (see |
| 190 |
"The \f(CW\*(C`substr\*(C'\fR operator"), and since the behavior |
| 191 |
of negative offsets has never been specified, the |
| 192 |
change was deemed acceptable. |
| 193 |
.PP |
| 194 |
\fIReturn value of \fR\f(CI\*(C`tied\*(C'\fR |
| 195 |
.IX Subsection "Return value of tied" |
| 196 |
.PP |
| 197 |
The value returned by \f(CW\*(C`tied\*(C'\fR on a tied variable is now the actual |
| 198 |
scalar that holds the object to which the variable is tied. This |
| 199 |
lets ties be weakened with \f(CW\*(C`Scalar::Util::weaken(tied |
| 200 |
$tied_variable)\*(C'\fR. |
| 201 |
.SS "Unicode Support" |
| 202 |
.IX Subsection "Unicode Support" |
| 203 |
\fISupports (almost) Unicode 6.1\fR |
| 204 |
.IX Subsection "Supports (almost) Unicode 6.1" |
| 205 |
.PP |
| 206 |
Besides the addition of whole new scripts, and new characters in |
| 207 |
existing scripts, this new version of Unicode, as always, makes some |
| 208 |
changes to existing characters. One change that may trip up some |
| 209 |
applications is that the General Category of two characters in the |
| 210 |
Latin\-1 range, PILCROW SIGN and SECTION SIGN, has been changed from |
| 211 |
Other_Symbol to Other_Punctuation. The same change has been made for |
| 212 |
a character in each of Tibetan, Ethiopic, and Aegean. |
| 213 |
The code points U+3248..U+324F (CIRCLED NUMBER TEN ON BLACK SQUARE |
| 214 |
through CIRCLED NUMBER EIGHTY ON BLACK SQUARE) have had their General |
| 215 |
Category changed from Other_Symbol to Other_Numeric. The Line Break |
| 216 |
property has changes for Hebrew and Japanese; and because of |
| 217 |
other changes in 6.1, the Perl regular expression construct \f(CW\*(C`\eX\*(C'\fR now |
| 218 |
works differently for some characters in Thai and Lao. |
| 219 |
.PP |
| 220 |
New aliases (synonyms) have been defined for many property values; |
| 221 |
these, along with the previously existing ones, are all cross-indexed in |
| 222 |
perluniprops. |
| 223 |
.PP |
| 224 |
The return value of \f(CWcharnames::viacode()\fR is affected by other |
| 225 |
changes: |
| 226 |
.PP |
| 227 |
.Vb 10 |
| 228 |
\& Code point Old Name New Name |
| 229 |
\& U+000A LINE FEED (LF) LINE FEED |
| 230 |
\& U+000C FORM FEED (FF) FORM FEED |
| 231 |
\& U+000D CARRIAGE RETURN (CR) CARRIAGE RETURN |
| 232 |
\& U+0085 NEXT LINE (NEL) NEXT LINE |
| 233 |
\& U+008E SINGLE\-SHIFT 2 SINGLE\-SHIFT\-2 |
| 234 |
\& U+008F SINGLE\-SHIFT 3 SINGLE\-SHIFT\-3 |
| 235 |
\& U+0091 PRIVATE USE 1 PRIVATE USE\-1 |
| 236 |
\& U+0092 PRIVATE USE 2 PRIVATE USE\-2 |
| 237 |
\& U+2118 SCRIPT CAPITAL P WEIERSTRASS ELLIPTIC FUNCTION |
| 238 |
.Ve |
| 239 |
.PP |
| 240 |
Perl will accept any of these names as input, but |
| 241 |
\&\f(CWcharnames::viacode()\fR now returns the new name of each pair. The |
| 242 |
change for U+2118 is considered by Unicode to be a correction, that is |
| 243 |
the original name was a mistake (but again, it will remain forever valid |
| 244 |
to use it to refer to U+2118). But most of these changes are the |
| 245 |
fallout of the mistake Unicode 6.0 made in naming a character used in |
| 246 |
Japanese cell phones to be "BELL", which conflicts with the longstanding |
| 247 |
industry use of (and Unicode's recommendation to use) that name |
| 248 |
to mean the ASCII control character at U+0007. Therefore, that name |
| 249 |
has been deprecated in Perl since v5.14, and any use of it will raise a |
| 250 |
warning message (unless turned off). The name "ALERT" is now the |
| 251 |
preferred name for this code point, with "BEL" an acceptable short |
| 252 |
form. The name for the new cell phone character, at code point U+1F514, |
| 253 |
remains undefined in this version of Perl (hence we don't |
| 254 |
implement quite all of Unicode 6.1), but starting in v5.18, BELL will mean |
| 255 |
this character, and not U+0007. |
| 256 |
.PP |
| 257 |
Unicode has taken steps to make sure that this sort of mistake does not |
| 258 |
happen again. The Standard now includes all generally accepted |
| 259 |
names and abbreviations for control characters, whereas previously it |
| 260 |
didn't (though there were recommended names for most of them, which Perl |
| 261 |
used). This means that most of those recommended names are now |
| 262 |
officially in the Standard. Unicode did not recommend names for the |
| 263 |
four code points listed above between U+008E and U+008F, and in |
| 264 |
standardizing them Unicode subtly changed the names that Perl had |
| 265 |
previously given them, by replacing the final blank in each name by a |
| 266 |
hyphen. Unicode also officially accepts names that Perl had deprecated, |
| 267 |
such as FILE SEPARATOR. Now the only deprecated name is BELL. |
| 268 |
Finally, Perl now uses the new official names instead of the old |
| 269 |
(now considered obsolete) names for the first four code points in the |
| 270 |
list above (the ones which have the parentheses in them). |
| 271 |
.PP |
| 272 |
Now that the names have been placed in the Unicode standard, these kinds |
| 273 |
of changes should not happen again, though corrections, such as to |
| 274 |
U+2118, are still possible. |
| 275 |
.PP |
| 276 |
Unicode also added some name abbreviations, which Perl now accepts: |
| 277 |
SP for SPACE; |
| 278 |
TAB for CHARACTER TABULATION; |
| 279 |
NEW LINE, END OF LINE, NL, and EOL for LINE FEED; |
| 280 |
LOCKING-SHIFT ONE for SHIFT OUT; |
| 281 |
LOCKING-SHIFT ZERO for SHIFT IN; |
| 282 |
and ZWNBSP for ZERO WIDTH NO-BREAK SPACE. |
| 283 |
.PP |
| 284 |
More details on this version of Unicode are provided in |
| 285 |
<http://www.unicode.org/versions/Unicode6.1.0/>. |
| 286 |
.PP |
| 287 |
\fR\f(CI\*(C`use charnames\*(C'\fR\fI is no longer needed for \fR\f(CI\*(C`\eN{name}\*(C'\fR\fI\fR |
| 288 |
.IX Subsection "use charnames is no longer needed for N{name}" |
| 289 |
.PP |
| 290 |
When \f(CW\*(C`\eN{\fR\f(CIname\fR\f(CW}\*(C'\fR is encountered, the \f(CW\*(C`charnames\*(C'\fR module is now |
| 291 |
automatically loaded when needed as if the \f(CW\*(C`:full\*(C'\fR and \f(CW\*(C`:short\*(C'\fR |
| 292 |
options had been specified. See charnames for more information. |
| 293 |
.PP |
| 294 |
\fR\f(CI\*(C`\eN{...}\*(C'\fR\fI can now have Unicode loose name matching\fR |
| 295 |
.IX Subsection "N{...} can now have Unicode loose name matching" |
| 296 |
.PP |
| 297 |
This is described in the \f(CW\*(C`charnames\*(C'\fR item in |
| 298 |
"Updated Modules and Pragmata" below. |
| 299 |
.PP |
| 300 |
\fIUnicode Symbol Names\fR |
| 301 |
.IX Subsection "Unicode Symbol Names" |
| 302 |
.PP |
| 303 |
Perl now has proper support for Unicode in symbol names. It used to be |
| 304 |
that \f(CW\*(C`*{$foo}\*(C'\fR would ignore the internal UTF8 flag and use the bytes of |
| 305 |
the underlying representation to look up the symbol. That meant that |
| 306 |
\&\f(CW\*(C`*{"\ex{100}"}\*(C'\fR and \f(CW\*(C`*{"\exc4\ex80"}\*(C'\fR would return the same thing. All |
| 307 |
these parts of Perl have been fixed to account for Unicode: |
| 308 |
.IP \(bu 4 |
| 309 |
Method names (including those passed to \f(CW\*(C`use overload\*(C'\fR) |
| 310 |
.IP \(bu 4 |
| 311 |
Typeglob names (including names of variables, subroutines, and filehandles) |
| 312 |
.IP \(bu 4 |
| 313 |
Package names |
| 314 |
.IP \(bu 4 |
| 315 |
\&\f(CW\*(C`goto\*(C'\fR |
| 316 |
.IP \(bu 4 |
| 317 |
Symbolic dereferencing |
| 318 |
.IP \(bu 4 |
| 319 |
Second argument to \f(CWbless()\fR and \f(CWtie()\fR |
| 320 |
.IP \(bu 4 |
| 321 |
Return value of \f(CWref()\fR |
| 322 |
.IP \(bu 4 |
| 323 |
Subroutine prototypes |
| 324 |
.IP \(bu 4 |
| 325 |
Attributes |
| 326 |
.IP \(bu 4 |
| 327 |
Various warnings and error messages that mention variable names or values, |
| 328 |
methods, etc. |
| 329 |
.PP |
| 330 |
In addition, a parsing bug has been fixed that prevented \f(CW\*(C`*{é}\*(C'\fR from |
| 331 |
implicitly quoting the name, but instead interpreted it as \f(CW\*(C`*{+é}\*(C'\fR, which |
| 332 |
would cause a strict violation. |
| 333 |
.PP |
| 334 |
\&\f(CW\*(C`*{"*a::b"}\*(C'\fR automatically strips off the * if it is followed by an ASCII |
| 335 |
letter. That has been extended to all Unicode identifier characters. |
| 336 |
.PP |
| 337 |
One-character non-ASCII non-punctuation variables (like \f(CW$é\fR) are now |
| 338 |
subject to "Used only once" warnings. They used to be exempt, as they |
| 339 |
were treated as punctuation variables. |
| 340 |
.PP |
| 341 |
Also, single-character Unicode punctuation variables (like \f(CW$‰\fR) are now |
| 342 |
supported [perl #69032]. |
| 343 |
.PP |
| 344 |
\fIImproved ability to mix locales and Unicode, including UTF\-8 locales\fR |
| 345 |
.IX Subsection "Improved ability to mix locales and Unicode, including UTF-8 locales" |
| 346 |
.PP |
| 347 |
An optional parameter has been added to \f(CW\*(C`use locale\*(C'\fR |
| 348 |
.PP |
| 349 |
.Vb 1 |
| 350 |
\& use locale \*(Aq:not_characters\*(Aq; |
| 351 |
.Ve |
| 352 |
.PP |
| 353 |
which tells Perl to use all but the \f(CW\*(C`LC_CTYPE\*(C'\fR and \f(CW\*(C`LC_COLLATE\*(C'\fR |
| 354 |
portions of the current locale. Instead, the character set is assumed |
| 355 |
to be Unicode. This lets locales and Unicode be seamlessly mixed, |
| 356 |
including the increasingly frequent UTF\-8 locales. When using this |
| 357 |
hybrid form of locales, the \f(CW\*(C`:locale\*(C'\fR layer to the open pragma can |
| 358 |
be used to interface with the file system, and there are CPAN modules |
| 359 |
available for ARGV and environment variable conversions. |
| 360 |
.PP |
| 361 |
Full details are in perllocale. |
| 362 |
.PP |
| 363 |
\fINew function \fR\f(CI\*(C`fc\*(C'\fR\fI and corresponding escape sequence \fR\f(CI\*(C`\eF\*(C'\fR\fI for Unicode foldcase\fR |
| 364 |
.IX Subsection "New function fc and corresponding escape sequence F for Unicode foldcase" |
| 365 |
.PP |
| 366 |
Unicode foldcase is an extension to lowercase that gives better results |
| 367 |
when comparing two strings case-insensitively. It has long been used |
| 368 |
internally in regular expression \f(CW\*(C`/i\*(C'\fR matching. Now it is available |
| 369 |
explicitly through the new \f(CW\*(C`fc\*(C'\fR function call (enabled by |
| 370 |
\&\f(CW"use\ feature\ \*(Aqfc\*(Aq"\fR, or \f(CW\*(C`use v5.16\*(C'\fR, or explicitly callable via |
| 371 |
\&\f(CW\*(C`CORE::fc\*(C'\fR) or through the new \f(CW\*(C`\eF\*(C'\fR sequence in double-quotish |
| 372 |
strings. |
| 373 |
.PP |
| 374 |
Full details are in "fc" in perlfunc. |
| 375 |
.PP |
| 376 |
\fIThe Unicode \fR\f(CI\*(C`Script_Extensions\*(C'\fR\fI property is now supported\fR |
| 377 |
.IX Subsection "The Unicode Script_Extensions property is now supported" |
| 378 |
.PP |
| 379 |
New in Unicode 6.0, this is an improved \f(CW\*(C`Script\*(C'\fR property. Details |
| 380 |
are in "Scripts" in perlunicode. |
| 381 |
.SS "XS Changes" |
| 382 |
.IX Subsection "XS Changes" |
| 383 |
\fIImproved typemaps for Some Builtin Types\fR |
| 384 |
.IX Subsection "Improved typemaps for Some Builtin Types" |
| 385 |
.PP |
| 386 |
Most XS authors will know there is a longstanding bug in the |
| 387 |
OUTPUT typemap for T_AVREF (\f(CW\*(C`AV*\*(C'\fR), T_HVREF (\f(CW\*(C`HV*\*(C'\fR), T_CVREF (\f(CW\*(C`CV*\*(C'\fR), |
| 388 |
and T_SVREF (\f(CW\*(C`SVREF\*(C'\fR or \f(CW\*(C`\e$foo\*(C'\fR) that requires manually decrementing |
| 389 |
the reference count of the return value instead of the typemap taking |
| 390 |
care of this. For backwards-compatibility, this cannot be changed in the |
| 391 |
default typemaps. But we now provide additional typemaps |
| 392 |
\&\f(CW\*(C`T_AVREF_REFCOUNT_FIXED\*(C'\fR, etc. that do not exhibit this bug. Using |
| 393 |
them in your extension is as simple as having one line in your |
| 394 |
\&\f(CW\*(C`TYPEMAP\*(C'\fR section: |
| 395 |
.PP |
| 396 |
.Vb 1 |
| 397 |
\& HV* T_HVREF_REFCOUNT_FIXED |
| 398 |
.Ve |
| 399 |
.PP |
| 400 |
\fR\f(CIis_utf8_char()\fR\fI\fR |
| 401 |
.IX Subsection "is_utf8_char()" |
| 402 |
.PP |
| 403 |
The XS-callable function \f(CWis_utf8_char()\fR, when presented with |
| 404 |
malformed UTF\-8 input, can read up to 12 bytes beyond the end of the |
| 405 |
string. This cannot be fixed without changing its API, and so its |
| 406 |
use is now deprecated. Use \f(CWis_utf8_char_buf()\fR (described just below) |
| 407 |
instead. |
| 408 |
.PP |
| 409 |
\fIAdded \fR\f(CIis_utf8_char_buf()\fR |
| 410 |
.IX Subsection "Added is_utf8_char_buf()" |
| 411 |
.PP |
| 412 |
This function is designed to replace the deprecated "\fBis_utf8_char()\fR" |
| 413 |
function. It includes an extra parameter to make sure it doesn't read |
| 414 |
past the end of the input buffer. |
| 415 |
.PP |
| 416 |
\fIOther \fR\f(CIis_utf8_foo()\fR\fI functions, as well as \fR\f(CIutf8_to_foo()\fR\fI, etc.\fR |
| 417 |
.IX Subsection "Other is_utf8_foo() functions, as well as utf8_to_foo(), etc." |
| 418 |
.PP |
| 419 |
Most other XS-callable functions that take UTF\-8 encoded input |
| 420 |
implicitly assume that the UTF\-8 is valid (not malformed) with respect to |
| 421 |
buffer length. Do not do things such as change a character's case or |
| 422 |
see if it is alphanumeric without first being sure that it is valid |
| 423 |
UTF\-8. This can be safely done for a whole string by using one of the |
| 424 |
functions \f(CWis_utf8_string()\fR, \f(CWis_utf8_string_loc()\fR, and |
| 425 |
\&\f(CWis_utf8_string_loclen()\fR. |
| 426 |
.PP |
| 427 |
\fINew Pad API\fR |
| 428 |
.IX Subsection "New Pad API" |
| 429 |
.PP |
| 430 |
Many new functions have been added to the API for manipulating lexical |
| 431 |
pads. See "Pad Data Structures" in perlapi for more information. |
| 432 |
.SS "Changes to Special Variables" |
| 433 |
.IX Subsection "Changes to Special Variables" |
| 434 |
\fR\f(CI$$\fR\fI can be assigned to\fR |
| 435 |
.IX Subsection "$$ can be assigned to" |
| 436 |
.PP |
| 437 |
\&\f(CW$$\fR was made read-only in Perl 5.8.0. But only sometimes: \f(CW\*(C`local $$\*(C'\fR |
| 438 |
would make it writable again. Some CPAN modules were using \f(CW\*(C`local $$\*(C'\fR or |
| 439 |
XS code to bypass the read-only check, so there is no reason to keep \f(CW$$\fR |
| 440 |
read-only. (This change also allowed a bug to be fixed while maintaining |
| 441 |
backward compatibility.) |
| 442 |
.PP |
| 443 |
\fR\f(CI$^X\fR\fI converted to an absolute path on FreeBSD, OS X and Solaris\fR |
| 444 |
.IX Subsection "$^X converted to an absolute path on FreeBSD, OS X and Solaris" |
| 445 |
.PP |
| 446 |
\&\f(CW$^X\fR is now converted to an absolute path on OS X, FreeBSD (without |
| 447 |
needing \fI/proc\fR mounted) and Solaris 10 and 11. This augments the |
| 448 |
previous approach of using \fI/proc\fR on Linux, FreeBSD, and NetBSD |
| 449 |
(in all cases, where mounted). |
| 450 |
.PP |
| 451 |
This makes relocatable perl installations more useful on these platforms. |
| 452 |
(See "Relocatable \f(CW@INC\fR" in \fIINSTALL\fR) |
| 453 |
.SS "Debugger Changes" |
| 454 |
.IX Subsection "Debugger Changes" |
| 455 |
\fIFeatures inside the debugger\fR |
| 456 |
.IX Subsection "Features inside the debugger" |
| 457 |
.PP |
| 458 |
The current Perl's feature bundle is now enabled for commands entered |
| 459 |
in the interactive debugger. |
| 460 |
.PP |
| 461 |
\fINew option for the debugger's \fR\f(BIt\fR\fI command\fR |
| 462 |
.IX Subsection "New option for the debugger's t command" |
| 463 |
.PP |
| 464 |
The \fBt\fR command in the debugger, which toggles tracing mode, now |
| 465 |
accepts a numeric argument that determines how many levels of subroutine |
| 466 |
calls to trace. |
| 467 |
.PP |
| 468 |
\fR\f(CI\*(C`enable\*(C'\fR\fI and \fR\f(CI\*(C`disable\*(C'\fR\fI\fR |
| 469 |
.IX Subsection "enable and disable" |
| 470 |
.PP |
| 471 |
The debugger now has \f(CW\*(C`disable\*(C'\fR and \f(CW\*(C`enable\*(C'\fR commands for disabling |
| 472 |
existing breakpoints and re-enabling them. See perldebug. |
| 473 |
.PP |
| 474 |
\fIBreakpoints with file names\fR |
| 475 |
.IX Subsection "Breakpoints with file names" |
| 476 |
.PP |
| 477 |
The debugger's "b" command for setting breakpoints now lets a line |
| 478 |
number be prefixed with a file name. See |
| 479 |
"b [file]:[line] [condition]" in perldebug. |
| 480 |
.ie n .SS "The ""CORE"" Namespace" |
| 481 |
.el .SS "The \f(CWCORE\fP Namespace" |
| 482 |
.IX Subsection "The CORE Namespace" |
| 483 |
\fIThe \fR\f(CI\*(C`CORE::\*(C'\fR\fI prefix\fR |
| 484 |
.IX Subsection "The CORE:: prefix" |
| 485 |
.PP |
| 486 |
The \f(CW\*(C`CORE::\*(C'\fR prefix can now be used on keywords enabled by |
| 487 |
feature.pm, even outside the scope of \f(CW\*(C`use feature\*(C'\fR. |
| 488 |
.PP |
| 489 |
\fISubroutines in the \fR\f(CI\*(C`CORE\*(C'\fR\fI namespace\fR |
| 490 |
.IX Subsection "Subroutines in the CORE namespace" |
| 491 |
.PP |
| 492 |
Many Perl keywords are now available as subroutines in the CORE namespace. |
| 493 |
This lets them be aliased: |
| 494 |
.PP |
| 495 |
.Vb 2 |
| 496 |
\& BEGIN { *entangle = \e&CORE::tie } |
| 497 |
\& entangle $variable, $package, @args; |
| 498 |
.Ve |
| 499 |
.PP |
| 500 |
And for prototypes to be bypassed: |
| 501 |
.PP |
| 502 |
.Vb 5 |
| 503 |
\& sub mytie(\e[%$*@]$@) { |
| 504 |
\& my ($ref, $pack, @args) = @_; |
| 505 |
\& ... do something ... |
| 506 |
\& goto &CORE::tie; |
| 507 |
\& } |
| 508 |
.Ve |
| 509 |
.PP |
| 510 |
Some of these cannot be called through references or via \f(CW&foo\fR syntax, |
| 511 |
but must be called as barewords. |
| 512 |
.PP |
| 513 |
See CORE for details. |
| 514 |
.SS "Other Changes" |
| 515 |
.IX Subsection "Other Changes" |
| 516 |
\fIAnonymous handles\fR |
| 517 |
.IX Subsection "Anonymous handles" |
| 518 |
.PP |
| 519 |
Automatically generated file handles are now named _\|_ANONIO_\|_ when the |
| 520 |
variable name cannot be determined, rather than \f(CW$_\fR\|_ANONIO_\|_. |
| 521 |
.PP |
| 522 |
\fIAutoloaded sort Subroutines\fR |
| 523 |
.IX Subsection "Autoloaded sort Subroutines" |
| 524 |
.PP |
| 525 |
Custom sort subroutines can now be autoloaded [perl #30661]: |
| 526 |
.PP |
| 527 |
.Vb 2 |
| 528 |
\& sub AUTOLOAD { ... } |
| 529 |
\& @sorted = sort foo @list; # uses AUTOLOAD |
| 530 |
.Ve |
| 531 |
.PP |
| 532 |
\fR\f(CI\*(C`continue\*(C'\fR\fI no longer requires the "switch" feature\fR |
| 533 |
.IX Subsection "continue no longer requires the ""switch"" feature" |
| 534 |
.PP |
| 535 |
The \f(CW\*(C`continue\*(C'\fR keyword has two meanings. It can introduce a \f(CW\*(C`continue\*(C'\fR |
| 536 |
block after a loop, or it can exit the current \f(CW\*(C`when\*(C'\fR block. Up to now, |
| 537 |
the latter meaning was valid only with the "switch" feature enabled, and |
| 538 |
was a syntax error otherwise. Since the main purpose of feature.pm is to |
| 539 |
avoid conflicts with user-defined subroutines, there is no reason for |
| 540 |
\&\f(CW\*(C`continue\*(C'\fR to depend on it. |
| 541 |
.PP |
| 542 |
\fIDTrace probes for interpreter phase change\fR |
| 543 |
.IX Subsection "DTrace probes for interpreter phase change" |
| 544 |
.PP |
| 545 |
The \f(CW\*(C`phase\-change\*(C'\fR probes will fire when the interpreter's phase |
| 546 |
changes, which tracks the \f(CW\*(C`${^GLOBAL_PHASE}\*(C'\fR variable. \f(CW\*(C`arg0\*(C'\fR is |
| 547 |
the new phase name; \f(CW\*(C`arg1\*(C'\fR is the old one. This is useful |
| 548 |
for limiting your instrumentation to one or more of: compile time, |
| 549 |
run time, or destruct time. |
| 550 |
.PP |
| 551 |
\fR\f(CI\*(C`_\|_FILE_\|_()\*(C'\fR\fI Syntax\fR |
| 552 |
.IX Subsection "__FILE__() Syntax" |
| 553 |
.PP |
| 554 |
The \f(CW\*(C`_\|_FILE_\|_\*(C'\fR, \f(CW\*(C`_\|_LINE_\|_\*(C'\fR and \f(CW\*(C`_\|_PACKAGE_\|_\*(C'\fR tokens can now be written |
| 555 |
with an empty pair of parentheses after them. This makes them parse the |
| 556 |
same way as \f(CW\*(C`time\*(C'\fR, \f(CW\*(C`fork\*(C'\fR and other built-in functions. |
| 557 |
.PP |
| 558 |
\fIThe \fR\f(CI\*(C`\e$\*(C'\fR\fI prototype accepts any scalar lvalue\fR |
| 559 |
.IX Subsection "The $ prototype accepts any scalar lvalue" |
| 560 |
.PP |
| 561 |
The \f(CW\*(C`\e$\*(C'\fR and \f(CW\*(C`\e[$]\*(C'\fR subroutine prototypes now accept any scalar lvalue |
| 562 |
argument. Previously they accepted only scalars beginning with \f(CW\*(C`$\*(C'\fR and |
| 563 |
hash and array elements. This change makes them consistent with the way |
| 564 |
the built-in \f(CW\*(C`read\*(C'\fR and \f(CW\*(C`recv\*(C'\fR functions (among others) parse their |
| 565 |
arguments. This means that one can override the built-in functions with |
| 566 |
custom subroutines that parse their arguments the same way. |
| 567 |
.PP |
| 568 |
\fR\f(CI\*(C`_\*(C'\fR\fI in subroutine prototypes\fR |
| 569 |
.IX Subsection "_ in subroutine prototypes" |
| 570 |
.PP |
| 571 |
The \f(CW\*(C`_\*(C'\fR character in subroutine prototypes is now allowed before \f(CW\*(C`@\*(C'\fR or |
| 572 |
\&\f(CW\*(C`%\*(C'\fR. |
| 573 |
.SH Security |
| 574 |
.IX Header "Security" |
| 575 |
.ie n .SS "Use is_utf8_char_buf() and not is_utf8_char()" |
| 576 |
.el .SS "Use \f(CWis_utf8_char_buf()\fP and not \f(CWis_utf8_char()\fP" |
| 577 |
.IX Subsection "Use is_utf8_char_buf() and not is_utf8_char()" |
| 578 |
The latter function is now deprecated because its API is insufficient to |
| 579 |
guarantee that it doesn't read (up to 12 bytes in the worst case) beyond |
| 580 |
the end of its input string. See |
| 581 |
\&\fBis_utf8_char_buf()\fR. |
| 582 |
.SS "Malformed UTF\-8 input could cause attempts to read beyond the end of the buffer" |
| 583 |
.IX Subsection "Malformed UTF-8 input could cause attempts to read beyond the end of the buffer" |
| 584 |
Two new XS-accessible functions, \f(CWutf8_to_uvchr_buf()\fR and |
| 585 |
\&\f(CWutf8_to_uvuni_buf()\fR are now available to prevent this, and the Perl |
| 586 |
core has been converted to use them. |
| 587 |
See "Internal Changes". |
| 588 |
.ie n .SS "File::Glob::bsd_glob() memory error with GLOB_ALTDIRFUNC (CVE\-2011\-2728)" |
| 589 |
.el .SS "\f(CWFile::Glob::bsd_glob()\fP memory error with GLOB_ALTDIRFUNC (CVE\-2011\-2728)" |
| 590 |
.IX Subsection "File::Glob::bsd_glob() memory error with GLOB_ALTDIRFUNC (CVE-2011-2728)" |
| 591 |
Calling \f(CW\*(C`File::Glob::bsd_glob\*(C'\fR with the unsupported flag |
| 592 |
GLOB_ALTDIRFUNC would cause an access violation / segfault. A Perl |
| 593 |
program that accepts a flags value from an external source could expose |
| 594 |
itself to denial of service or arbitrary code execution attacks. There |
| 595 |
are no known exploits in the wild. The problem has been corrected by |
| 596 |
explicitly disabling all unsupported flags and setting unused function |
| 597 |
pointers to null. Bug reported by Clément Lecigne. (5.14.2) |
| 598 |
.ie n .SS "Privileges are now set correctly when assigning to $(" |
| 599 |
.el .SS "Privileges are now set correctly when assigning to \f(CW$(\fP" |
| 600 |
.IX Subsection "Privileges are now set correctly when assigning to $(" |
| 601 |
A hypothetical bug (probably unexploitable in practice) because the |
| 602 |
incorrect setting of the effective group ID while setting \f(CW$(\fR has been |
| 603 |
fixed. The bug would have affected only systems that have \f(CWsetresgid()\fR |
| 604 |
but not \f(CWsetregid()\fR, but no such systems are known to exist. |
| 605 |
.SH Deprecations |
| 606 |
.IX Header "Deprecations" |
| 607 |
.SS "Don't read the Unicode data base files in \fIlib/unicore\fP" |
| 608 |
.IX Subsection "Don't read the Unicode data base files in lib/unicore" |
| 609 |
It is now deprecated to directly read the Unicode data base files. |
| 610 |
These are stored in the \fIlib/unicore\fR directory. Instead, you should |
| 611 |
use the new functions in Unicode::UCD. These provide a stable API, |
| 612 |
and give complete information. |
| 613 |
.PP |
| 614 |
Perl may at some point in the future change or remove these files. The |
| 615 |
file which applications were most likely to have used is |
| 616 |
\&\fIlib/unicore/ToDigit.pl\fR. "\fBprop_invmap()\fR" in Unicode::UCD can be used to |
| 617 |
get at its data instead. |
| 618 |
.ie n .SS "XS functions is_utf8_char(), utf8_to_uvchr() and utf8_to_uvuni()" |
| 619 |
.el .SS "XS functions \f(CWis_utf8_char()\fP, \f(CWutf8_to_uvchr()\fP and \f(CWutf8_to_uvuni()\fP" |
| 620 |
.IX Subsection "XS functions is_utf8_char(), utf8_to_uvchr() and utf8_to_uvuni()" |
| 621 |
This function is deprecated because it could read beyond the end of the |
| 622 |
input string. Use the new \fBis_utf8_char_buf()\fR, |
| 623 |
\&\f(CWutf8_to_uvchr_buf()\fR and \f(CWutf8_to_uvuni_buf()\fR instead. |
| 624 |
.SH "Future Deprecations" |
| 625 |
.IX Header "Future Deprecations" |
| 626 |
This section serves as a notice of features that are \fIlikely\fR to be |
| 627 |
removed or deprecated in the next release of |
| 628 |
perl (5.18.0). If your code depends on these features, you should |
| 629 |
contact the Perl 5 Porters via the mailing |
| 630 |
list <http://lists.perl.org/list/perl5-porters.html> or perlbug to |
| 631 |
explain your use case and inform the deprecation process. |
| 632 |
.SS "Core Modules" |
| 633 |
.IX Subsection "Core Modules" |
| 634 |
These modules may be marked as deprecated \fIfrom the core\fR. This only |
| 635 |
means that they will no longer be installed by default with the core |
| 636 |
distribution, but will remain available on the CPAN. |
| 637 |
.IP \(bu 4 |
| 638 |
CPANPLUS |
| 639 |
.IP \(bu 4 |
| 640 |
Filter::Simple |
| 641 |
.IP \(bu 4 |
| 642 |
PerlIO::mmap |
| 643 |
.IP \(bu 4 |
| 644 |
Pod::LaTeX |
| 645 |
.IP \(bu 4 |
| 646 |
Pod::Parser |
| 647 |
.IP \(bu 4 |
| 648 |
SelfLoader |
| 649 |
.IP \(bu 4 |
| 650 |
Text::Soundex |
| 651 |
.IP \(bu 4 |
| 652 |
Thread.pm |
| 653 |
.SS "Platforms with no supporting programmers" |
| 654 |
.IX Subsection "Platforms with no supporting programmers" |
| 655 |
These platforms will probably have their |
| 656 |
special build support removed during the |
| 657 |
5.17.0 development series. |
| 658 |
.IP \(bu 4 |
| 659 |
BeOS |
| 660 |
.IP \(bu 4 |
| 661 |
djgpp |
| 662 |
.IP \(bu 4 |
| 663 |
dgux |
| 664 |
.IP \(bu 4 |
| 665 |
EPOC |
| 666 |
.IP \(bu 4 |
| 667 |
MPE/iX |
| 668 |
.IP \(bu 4 |
| 669 |
Rhapsody |
| 670 |
.IP \(bu 4 |
| 671 |
UTS |
| 672 |
.IP \(bu 4 |
| 673 |
VM/ESA |
| 674 |
.SS "Other Future Deprecations" |
| 675 |
.IX Subsection "Other Future Deprecations" |
| 676 |
.IP \(bu 4 |
| 677 |
Swapping of $< and $> |
| 678 |
.Sp |
| 679 |
For more information about this future deprecation, see the relevant RT |
| 680 |
ticket <https://github.com/Perl/perl5/issues/11547>. |
| 681 |
.IP \(bu 4 |
| 682 |
sfio, stdio |
| 683 |
.Sp |
| 684 |
Perl supports being built without PerlIO proper, using a stdio or sfio |
| 685 |
wrapper instead. A perl build like this will not support IO layers and |
| 686 |
thus Unicode IO, making it rather handicapped. |
| 687 |
.Sp |
| 688 |
PerlIO supports a \f(CW\*(C`stdio\*(C'\fR layer if stdio use is desired, and similarly a |
| 689 |
sfio layer could be produced. |
| 690 |
.IP \(bu 4 |
| 691 |
Unescaped literal \f(CW"{"\fR in regular expressions. |
| 692 |
.Sp |
| 693 |
Starting with v5.20, it is planned to require a literal \f(CW"{"\fR to be |
| 694 |
escaped, for example by preceding it with a backslash. In v5.18, a |
| 695 |
deprecated warning message will be emitted for all such uses. |
| 696 |
This affects only patterns that are to match a literal \f(CW"{"\fR. Other |
| 697 |
uses of this character, such as part of a quantifier or sequence as in |
| 698 |
those below, are completely unaffected: |
| 699 |
.Sp |
| 700 |
.Vb 3 |
| 701 |
\& /foo{3,5}/ |
| 702 |
\& /\ep{Alphabetic}/ |
| 703 |
\& /\eN{DIGIT ZERO} |
| 704 |
.Ve |
| 705 |
.Sp |
| 706 |
Removing this will permit extensions to Perl's pattern syntax and better |
| 707 |
error checking for existing syntax. See "Quantifiers" in perlre for an |
| 708 |
example. |
| 709 |
.IP \(bu 4 |
| 710 |
Revamping \f(CW"\eQ"\fR semantics in double-quotish strings when combined with other escapes. |
| 711 |
.Sp |
| 712 |
There are several bugs and inconsistencies involving combinations |
| 713 |
of \f(CW\*(C`\eQ\*(C'\fR and escapes like \f(CW\*(C`\ex\*(C'\fR, \f(CW\*(C`\eL\*(C'\fR, etc., within a \f(CW\*(C`\eQ...\eE\*(C'\fR pair. |
| 714 |
These need to be fixed, and doing so will necessarily change current |
| 715 |
behavior. The changes have not yet been settled. |
| 716 |
.SH "Incompatible Changes" |
| 717 |
.IX Header "Incompatible Changes" |
| 718 |
.SS "Special blocks called in void context" |
| 719 |
.IX Subsection "Special blocks called in void context" |
| 720 |
Special blocks (\f(CW\*(C`BEGIN\*(C'\fR, \f(CW\*(C`CHECK\*(C'\fR, \f(CW\*(C`INIT\*(C'\fR, \f(CW\*(C`UNITCHECK\*(C'\fR, \f(CW\*(C`END\*(C'\fR) are now |
| 721 |
called in void context. This avoids wasteful copying of the result of the |
| 722 |
last statement [perl #108794]. |
| 723 |
.ie n .SS "The ""overloading"" pragma and regexp objects" |
| 724 |
.el .SS "The \f(CWoverloading\fP pragma and regexp objects" |
| 725 |
.IX Subsection "The overloading pragma and regexp objects" |
| 726 |
With \f(CW\*(C`no overloading\*(C'\fR, regular expression objects returned by \f(CW\*(C`qr//\*(C'\fR are |
| 727 |
now stringified as "Regexp=REGEXP(0xbe600d)" instead of the regular |
| 728 |
expression itself [perl #108780]. |
| 729 |
.SS "Two XS typemap Entries removed" |
| 730 |
.IX Subsection "Two XS typemap Entries removed" |
| 731 |
Two presumably unused XS typemap entries have been removed from the |
| 732 |
core typemap: T_DATAUNIT and T_CALLBACK. If you are, against all odds, |
| 733 |
a user of these, please see the instructions on how to restore them |
| 734 |
in perlxstypemap. |
| 735 |
.SS "Unicode 6.1 has incompatibilities with Unicode 6.0" |
| 736 |
.IX Subsection "Unicode 6.1 has incompatibilities with Unicode 6.0" |
| 737 |
These are detailed in "Supports (almost) Unicode 6.1" above. |
| 738 |
You can compile this version of Perl to use Unicode 6.0. See |
| 739 |
"Hacking Perl to work on earlier Unicode versions (for very serious hackers only)" in perlunicode. |
| 740 |
.SS "Borland compiler" |
| 741 |
.IX Subsection "Borland compiler" |
| 742 |
All support for the Borland compiler has been dropped. The code had not |
| 743 |
worked for a long time anyway. |
| 744 |
.SS "Certain deprecated Unicode properties are no longer supported by default" |
| 745 |
.IX Subsection "Certain deprecated Unicode properties are no longer supported by default" |
| 746 |
Perl should never have exposed certain Unicode properties that are used |
| 747 |
by Unicode internally and not meant to be publicly available. Use of |
| 748 |
these has generated deprecated warning messages since Perl 5.12. The |
| 749 |
removed properties are Other_Alphabetic, |
| 750 |
Other_Default_Ignorable_Code_Point, Other_Grapheme_Extend, |
| 751 |
Other_ID_Continue, Other_ID_Start, Other_Lowercase, Other_Math, and |
| 752 |
Other_Uppercase. |
| 753 |
.PP |
| 754 |
Perl may be recompiled to include any or all of them; instructions are |
| 755 |
given in |
| 756 |
"Unicode character properties that are NOT accepted by Perl" in perluniprops. |
| 757 |
.SS "Dereferencing IO thingies as typeglobs" |
| 758 |
.IX Subsection "Dereferencing IO thingies as typeglobs" |
| 759 |
The \f(CW\*(C`*{...}\*(C'\fR operator, when passed a reference to an IO thingy (as in |
| 760 |
\&\f(CW\*(C`*{*STDIN{IO}}\*(C'\fR), creates a new typeglob containing just that IO object. |
| 761 |
Previously, it would stringify as an empty string, but some operators would |
| 762 |
treat it as undefined, producing an "uninitialized" warning. |
| 763 |
Now it stringifies as _\|_ANONIO_\|_ [perl #96326]. |
| 764 |
.SS "User-defined case-changing operations" |
| 765 |
.IX Subsection "User-defined case-changing operations" |
| 766 |
This feature was deprecated in Perl 5.14, and has now been removed. |
| 767 |
The CPAN module Unicode::Casing provides better functionality without |
| 768 |
the drawbacks that this feature had, as are detailed in the 5.14 |
| 769 |
documentation: |
| 770 |
<http://perldoc.perl.org/5.14.0/perlunicode.html#User\-Defined\-Case\-Mappings\-%28for\-serious\-hackers\-only%29> |
| 771 |
.SS "XSUBs are now 'static'" |
| 772 |
.IX Subsection "XSUBs are now 'static'" |
| 773 |
XSUB C functions are now 'static', that is, they are not visible from |
| 774 |
outside the compilation unit. Users can use the new \f(CWXS_EXTERNAL(name)\fR |
| 775 |
and \f(CWXS_INTERNAL(name)\fR macros to pick the desired linking behavior. |
| 776 |
The ordinary \f(CWXS(name)\fR declaration for XSUBs will continue to declare |
| 777 |
non\-'static' XSUBs for compatibility, but the XS compiler, |
| 778 |
ExtUtils::ParseXS (\f(CW\*(C`xsubpp\*(C'\fR) will emit 'static' XSUBs by default. |
| 779 |
ExtUtils::ParseXS's behavior can be reconfigured from XS using the |
| 780 |
\&\f(CW\*(C`EXPORT_XSUB_SYMBOLS\*(C'\fR keyword. See perlxs for details. |
| 781 |
.SS "Weakening read-only references" |
| 782 |
.IX Subsection "Weakening read-only references" |
| 783 |
Weakening read-only references is no longer permitted. It should never |
| 784 |
have worked anyway, and could sometimes result in crashes. |
| 785 |
.SS "Tying scalars that hold typeglobs" |
| 786 |
.IX Subsection "Tying scalars that hold typeglobs" |
| 787 |
Attempting to tie a scalar after a typeglob was assigned to it would |
| 788 |
instead tie the handle in the typeglob's IO slot. This meant that it was |
| 789 |
impossible to tie the scalar itself. Similar problems affected \f(CW\*(C`tied\*(C'\fR and |
| 790 |
\&\f(CW\*(C`untie\*(C'\fR: \f(CW\*(C`tied $scalar\*(C'\fR would return false on a tied scalar if the last |
| 791 |
thing returned was a typeglob, and \f(CW\*(C`untie $scalar\*(C'\fR on such a tied scalar |
| 792 |
would do nothing. |
| 793 |
.PP |
| 794 |
We fixed this problem before Perl 5.14.0, but it caused problems with some |
| 795 |
CPAN modules, so we put in a deprecation cycle instead. |
| 796 |
.PP |
| 797 |
Now the deprecation has been removed and this bug has been fixed. So |
| 798 |
\&\f(CW\*(C`tie $scalar\*(C'\fR will always tie the scalar, not the handle it holds. To tie |
| 799 |
the handle, use \f(CW\*(C`tie *$scalar\*(C'\fR (with an explicit asterisk). The same |
| 800 |
applies to \f(CW\*(C`tied *$scalar\*(C'\fR and \f(CW\*(C`untie *$scalar\*(C'\fR. |
| 801 |
.ie n .SS "IPC::Open3 no longer provides xfork(), xclose_on_exec() and xpipe_anon()" |
| 802 |
.el .SS "IPC::Open3 no longer provides \f(CWxfork()\fP, \f(CWxclose_on_exec()\fP and \f(CWxpipe_anon()\fP" |
| 803 |
.IX Subsection "IPC::Open3 no longer provides xfork(), xclose_on_exec() and xpipe_anon()" |
| 804 |
All three functions were private, undocumented, and unexported. They do |
| 805 |
not appear to be used by any code on CPAN. Two have been inlined and one |
| 806 |
deleted entirely. |
| 807 |
.ie n .SS "$$ no longer caches PID" |
| 808 |
.el .SS "\f(CW$$\fP no longer caches PID" |
| 809 |
.IX Subsection "$$ no longer caches PID" |
| 810 |
Previously, if one called \fBfork\fR\|(3) from C, Perl's |
| 811 |
notion of \f(CW$$\fR could go out of sync with what \fBgetpid()\fR returns. By always |
| 812 |
fetching the value of \f(CW$$\fR via \fBgetpid()\fR, this potential bug is eliminated. |
| 813 |
Code that depends on the caching behavior will break. As described in |
| 814 |
Core Enhancements, |
| 815 |
\&\f(CW$$\fR is now writable, but it will be reset during a |
| 816 |
fork. |
| 817 |
.ie n .SS "$$ and getppid() no longer emulate POSIX semantics under LinuxThreads" |
| 818 |
.el .SS "\f(CW$$\fP and \f(CWgetppid()\fP no longer emulate POSIX semantics under LinuxThreads" |
| 819 |
.IX Subsection "$$ and getppid() no longer emulate POSIX semantics under LinuxThreads" |
| 820 |
The POSIX emulation of \f(CW$$\fR and \f(CWgetppid()\fR under the obsolete |
| 821 |
LinuxThreads implementation has been removed. |
| 822 |
This only impacts users of Linux 2.4 and |
| 823 |
users of Debian GNU/kFreeBSD up to and including 6.0, not the vast |
| 824 |
majority of Linux installations that use NPTL threads. |
| 825 |
.PP |
| 826 |
This means that \f(CWgetppid()\fR, like \f(CW$$\fR, is now always guaranteed to |
| 827 |
return the OS's idea of the current state of the process, not perl's |
| 828 |
cached version of it. |
| 829 |
.PP |
| 830 |
See the documentation for $$ for details. |
| 831 |
.ie n .SS "$<, $>, $( and $) are no longer cached" |
| 832 |
.el .SS "\f(CW$<\fP, \f(CW$>\fP, \f(CW$(\fP and \f(CW$)\fP are no longer cached" |
| 833 |
.IX Subsection "$<, $>, $( and $) are no longer cached" |
| 834 |
Similarly to the changes to \f(CW$$\fR and \f(CWgetppid()\fR, the internal |
| 835 |
caching of \f(CW$<\fR, \f(CW$>\fR, \f(CW$(\fR and \f(CW$)\fR has been removed. |
| 836 |
.PP |
| 837 |
When we cached these values our idea of what they were would drift out |
| 838 |
of sync with reality if someone (e.g., someone embedding perl) called |
| 839 |
\&\f(CW\*(C`sete?[ug]id()\*(C'\fR without updating \f(CW\*(C`PL_e?[ug]id\*(C'\fR. Having to deal with |
| 840 |
this complexity wasn't worth it given how cheap the \f(CW\*(C`gete?[ug]id()\*(C'\fR |
| 841 |
system call is. |
| 842 |
.PP |
| 843 |
This change will break a handful of CPAN modules that use the XS-level |
| 844 |
\&\f(CW\*(C`PL_uid\*(C'\fR, \f(CW\*(C`PL_gid\*(C'\fR, \f(CW\*(C`PL_euid\*(C'\fR or \f(CW\*(C`PL_egid\*(C'\fR variables. |
| 845 |
.PP |
| 846 |
The fix for those breakages is to use \f(CW\*(C`PerlProc_gete?[ug]id()\*(C'\fR to |
| 847 |
retrieve them (e.g., \f(CWPerlProc_getuid()\fR), and not to assign to |
| 848 |
\&\f(CW\*(C`PL_e?[ug]id\*(C'\fR if you change the UID/GID/EUID/EGID. There is no longer |
| 849 |
any need to do so since perl will always retrieve the up-to-date |
| 850 |
version of those values from the OS. |
| 851 |
.ie n .SS "Which Non-ASCII characters get quoted by ""quotemeta"" and ""\eQ"" has changed" |
| 852 |
.el .SS "Which Non-ASCII characters get quoted by \f(CWquotemeta\fP and \f(CW\eQ\fP has changed" |
| 853 |
.IX Subsection "Which Non-ASCII characters get quoted by quotemeta and Q has changed" |
| 854 |
This is unlikely to result in a real problem, as Perl does not attach |
| 855 |
special meaning to any non-ASCII character, so it is currently |
| 856 |
irrelevant which are quoted or not. This change fixes bug [perl #77654] and |
| 857 |
brings Perl's behavior more into line with Unicode's recommendations. |
| 858 |
See "quotemeta" in perlfunc. |
| 859 |
.SH "Performance Enhancements" |
| 860 |
.IX Header "Performance Enhancements" |
| 861 |
.IP \(bu 4 |
| 862 |
Improved performance for Unicode properties in regular expressions |
| 863 |
.Sp |
| 864 |
Matching a code point against a Unicode property is now done via a |
| 865 |
binary search instead of linear. This means for example that the worst |
| 866 |
case for a 1000 item property is 10 probes instead of 1000. This |
| 867 |
inefficiency has been compensated for in the past by permanently storing |
| 868 |
in a hash the results of a given probe plus the results for the adjacent |
| 869 |
64 code points, under the theory that near-by code points are likely to |
| 870 |
be searched for. A separate hash was used for each mention of a Unicode |
| 871 |
property in each regular expression. Thus, \f(CW\*(C`qr/\ep{foo}abc\ep{foo}/\*(C'\fR |
| 872 |
would generate two hashes. Any probes in one instance would be unknown |
| 873 |
to the other, and the hashes could expand separately to be quite large |
| 874 |
if the regular expression were used on many different widely-separated |
| 875 |
code points. |
| 876 |
Now, however, there is just one hash shared by all instances of a given |
| 877 |
property. This means that if \f(CW\*(C`\ep{foo}\*(C'\fR is matched against "A" in one |
| 878 |
regular expression in a thread, the result will be known immediately to |
| 879 |
all regular expressions, and the relentless march of using up memory is |
| 880 |
slowed considerably. |
| 881 |
.IP \(bu 4 |
| 882 |
Version declarations with the \f(CW\*(C`use\*(C'\fR keyword (e.g., \f(CW\*(C`use 5.012\*(C'\fR) are now |
| 883 |
faster, as they enable features without loading \fIfeature.pm\fR. |
| 884 |
.IP \(bu 4 |
| 885 |
\&\f(CW\*(C`local $_\*(C'\fR is faster now, as it no longer iterates through magic that it |
| 886 |
is not going to copy anyway. |
| 887 |
.IP \(bu 4 |
| 888 |
Perl 5.12.0 sped up the destruction of objects whose classes define |
| 889 |
empty \f(CW\*(C`DESTROY\*(C'\fR methods (to prevent autoloading), by simply not |
| 890 |
calling such empty methods. This release takes this optimization a |
| 891 |
step further, by not calling any \f(CW\*(C`DESTROY\*(C'\fR method that begins with a |
| 892 |
\&\f(CW\*(C`return\*(C'\fR statement. This can be useful for destructors that are only |
| 893 |
used for debugging: |
| 894 |
.Sp |
| 895 |
.Vb 2 |
| 896 |
\& use constant DEBUG => 1; |
| 897 |
\& sub DESTROY { return unless DEBUG; ... } |
| 898 |
.Ve |
| 899 |
.Sp |
| 900 |
Constant-folding will reduce the first statement to \f(CW\*(C`return;\*(C'\fR if DEBUG |
| 901 |
is set to 0, triggering this optimization. |
| 902 |
.IP \(bu 4 |
| 903 |
Assigning to a variable that holds a typeglob or copy-on-write scalar |
| 904 |
is now much faster. Previously the typeglob would be stringified or |
| 905 |
the copy-on-write scalar would be copied before being clobbered. |
| 906 |
.IP \(bu 4 |
| 907 |
Assignment to \f(CW\*(C`substr\*(C'\fR in void context is now more than twice its |
| 908 |
previous speed. Instead of creating and returning a special lvalue |
| 909 |
scalar that is then assigned to, \f(CW\*(C`substr\*(C'\fR modifies the original string |
| 910 |
itself. |
| 911 |
.IP \(bu 4 |
| 912 |
\&\f(CW\*(C`substr\*(C'\fR no longer calculates a value to return when called in void |
| 913 |
context. |
| 914 |
.IP \(bu 4 |
| 915 |
Due to changes in File::Glob, Perl's \f(CW\*(C`glob\*(C'\fR function and its \f(CW\*(C`<...>\*(C'\fR equivalent are now much faster. The splitting of the pattern |
| 916 |
into words has been rewritten in C, resulting in speed-ups of 20% for |
| 917 |
some cases. |
| 918 |
.Sp |
| 919 |
This does not affect \f(CW\*(C`glob\*(C'\fR on VMS, as it does not use File::Glob. |
| 920 |
.IP \(bu 4 |
| 921 |
The short-circuiting operators \f(CW\*(C`&&\*(C'\fR, \f(CW\*(C`||\*(C'\fR, and \f(CW\*(C`//\*(C'\fR, when chained |
| 922 |
(such as \f(CW\*(C`$a || $b || $c\*(C'\fR), are now considerably faster to short-circuit, |
| 923 |
due to reduced optree traversal. |
| 924 |
.IP \(bu 4 |
| 925 |
The implementation of \f(CW\*(C`s///r\*(C'\fR makes one fewer copy of the scalar's value. |
| 926 |
.IP \(bu 4 |
| 927 |
Recursive calls to lvalue subroutines in lvalue scalar context use less |
| 928 |
memory. |
| 929 |
.SH "Modules and Pragmata" |
| 930 |
.IX Header "Modules and Pragmata" |
| 931 |
.SS "Deprecated Modules" |
| 932 |
.IX Subsection "Deprecated Modules" |
| 933 |
.IP Version::Requirements 4 |
| 934 |
.IX Item "Version::Requirements" |
| 935 |
Version::Requirements is now DEPRECATED, use CPAN::Meta::Requirements, |
| 936 |
which is a drop-in replacement. It will be deleted from perl.git blead |
| 937 |
in v5.17.0. |
| 938 |
.SS "New Modules and Pragmata" |
| 939 |
.IX Subsection "New Modules and Pragmata" |
| 940 |
.IP \(bu 4 |
| 941 |
arybase \-\- this new module implements the \f(CW$[\fR variable. |
| 942 |
.IP \(bu 4 |
| 943 |
PerlIO::mmap 0.010 has been added to the Perl core. |
| 944 |
.Sp |
| 945 |
The \f(CW\*(C`mmap\*(C'\fR PerlIO layer is no longer implemented by perl itself, but has |
| 946 |
been moved out into the new PerlIO::mmap module. |
| 947 |
.SS "Updated Modules and Pragmata" |
| 948 |
.IX Subsection "Updated Modules and Pragmata" |
| 949 |
This is only an overview of selected module updates. For a complete list of |
| 950 |
updates, run: |
| 951 |
.PP |
| 952 |
.Vb 1 |
| 953 |
\& $ corelist \-\-diff 5.14.0 5.16.0 |
| 954 |
.Ve |
| 955 |
.PP |
| 956 |
You can substitute your favorite version in place of 5.14.0, too. |
| 957 |
.IP \(bu 4 |
| 958 |
Archive::Extract has been upgraded from version 0.48 to 0.58. |
| 959 |
.Sp |
| 960 |
Includes a fix for FreeBSD to only use \f(CW\*(C`unzip\*(C'\fR if it is located in |
| 961 |
\&\f(CW\*(C`/usr/local/bin\*(C'\fR, as FreeBSD 9.0 will ship with a limited \f(CW\*(C`unzip\*(C'\fR in |
| 962 |
\&\f(CW\*(C`/usr/bin\*(C'\fR. |
| 963 |
.IP \(bu 4 |
| 964 |
Archive::Tar has been upgraded from version 1.76 to 1.82. |
| 965 |
.Sp |
| 966 |
Adjustments to handle files >8gb (>0777777777777 octal) and a feature |
| 967 |
to return the MD5SUM of files in the archive. |
| 968 |
.IP \(bu 4 |
| 969 |
base has been upgraded from version 2.16 to 2.18. |
| 970 |
.Sp |
| 971 |
\&\f(CW\*(C`base\*(C'\fR no longer sets a module's \f(CW$VERSION\fR to "\-1" when a module it |
| 972 |
loads does not define a \f(CW$VERSION\fR. This change has been made because |
| 973 |
"\-1" is not a valid version number under the new "lax" criteria used |
| 974 |
internally by \f(CW\*(C`UNIVERSAL::VERSION\*(C'\fR. (See version for more on "lax" |
| 975 |
version criteria.) |
| 976 |
.Sp |
| 977 |
\&\f(CW\*(C`base\*(C'\fR no longer internally skips loading modules it has already loaded |
| 978 |
and instead relies on \f(CW\*(C`require\*(C'\fR to inspect \f(CW%INC\fR. This fixes a bug |
| 979 |
when \f(CW\*(C`base\*(C'\fR is used with code that clear \f(CW%INC\fR to force a module to |
| 980 |
be reloaded. |
| 981 |
.IP \(bu 4 |
| 982 |
Carp has been upgraded from version 1.20 to 1.26. |
| 983 |
.Sp |
| 984 |
It now includes last read filehandle info and puts a dot after the file |
| 985 |
and line number, just like errors from \f(CW\*(C`die\*(C'\fR [perl #106538]. |
| 986 |
.IP \(bu 4 |
| 987 |
charnames has been updated from version 1.18 to 1.30. |
| 988 |
.Sp |
| 989 |
\&\f(CW\*(C`charnames\*(C'\fR can now be invoked with a new option, \f(CW\*(C`:loose\*(C'\fR, |
| 990 |
which is like the existing \f(CW\*(C`:full\*(C'\fR option, but enables Unicode loose |
| 991 |
name matching. Details are in "LOOSE MATCHES" in charnames. |
| 992 |
.IP \(bu 4 |
| 993 |
B::Deparse has been upgraded from version 1.03 to 1.14. This fixes |
| 994 |
numerous deparsing bugs. |
| 995 |
.IP \(bu 4 |
| 996 |
CGI has been upgraded from version 3.52 to 3.59. |
| 997 |
.Sp |
| 998 |
It uses the public and documented FCGI.pm API in CGI::Fast. CGI::Fast was |
| 999 |
using an FCGI API that was deprecated and removed from documentation |
| 1000 |
more than ten years ago. Usage of this deprecated API with FCGI >= |
| 1001 |
0.70 or FCGI <= 0.73 introduces a security issue. |
| 1002 |
<https://rt.cpan.org/Public/Bug/Display.html?id=68380> |
| 1003 |
<http://web.nvd.nist.gov/view/vuln/detail?vulnId=CVE\-2011\-2766> |
| 1004 |
.Sp |
| 1005 |
Things that may break your code: |
| 1006 |
.Sp |
| 1007 |
\&\f(CWurl()\fR was fixed to return \f(CW\*(C`PATH_INFO\*(C'\fR when it is explicitly requested |
| 1008 |
with either the \f(CW\*(C`path=>1\*(C'\fR or \f(CW\*(C`path_info=>1\*(C'\fR flag. |
| 1009 |
.Sp |
| 1010 |
If your code is running under mod_rewrite (or compatible) and you are |
| 1011 |
calling \f(CWself_url()\fR or you are calling \f(CWurl()\fR and passing |
| 1012 |
\&\f(CW\*(C`path_info=>1\*(C'\fR, these methods will actually be returning |
| 1013 |
\&\f(CW\*(C`PATH_INFO\*(C'\fR now, as you have explicitly requested or \f(CWself_url()\fR |
| 1014 |
has requested on your behalf. |
| 1015 |
.Sp |
| 1016 |
The \f(CW\*(C`PATH_INFO\*(C'\fR has been omitted in such URLs since the issue was |
| 1017 |
introduced in the 3.12 release in December, 2005. |
| 1018 |
.Sp |
| 1019 |
This bug is so old your application may have come to depend on it or |
| 1020 |
workaround it. Check for application before upgrading to this release. |
| 1021 |
.Sp |
| 1022 |
Examples of affected method calls: |
| 1023 |
.Sp |
| 1024 |
.Vb 5 |
| 1025 |
\& $q\->url(\-absolute => 1, \-query => 1, \-path_info => 1); |
| 1026 |
\& $q\->url(\-path=>1); |
| 1027 |
\& $q\->url(\-full=>1,\-path=>1); |
| 1028 |
\& $q\->url(\-rewrite=>1,\-path=>1); |
| 1029 |
\& $q\->self_url(); |
| 1030 |
.Ve |
| 1031 |
.Sp |
| 1032 |
We no longer read from STDIN when the Content-Length is not set, |
| 1033 |
preventing requests with no Content-Length from sometimes freezing. |
| 1034 |
This is consistent with the CGI RFC 3875, and is also consistent with |
| 1035 |
CGI::Simple. However, the old behavior may have been expected by some |
| 1036 |
command-line uses of CGI.pm. |
| 1037 |
.Sp |
| 1038 |
In addition, the DELETE HTTP verb is now supported. |
| 1039 |
.IP \(bu 4 |
| 1040 |
Compress::Zlib has been upgraded from version 2.035 to 2.048. |
| 1041 |
.Sp |
| 1042 |
IO::Compress::Zip and IO::Uncompress::Unzip now have support for LZMA |
| 1043 |
(method 14). There is a fix for a CRC issue in IO::Compress::Unzip and |
| 1044 |
it supports Streamed Stored context now. And fixed a Zip64 issue in |
| 1045 |
IO::Compress::Zip when the content size was exactly 0xFFFFFFFF. |
| 1046 |
.IP \(bu 4 |
| 1047 |
Digest::SHA has been upgraded from version 5.61 to 5.71. |
| 1048 |
.Sp |
| 1049 |
Added BITS mode to the addfile method and shasum. This makes |
| 1050 |
partial-byte inputs possible via files/STDIN and lets shasum check |
| 1051 |
all 8074 NIST Msg vectors, where previously special programming was |
| 1052 |
required to do this. |
| 1053 |
.IP \(bu 4 |
| 1054 |
Encode has been upgraded from version 2.42 to 2.44. |
| 1055 |
.Sp |
| 1056 |
Missing aliases added, a deep recursion error fixed and various |
| 1057 |
documentation updates. |
| 1058 |
.Sp |
| 1059 |
Addressed 'decode_xs n\-byte heap\-overflow' security bug in Unicode.xs |
| 1060 |
(CVE\-2011\-2939). (5.14.2) |
| 1061 |
.IP \(bu 4 |
| 1062 |
ExtUtils::CBuilder updated from version 0.280203 to 0.280206. |
| 1063 |
.Sp |
| 1064 |
The new version appends CFLAGS and LDFLAGS to their Config.pm |
| 1065 |
counterparts. |
| 1066 |
.IP \(bu 4 |
| 1067 |
ExtUtils::ParseXS has been upgraded from version 2.2210 to 3.16. |
| 1068 |
.Sp |
| 1069 |
Much of ExtUtils::ParseXS, the module behind the XS compiler \f(CW\*(C`xsubpp\*(C'\fR, |
| 1070 |
was rewritten and cleaned up. It has been made somewhat more extensible |
| 1071 |
and now finally uses strictures. |
| 1072 |
.Sp |
| 1073 |
The typemap logic has been moved into a separate module, |
| 1074 |
ExtUtils::Typemaps. See "New Modules and Pragmata", above. |
| 1075 |
.Sp |
| 1076 |
For a complete set of changes, please see the ExtUtils::ParseXS |
| 1077 |
changelog, available on the CPAN. |
| 1078 |
.IP \(bu 4 |
| 1079 |
File::Glob has been upgraded from version 1.12 to 1.17. |
| 1080 |
.Sp |
| 1081 |
On Windows, tilde (~) expansion now checks the \f(CW\*(C`USERPROFILE\*(C'\fR environment |
| 1082 |
variable, after checking \f(CW\*(C`HOME\*(C'\fR. |
| 1083 |
.Sp |
| 1084 |
It has a new \f(CW\*(C`:bsd_glob\*(C'\fR export tag, intended to replace \f(CW\*(C`:glob\*(C'\fR. Like |
| 1085 |
\&\f(CW\*(C`:glob\*(C'\fR it overrides \f(CW\*(C`glob\*(C'\fR with a function that does not split the glob |
| 1086 |
pattern into words, but, unlike \f(CW\*(C`:glob\*(C'\fR, it iterates properly in scalar |
| 1087 |
context, instead of returning the last file. |
| 1088 |
.Sp |
| 1089 |
There are other changes affecting Perl's own \f(CW\*(C`glob\*(C'\fR operator (which uses |
| 1090 |
File::Glob internally, except on VMS). See "Performance Enhancements" |
| 1091 |
and "Selected Bug Fixes". |
| 1092 |
.IP \(bu 4 |
| 1093 |
FindBin updated from version 1.50 to 1.51. |
| 1094 |
.Sp |
| 1095 |
It no longer returns a wrong result if a script of the same name as the |
| 1096 |
current one exists in the path and is executable. |
| 1097 |
.IP \(bu 4 |
| 1098 |
HTTP::Tiny has been upgraded from version 0.012 to 0.017. |
| 1099 |
.Sp |
| 1100 |
Added support for using \f(CW$ENV{http_proxy}\fR to set the default proxy host. |
| 1101 |
.Sp |
| 1102 |
Adds additional shorthand methods for all common HTTP verbs, |
| 1103 |
a \f(CWpost_form()\fR method for POST-ing x\-www-form-urlencoded data and |
| 1104 |
a \f(CWwww_form_urlencode()\fR utility method. |
| 1105 |
.IP \(bu 4 |
| 1106 |
IO has been upgraded from version 1.25_04 to 1.25_06, and IO::Handle |
| 1107 |
from version 1.31 to 1.33. |
| 1108 |
.Sp |
| 1109 |
Together, these upgrades fix a problem with IO::Handle's \f(CW\*(C`getline\*(C'\fR and |
| 1110 |
\&\f(CW\*(C`getlines\*(C'\fR methods. When these methods are called on the special ARGV |
| 1111 |
handle, the next file is automatically opened, as happens with the built-in |
| 1112 |
\&\f(CW\*(C`<>\*(C'\fR and \f(CW\*(C`readline\*(C'\fR functions. But, unlike the built-ins, these |
| 1113 |
methods were not respecting the caller's use of the open pragma and |
| 1114 |
applying the appropriate I/O layers to the newly-opened file |
| 1115 |
[rt.cpan.org #66474]. |
| 1116 |
.IP \(bu 4 |
| 1117 |
IPC::Cmd has been upgraded from version 0.70 to 0.76. |
| 1118 |
.Sp |
| 1119 |
Capturing of command output (both \f(CW\*(C`STDOUT\*(C'\fR and \f(CW\*(C`STDERR\*(C'\fR) is now supported |
| 1120 |
using IPC::Open3 on MSWin32 without requiring IPC::Run. |
| 1121 |
.IP \(bu 4 |
| 1122 |
IPC::Open3 has been upgraded from version 1.09 to 1.12. |
| 1123 |
.Sp |
| 1124 |
Fixes a bug which prevented use of \f(CW\*(C`open3\*(C'\fR on Windows when \f(CW*STDIN\fR, |
| 1125 |
\&\f(CW*STDOUT\fR or \f(CW*STDERR\fR had been localized. |
| 1126 |
.Sp |
| 1127 |
Fixes a bug which prevented duplicating numeric file descriptors on Windows. |
| 1128 |
.Sp |
| 1129 |
\&\f(CW\*(C`open3\*(C'\fR with "\-" for the program name works once more. This was broken in |
| 1130 |
version 1.06 (and hence in Perl 5.14.0) [perl #95748]. |
| 1131 |
.IP \(bu 4 |
| 1132 |
Locale::Codes has been upgraded from version 3.16 to 3.21. |
| 1133 |
.Sp |
| 1134 |
Added Language Extension codes (langext) and Language Variation codes (langvar) |
| 1135 |
as defined in the IANA language registry. |
| 1136 |
.Sp |
| 1137 |
Added language codes from ISO 639\-5 |
| 1138 |
.Sp |
| 1139 |
Added language/script codes from the IANA language subtag registry |
| 1140 |
.Sp |
| 1141 |
Fixed an uninitialized value warning [rt.cpan.org #67438]. |
| 1142 |
.Sp |
| 1143 |
Fixed the return value for the all_XXX_codes and all_XXX_names functions |
| 1144 |
[rt.cpan.org #69100]. |
| 1145 |
.Sp |
| 1146 |
Reorganized modules to move Locale::MODULE to Locale::Codes::MODULE to allow |
| 1147 |
for cleaner future additions. The original four modules (Locale::Language, |
| 1148 |
Locale::Currency, Locale::Country, Locale::Script) will continue to work, but |
| 1149 |
all new sets of codes will be added in the Locale::Codes namespace. |
| 1150 |
.Sp |
| 1151 |
The code2XXX, XXX2code, all_XXX_codes, and all_XXX_names functions now |
| 1152 |
support retired codes. All codesets may be specified by a constant or |
| 1153 |
by their name now. Previously, they were specified only by a constant. |
| 1154 |
.Sp |
| 1155 |
The alias_code function exists for backward compatibility. It has been |
| 1156 |
replaced by rename_country_code. The alias_code function will be |
| 1157 |
removed some time after September, 2013. |
| 1158 |
.Sp |
| 1159 |
All work is now done in the central module (Locale::Codes). Previously, |
| 1160 |
some was still done in the wrapper modules (Locale::Codes::*). Added |
| 1161 |
Language Family codes (langfam) as defined in ISO 639\-5. |
| 1162 |
.IP \(bu 4 |
| 1163 |
Math::BigFloat has been upgraded from version 1.993 to 1.997. |
| 1164 |
.Sp |
| 1165 |
The \f(CW\*(C`numify\*(C'\fR method has been corrected to return a normalized Perl number |
| 1166 |
(the result of \f(CW\*(C`0 + $thing\*(C'\fR), instead of a string [rt.cpan.org #66732]. |
| 1167 |
.IP \(bu 4 |
| 1168 |
Math::BigInt has been upgraded from version 1.994 to 1.998. |
| 1169 |
.Sp |
| 1170 |
It provides a new \f(CW\*(C`bsgn\*(C'\fR method that complements the \f(CW\*(C`babs\*(C'\fR method. |
| 1171 |
.Sp |
| 1172 |
It fixes the internal \f(CW\*(C`objectify\*(C'\fR function's handling of "foreign objects" |
| 1173 |
so they are converted to the appropriate class (Math::BigInt or |
| 1174 |
Math::BigFloat). |
| 1175 |
.IP \(bu 4 |
| 1176 |
Math::BigRat has been upgraded from version 0.2602 to 0.2603. |
| 1177 |
.Sp |
| 1178 |
\&\f(CWint()\fR on a Math::BigRat object containing \-1/2 now creates a |
| 1179 |
Math::BigInt containing 0, rather than \-0. Math::BigInt does not even |
| 1180 |
support negative zero, so the resulting object was actually malformed |
| 1181 |
[perl #95530]. |
| 1182 |
.IP \(bu 4 |
| 1183 |
Math::Complex has been upgraded from version 1.56 to 1.59 |
| 1184 |
and Math::Trig from version 1.2 to 1.22. |
| 1185 |
.Sp |
| 1186 |
Fixes include: correct copy constructor usage; fix polarwise formatting with |
| 1187 |
numeric format specifier; and more stable \f(CW\*(C`great_circle_direction\*(C'\fR algorithm. |
| 1188 |
.IP \(bu 4 |
| 1189 |
Module::CoreList has been upgraded from version 2.51 to 2.66. |
| 1190 |
.Sp |
| 1191 |
The \f(CW\*(C`corelist\*(C'\fR utility now understands the \f(CW\*(C`\-r\*(C'\fR option for displaying |
| 1192 |
Perl release dates and the \f(CW\*(C`\-\-diff\*(C'\fR option to print the set of modlib |
| 1193 |
changes between two perl distributions. |
| 1194 |
.IP \(bu 4 |
| 1195 |
Module::Metadata has been upgraded from version 1.000004 to 1.000009. |
| 1196 |
.Sp |
| 1197 |
Adds \f(CW\*(C`provides\*(C'\fR method to generate a CPAN META provides data structure |
| 1198 |
correctly; use of \f(CW\*(C`package_versions_from_directory\*(C'\fR is discouraged. |
| 1199 |
.IP \(bu 4 |
| 1200 |
ODBM_File has been upgraded from version 1.10 to 1.12. |
| 1201 |
.Sp |
| 1202 |
The XS code is now compiled with \f(CW\*(C`PERL_NO_GET_CONTEXT\*(C'\fR, which will aid |
| 1203 |
performance under ithreads. |
| 1204 |
.IP \(bu 4 |
| 1205 |
open has been upgraded from version 1.08 to 1.10. |
| 1206 |
.Sp |
| 1207 |
It no longer turns off layers on standard handles when invoked without the |
| 1208 |
":std" directive. Similarly, when invoked \fIwith\fR the ":std" directive, it |
| 1209 |
now clears layers on STDERR before applying the new ones, and not just on |
| 1210 |
STDIN and STDOUT [perl #92728]. |
| 1211 |
.IP \(bu 4 |
| 1212 |
overload has been upgraded from version 1.13 to 1.18. |
| 1213 |
.Sp |
| 1214 |
\&\f(CW\*(C`overload::Overloaded\*(C'\fR no longer calls \f(CW\*(C`can\*(C'\fR on the class, but uses |
| 1215 |
another means to determine whether the object has overloading. It was |
| 1216 |
never correct for it to call \f(CW\*(C`can\*(C'\fR, as overloading does not respect |
| 1217 |
AUTOLOAD. So classes that autoload methods and implement \f(CW\*(C`can\*(C'\fR no longer |
| 1218 |
have to account for overloading [perl #40333]. |
| 1219 |
.Sp |
| 1220 |
A warning is now produced for invalid arguments. See "New Diagnostics". |
| 1221 |
.IP \(bu 4 |
| 1222 |
PerlIO::scalar has been upgraded from version 0.11 to 0.14. |
| 1223 |
.Sp |
| 1224 |
(This is the module that implements \f(CW\*(C`open $fh, \*(Aq>\*(Aq, \e$scalar\*(C'\fR.) |
| 1225 |
.Sp |
| 1226 |
It fixes a problem with \f(CW\*(C`open my $fh, ">", \e$scalar\*(C'\fR not working if |
| 1227 |
\&\f(CW$scalar\fR is a copy-on-write scalar. (5.14.2) |
| 1228 |
.Sp |
| 1229 |
It also fixes a hang that occurs with \f(CW\*(C`readline\*(C'\fR or \f(CW\*(C`<$fh>\*(C'\fR if a |
| 1230 |
typeglob has been assigned to \f(CW$scalar\fR [perl #92258]. |
| 1231 |
.Sp |
| 1232 |
It no longer assumes during \f(CW\*(C`seek\*(C'\fR that \f(CW$scalar\fR is a string internally. |
| 1233 |
If it didn't crash, it was close to doing so [perl #92706]. Also, the |
| 1234 |
internal print routine no longer assumes that the position set by \f(CW\*(C`seek\*(C'\fR |
| 1235 |
is valid, but extends the string to that position, filling the intervening |
| 1236 |
bytes (between the old length and the seek position) with nulls |
| 1237 |
[perl #78980]. |
| 1238 |
.Sp |
| 1239 |
Printing to an in-memory handle now works if the \f(CW$scalar\fR holds a reference, |
| 1240 |
stringifying the reference before modifying it. References used to be |
| 1241 |
treated as empty strings. |
| 1242 |
.Sp |
| 1243 |
Printing to an in-memory handle no longer crashes if the \f(CW$scalar\fR happens to |
| 1244 |
hold a number internally, but no string buffer. |
| 1245 |
.Sp |
| 1246 |
Printing to an in-memory handle no longer creates scalars that confuse |
| 1247 |
the regular expression engine [perl #108398]. |
| 1248 |
.IP \(bu 4 |
| 1249 |
Pod::Functions has been upgraded from version 1.04 to 1.05. |
| 1250 |
.Sp |
| 1251 |
\&\fIFunctions.pm\fR is now generated at perl build time from annotations in |
| 1252 |
\&\fIperlfunc.pod\fR. This will ensure that Pod::Functions and perlfunc |
| 1253 |
remain in synchronisation. |
| 1254 |
.IP \(bu 4 |
| 1255 |
Pod::Html has been upgraded from version 1.11 to 1.1502. |
| 1256 |
.Sp |
| 1257 |
This is an extensive rewrite of Pod::Html to use Pod::Simple under |
| 1258 |
the hood. The output has changed significantly. |
| 1259 |
.IP \(bu 4 |
| 1260 |
Pod::Perldoc has been upgraded from version 3.15_03 to 3.17. |
| 1261 |
.Sp |
| 1262 |
It corrects the search paths on VMS [perl #90640]. (5.14.1) |
| 1263 |
.Sp |
| 1264 |
The \fB\-v\fR option now fetches the right section for \f(CW$0\fR. |
| 1265 |
.Sp |
| 1266 |
This upgrade has numerous significant fixes. Consult its changelog on |
| 1267 |
the CPAN for more information. |
| 1268 |
.IP \(bu 4 |
| 1269 |
POSIX has been upgraded from version 1.24 to 1.30. |
| 1270 |
.Sp |
| 1271 |
POSIX no longer uses AutoLoader. Any code which was relying on this |
| 1272 |
implementation detail was buggy, and may fail because of this change. |
| 1273 |
The module's Perl code has been considerably simplified, roughly halving |
| 1274 |
the number of lines, with no change in functionality. The XS code has |
| 1275 |
been refactored to reduce the size of the shared object by about 12%, |
| 1276 |
with no change in functionality. More POSIX functions now have tests. |
| 1277 |
.Sp |
| 1278 |
\&\f(CW\*(C`sigsuspend\*(C'\fR and \f(CW\*(C`pause\*(C'\fR now run signal handlers before returning, as the |
| 1279 |
whole point of these two functions is to wait until a signal has |
| 1280 |
arrived, and then return \fIafter\fR it has been triggered. Delayed, or |
| 1281 |
"safe", signals were preventing that from happening, possibly resulting in |
| 1282 |
race conditions [perl #107216]. |
| 1283 |
.Sp |
| 1284 |
\&\f(CW\*(C`POSIX::sleep\*(C'\fR is now a direct call into the underlying OS \f(CW\*(C`sleep\*(C'\fR |
| 1285 |
function, instead of being a Perl wrapper on \f(CW\*(C`CORE::sleep\*(C'\fR. |
| 1286 |
\&\f(CW\*(C`POSIX::dup2\*(C'\fR now returns the correct value on Win32 (\fIi.e.\fR, the file |
| 1287 |
descriptor). \f(CW\*(C`POSIX::SigSet\*(C'\fR \f(CW\*(C`sigsuspend\*(C'\fR and \f(CW\*(C`sigpending\*(C'\fR and |
| 1288 |
\&\f(CW\*(C`POSIX::pause\*(C'\fR now dispatch safe signals immediately before returning to |
| 1289 |
their caller. |
| 1290 |
.Sp |
| 1291 |
\&\f(CW\*(C`POSIX::Termios::setattr\*(C'\fR now defaults the third argument to \f(CW\*(C`TCSANOW\*(C'\fR, |
| 1292 |
instead of 0. On most platforms \f(CW\*(C`TCSANOW\*(C'\fR is defined to be 0, but on some |
| 1293 |
0 is not a valid parameter, which caused a call with defaults to fail. |
| 1294 |
.IP \(bu 4 |
| 1295 |
Socket has been upgraded from version 1.94 to 2.001. |
| 1296 |
.Sp |
| 1297 |
It has new functions and constants for handling IPv6 sockets: |
| 1298 |
.Sp |
| 1299 |
.Vb 11 |
| 1300 |
\& pack_ipv6_mreq |
| 1301 |
\& unpack_ipv6_mreq |
| 1302 |
\& IPV6_ADD_MEMBERSHIP |
| 1303 |
\& IPV6_DROP_MEMBERSHIP |
| 1304 |
\& IPV6_MTU |
| 1305 |
\& IPV6_MTU_DISCOVER |
| 1306 |
\& IPV6_MULTICAST_HOPS |
| 1307 |
\& IPV6_MULTICAST_IF |
| 1308 |
\& IPV6_MULTICAST_LOOP |
| 1309 |
\& IPV6_UNICAST_HOPS |
| 1310 |
\& IPV6_V6ONLY |
| 1311 |
.Ve |
| 1312 |
.IP \(bu 4 |
| 1313 |
Storable has been upgraded from version 2.27 to 2.34. |
| 1314 |
.Sp |
| 1315 |
It no longer turns copy-on-write scalars into read-only scalars when |
| 1316 |
freezing and thawing. |
| 1317 |
.IP \(bu 4 |
| 1318 |
Sys::Syslog has been upgraded from version 0.27 to 0.29. |
| 1319 |
.Sp |
| 1320 |
This upgrade closes many outstanding bugs. |
| 1321 |
.IP \(bu 4 |
| 1322 |
Term::ANSIColor has been upgraded from version 3.00 to 3.01. |
| 1323 |
.Sp |
| 1324 |
Only interpret an initial array reference as a list of colors, not any initial |
| 1325 |
reference, allowing the colored function to work properly on objects with |
| 1326 |
stringification defined. |
| 1327 |
.IP \(bu 4 |
| 1328 |
Term::ReadLine has been upgraded from version 1.07 to 1.09. |
| 1329 |
.Sp |
| 1330 |
Term::ReadLine now supports any event loop, including unpublished ones and |
| 1331 |
simple IO::Select, loops without the need to rewrite existing code for |
| 1332 |
any particular framework [perl #108470]. |
| 1333 |
.IP \(bu 4 |
| 1334 |
threads::shared has been upgraded from version 1.37 to 1.40. |
| 1335 |
.Sp |
| 1336 |
Destructors on shared objects used to be ignored sometimes if the objects |
| 1337 |
were referenced only by shared data structures. This has been mostly |
| 1338 |
fixed, but destructors may still be ignored if the objects still exist at |
| 1339 |
global destruction time [perl #98204]. |
| 1340 |
.IP \(bu 4 |
| 1341 |
Unicode::Collate has been upgraded from version 0.73 to 0.89. |
| 1342 |
.Sp |
| 1343 |
Updated to CLDR 1.9.1 |
| 1344 |
.Sp |
| 1345 |
Locales updated to CLDR 2.0: mk, mt, nb, nn, ro, ru, sk, sr, sv, uk, |
| 1346 |
zh_\|_pinyin, zh_\|_stroke |
| 1347 |
.Sp |
| 1348 |
Newly supported locales: bn, fa, ml, mr, or, pa, sa, si, si_\|_dictionary, |
| 1349 |
sr_Latn, sv_\|_reformed, ta, te, th, ur, wae. |
| 1350 |
.Sp |
| 1351 |
Tailored compatibility ideographs as well as unified ideographs for the |
| 1352 |
locales: ja, ko, zh_\|_big5han, zh_\|_gb2312han, zh_\|_pinyin, zh_\|_stroke. |
| 1353 |
.Sp |
| 1354 |
Locale/*.pl files are now searched for in \f(CW@INC\fR. |
| 1355 |
.IP \(bu 4 |
| 1356 |
Unicode::Normalize has been upgraded from version 1.10 to 1.14. |
| 1357 |
.Sp |
| 1358 |
Fixes for the removal of \fIunicore/CompositionExclusions.txt\fR from core. |
| 1359 |
.IP \(bu 4 |
| 1360 |
Unicode::UCD has been upgraded from version 0.32 to 0.43. |
| 1361 |
.Sp |
| 1362 |
This adds four new functions: \f(CWprop_aliases()\fR and |
| 1363 |
\&\f(CWprop_value_aliases()\fR, which are used to find all Unicode-approved |
| 1364 |
synonyms for property names, or to convert from one name to another; |
| 1365 |
\&\f(CW\*(C`prop_invlist\*(C'\fR which returns all code points matching a given |
| 1366 |
Unicode binary property; and \f(CW\*(C`prop_invmap\*(C'\fR which returns the complete |
| 1367 |
specification of a given Unicode property. |
| 1368 |
.IP \(bu 4 |
| 1369 |
Win32API::File has been upgraded from version 0.1101 to 0.1200. |
| 1370 |
.Sp |
| 1371 |
Added SetStdHandle and GetStdHandle functions |
| 1372 |
.SS "Removed Modules and Pragmata" |
| 1373 |
.IX Subsection "Removed Modules and Pragmata" |
| 1374 |
As promised in Perl 5.14.0's release notes, the following modules have |
| 1375 |
been removed from the core distribution, and if needed should be installed |
| 1376 |
from CPAN instead. |
| 1377 |
.IP \(bu 4 |
| 1378 |
Devel::DProf has been removed from the Perl core. Prior version was |
| 1379 |
20110228.00. |
| 1380 |
.IP \(bu 4 |
| 1381 |
Shell has been removed from the Perl core. Prior version was 0.72_01. |
| 1382 |
.IP \(bu 4 |
| 1383 |
Several old perl4\-style libraries which have been deprecated with 5.14 |
| 1384 |
are now removed: |
| 1385 |
.Sp |
| 1386 |
.Vb 5 |
| 1387 |
\& abbrev.pl assert.pl bigfloat.pl bigint.pl bigrat.pl cacheout.pl |
| 1388 |
\& complete.pl ctime.pl dotsh.pl exceptions.pl fastcwd.pl flush.pl |
| 1389 |
\& getcwd.pl getopt.pl getopts.pl hostname.pl importenv.pl |
| 1390 |
\& lib/find{,depth}.pl look.pl newgetopt.pl open2.pl open3.pl |
| 1391 |
\& pwd.pl shellwords.pl stat.pl tainted.pl termcap.pl timelocal.pl |
| 1392 |
.Ve |
| 1393 |
.Sp |
| 1394 |
They can be found on CPAN as Perl4::CoreLibs. |
| 1395 |
.SH Documentation |
| 1396 |
.IX Header "Documentation" |
| 1397 |
.SS "New Documentation" |
| 1398 |
.IX Subsection "New Documentation" |
| 1399 |
\fIperldtrace\fR |
| 1400 |
.IX Subsection "perldtrace" |
| 1401 |
.PP |
| 1402 |
perldtrace describes Perl's DTrace support, listing the provided probes |
| 1403 |
and gives examples of their use. |
| 1404 |
.PP |
| 1405 |
\fIperlexperiment\fR |
| 1406 |
.IX Subsection "perlexperiment" |
| 1407 |
.PP |
| 1408 |
This document is intended to provide a list of experimental features in |
| 1409 |
Perl. It is still a work in progress. |
| 1410 |
.PP |
| 1411 |
\fIperlootut\fR |
| 1412 |
.IX Subsection "perlootut" |
| 1413 |
.PP |
| 1414 |
This a new OO tutorial. It focuses on basic OO concepts, and then recommends |
| 1415 |
that readers choose an OO framework from CPAN. |
| 1416 |
.PP |
| 1417 |
\fIperlxstypemap\fR |
| 1418 |
.IX Subsection "perlxstypemap" |
| 1419 |
.PP |
| 1420 |
The new manual describes the XS typemapping mechanism in unprecedented |
| 1421 |
detail and combines new documentation with information extracted from |
| 1422 |
perlxs and the previously unofficial list of all core typemaps. |
| 1423 |
.SS "Changes to Existing Documentation" |
| 1424 |
.IX Subsection "Changes to Existing Documentation" |
| 1425 |
\fIperlapi\fR |
| 1426 |
.IX Subsection "perlapi" |
| 1427 |
.IP \(bu 4 |
| 1428 |
The HV API has long accepted negative lengths to show that the key is |
| 1429 |
in UTF8. This is now documented. |
| 1430 |
.IP \(bu 4 |
| 1431 |
The \f(CWboolSV()\fR macro is now documented. |
| 1432 |
.PP |
| 1433 |
\fIperlfunc\fR |
| 1434 |
.IX Subsection "perlfunc" |
| 1435 |
.IP \(bu 4 |
| 1436 |
\&\f(CW\*(C`dbmopen\*(C'\fR treats a 0 mode as a special case, that prevents a nonexistent |
| 1437 |
file from being created. This has been the case since Perl 5.000, but was |
| 1438 |
never documented anywhere. Now the perlfunc entry mentions it |
| 1439 |
[perl #90064]. |
| 1440 |
.IP \(bu 4 |
| 1441 |
As an accident of history, \f(CW\*(C`open $fh, \*(Aq<:\*(Aq, ...\*(C'\fR applies the default |
| 1442 |
layers for the platform (\f(CW\*(C`:raw\*(C'\fR on Unix, \f(CW\*(C`:crlf\*(C'\fR on Windows), ignoring |
| 1443 |
whatever is declared by open.pm. This seems such a useful feature |
| 1444 |
it has been documented in perlfunc and open. |
| 1445 |
.IP \(bu 4 |
| 1446 |
The entry for \f(CW\*(C`split\*(C'\fR has been rewritten. It is now far clearer than |
| 1447 |
before. |
| 1448 |
.PP |
| 1449 |
\fIperlguts\fR |
| 1450 |
.IX Subsection "perlguts" |
| 1451 |
.IP \(bu 4 |
| 1452 |
A new section, Autoloading with XSUBs, |
| 1453 |
has been added, which explains the two APIs for accessing the name of the |
| 1454 |
autoloaded sub. |
| 1455 |
.IP \(bu 4 |
| 1456 |
Some function descriptions in perlguts were confusing, as it was |
| 1457 |
not clear whether they referred to the function above or below the |
| 1458 |
description. This has been clarified [perl #91790]. |
| 1459 |
.PP |
| 1460 |
\fIperlobj\fR |
| 1461 |
.IX Subsection "perlobj" |
| 1462 |
.IP \(bu 4 |
| 1463 |
This document has been rewritten from scratch, and its coverage of various OO |
| 1464 |
concepts has been expanded. |
| 1465 |
.PP |
| 1466 |
\fIperlop\fR |
| 1467 |
.IX Subsection "perlop" |
| 1468 |
.IP \(bu 4 |
| 1469 |
Documentation of the smartmatch operator has been reworked and moved from |
| 1470 |
perlsyn to perlop where it belongs. |
| 1471 |
.Sp |
| 1472 |
It has also been corrected for the case of \f(CW\*(C`undef\*(C'\fR on the left-hand |
| 1473 |
side. The list of different smart match behaviors had an item in the |
| 1474 |
wrong place. |
| 1475 |
.IP \(bu 4 |
| 1476 |
Documentation of the ellipsis statement (\f(CW\*(C`...\*(C'\fR) has been reworked and |
| 1477 |
moved from perlop to perlsyn. |
| 1478 |
.IP \(bu 4 |
| 1479 |
The explanation of bitwise operators has been expanded to explain how they |
| 1480 |
work on Unicode strings (5.14.1). |
| 1481 |
.IP \(bu 4 |
| 1482 |
More examples for \f(CW\*(C`m//g\*(C'\fR have been added (5.14.1). |
| 1483 |
.IP \(bu 4 |
| 1484 |
The \f(CW\*(C`<<\eFOO\*(C'\fR here-doc syntax has been documented (5.14.1). |
| 1485 |
.PP |
| 1486 |
\fIperlpragma\fR |
| 1487 |
.IX Subsection "perlpragma" |
| 1488 |
.IP \(bu 4 |
| 1489 |
There is now a standard convention for naming keys in the \f(CW\*(C`%^H\*(C'\fR, |
| 1490 |
documented under Key naming. |
| 1491 |
.PP |
| 1492 |
\fI"Laundering and Detecting Tainted Data" in perlsec\fR |
| 1493 |
.IX Subsection """Laundering and Detecting Tainted Data"" in perlsec" |
| 1494 |
.IP \(bu 4 |
| 1495 |
The example function for checking for taintedness contained a subtle |
| 1496 |
error. \f(CW$@\fR needs to be localized to prevent its changing this |
| 1497 |
global's value outside the function. The preferred method to check for |
| 1498 |
this remains "tainted" in Scalar::Util. |
| 1499 |
.PP |
| 1500 |
\fIperllol\fR |
| 1501 |
.IX Subsection "perllol" |
| 1502 |
.IP \(bu 4 |
| 1503 |
perllol has been expanded with examples using the new \f(CW\*(C`push $scalar\*(C'\fR |
| 1504 |
syntax introduced in Perl 5.14.0 (5.14.1). |
| 1505 |
.PP |
| 1506 |
\fIperlmod\fR |
| 1507 |
.IX Subsection "perlmod" |
| 1508 |
.IP \(bu 4 |
| 1509 |
perlmod now states explicitly that some types of explicit symbol table |
| 1510 |
manipulation are not supported. This codifies what was effectively already |
| 1511 |
the case [perl #78074]. |
| 1512 |
.PP |
| 1513 |
\fIperlpodstyle\fR |
| 1514 |
.IX Subsection "perlpodstyle" |
| 1515 |
.IP \(bu 4 |
| 1516 |
The tips on which formatting codes to use have been corrected and greatly |
| 1517 |
expanded. |
| 1518 |
.IP \(bu 4 |
| 1519 |
There are now a couple of example one-liners for previewing POD files after |
| 1520 |
they have been edited. |
| 1521 |
.PP |
| 1522 |
\fIperlre\fR |
| 1523 |
.IX Subsection "perlre" |
| 1524 |
.IP \(bu 4 |
| 1525 |
The \f(CW\*(C`(*COMMIT)\*(C'\fR directive is now listed in the right section |
| 1526 |
(Verbs without an argument). |
| 1527 |
.PP |
| 1528 |
\fIperlrun\fR |
| 1529 |
.IX Subsection "perlrun" |
| 1530 |
.IP \(bu 4 |
| 1531 |
perlrun has undergone a significant clean-up. Most notably, the |
| 1532 |
\&\fB\-0x...\fR form of the \fB\-0\fR flag has been clarified, and the final section |
| 1533 |
on environment variables has been corrected and expanded (5.14.1). |
| 1534 |
.PP |
| 1535 |
\fIperlsub\fR |
| 1536 |
.IX Subsection "perlsub" |
| 1537 |
.IP \(bu 4 |
| 1538 |
The ($;) prototype syntax, which has existed for rather a long time, is now |
| 1539 |
documented in perlsub. It lets a unary function have the same |
| 1540 |
precedence as a list operator. |
| 1541 |
.PP |
| 1542 |
\fIperltie\fR |
| 1543 |
.IX Subsection "perltie" |
| 1544 |
.IP \(bu 4 |
| 1545 |
The required syntax for tying handles has been documented. |
| 1546 |
.PP |
| 1547 |
\fIperlvar\fR |
| 1548 |
.IX Subsection "perlvar" |
| 1549 |
.IP \(bu 4 |
| 1550 |
The documentation for $! has been corrected and clarified. |
| 1551 |
It used to state that $! could be \f(CW\*(C`undef\*(C'\fR, which is not the case. It was |
| 1552 |
also unclear whether system calls set C's \f(CW\*(C`errno\*(C'\fR or Perl's \f(CW$!\fR |
| 1553 |
[perl #91614]. |
| 1554 |
.IP \(bu 4 |
| 1555 |
Documentation for $$ has been amended with additional |
| 1556 |
cautions regarding changing the process ID. |
| 1557 |
.PP |
| 1558 |
\fIOther Changes\fR |
| 1559 |
.IX Subsection "Other Changes" |
| 1560 |
.IP \(bu 4 |
| 1561 |
perlxs was extended with documentation on inline typemaps. |
| 1562 |
.IP \(bu 4 |
| 1563 |
perlref has a new Circular References |
| 1564 |
section explaining how circularities may not be freed and how to solve that |
| 1565 |
with weak references. |
| 1566 |
.IP \(bu 4 |
| 1567 |
Parts of perlapi were clarified, and Perl equivalents of some C |
| 1568 |
functions have been added as an additional mode of exposition. |
| 1569 |
.IP \(bu 4 |
| 1570 |
A few parts of perlre and perlrecharclass were clarified. |
| 1571 |
.SS "Removed Documentation" |
| 1572 |
.IX Subsection "Removed Documentation" |
| 1573 |
\fIOld OO Documentation\fR |
| 1574 |
.IX Subsection "Old OO Documentation" |
| 1575 |
.PP |
| 1576 |
The old OO tutorials, perltoot, perltooc, and perlboot, have been |
| 1577 |
removed. The perlbot (bag of object tricks) document has been removed |
| 1578 |
as well. |
| 1579 |
.PP |
| 1580 |
\fIDevelopment Deltas\fR |
| 1581 |
.IX Subsection "Development Deltas" |
| 1582 |
.PP |
| 1583 |
The perldelta files for development releases are no longer packaged with |
| 1584 |
perl. These can still be found in the perl source code repository. |
| 1585 |
.SH Diagnostics |
| 1586 |
.IX Header "Diagnostics" |
| 1587 |
The following additions or changes have been made to diagnostic output, |
| 1588 |
including warnings and fatal error messages. For the complete list of |
| 1589 |
diagnostic messages, see perldiag. |
| 1590 |
.SS "New Diagnostics" |
| 1591 |
.IX Subsection "New Diagnostics" |
| 1592 |
\fINew Errors\fR |
| 1593 |
.IX Subsection "New Errors" |
| 1594 |
.IP \(bu 4 |
| 1595 |
Cannot set tied \f(CW@DB::args\fR |
| 1596 |
.Sp |
| 1597 |
This error occurs when \f(CW\*(C`caller\*(C'\fR tries to set \f(CW@DB::args\fR but finds it |
| 1598 |
tied. Before this error was added, it used to crash instead. |
| 1599 |
.IP \(bu 4 |
| 1600 |
Cannot tie unreifiable array |
| 1601 |
.Sp |
| 1602 |
This error is part of a safety check that the \f(CW\*(C`tie\*(C'\fR operator does before |
| 1603 |
tying a special array like \f(CW@_\fR. You should never see this message. |
| 1604 |
.IP \(bu 4 |
| 1605 |
&CORE::%s cannot be called directly |
| 1606 |
.Sp |
| 1607 |
This occurs when a subroutine in the \f(CW\*(C`CORE::\*(C'\fR namespace is called |
| 1608 |
with \f(CW&foo\fR syntax or through a reference. Some subroutines |
| 1609 |
in this package cannot yet be called that way, but must be |
| 1610 |
called as barewords. See "Subroutines in the \f(CW\*(C`CORE\*(C'\fR namespace", above. |
| 1611 |
.IP \(bu 4 |
| 1612 |
Source filters apply only to byte streams |
| 1613 |
.Sp |
| 1614 |
This new error occurs when you try to activate a source filter (usually by |
| 1615 |
loading a source filter module) within a string passed to \f(CW\*(C`eval\*(C'\fR under the |
| 1616 |
\&\f(CW\*(C`unicode_eval\*(C'\fR feature. |
| 1617 |
.PP |
| 1618 |
\fINew Warnings\fR |
| 1619 |
.IX Subsection "New Warnings" |
| 1620 |
.IP \(bu 4 |
| 1621 |
defined(@array) is deprecated |
| 1622 |
.Sp |
| 1623 |
The long-deprecated \f(CWdefined(@array)\fR now also warns for package variables. |
| 1624 |
Previously it issued a warning for lexical variables only. |
| 1625 |
.IP \(bu 4 |
| 1626 |
\&\fBlength()\fR used on \f(CW%s\fR |
| 1627 |
.Sp |
| 1628 |
This new warning occurs when \f(CW\*(C`length\*(C'\fR is used on an array or hash, instead |
| 1629 |
of \f(CWscalar(@array)\fR or \f(CW\*(C`scalar(keys %hash)\*(C'\fR. |
| 1630 |
.IP \(bu 4 |
| 1631 |
lvalue attribute \f(CW%s\fR already-defined subroutine |
| 1632 |
.Sp |
| 1633 |
attributes.pm now emits this warning when the :lvalue |
| 1634 |
attribute is applied to a Perl subroutine that has already been defined, as |
| 1635 |
doing so can have unexpected side-effects. |
| 1636 |
.IP \(bu 4 |
| 1637 |
overload arg '%s' is invalid |
| 1638 |
.Sp |
| 1639 |
This warning, in the "overload" category, is produced when the overload |
| 1640 |
pragma is given an argument it doesn't recognize, presumably a mistyped |
| 1641 |
operator. |
| 1642 |
.IP \(bu 4 |
| 1643 |
$[ used in \f(CW%s\fR (did you mean $] ?) |
| 1644 |
.Sp |
| 1645 |
This new warning exists to catch the mistaken use of \f(CW$[\fR in version |
| 1646 |
checks. \f(CW$]\fR, not \f(CW$[\fR, contains the version number. |
| 1647 |
.IP \(bu 4 |
| 1648 |
Useless assignment to a temporary |
| 1649 |
.Sp |
| 1650 |
Assigning to a temporary scalar returned |
| 1651 |
from an lvalue subroutine now produces this |
| 1652 |
warning [perl #31946]. |
| 1653 |
.IP \(bu 4 |
| 1654 |
Useless use of \eE |
| 1655 |
.Sp |
| 1656 |
\&\f(CW\*(C`\eE\*(C'\fR does nothing unless preceded by \f(CW\*(C`\eQ\*(C'\fR, \f(CW\*(C`\eL\*(C'\fR or \f(CW\*(C`\eU\*(C'\fR. |
| 1657 |
.SS "Removed Errors" |
| 1658 |
.IX Subsection "Removed Errors" |
| 1659 |
.IP \(bu 4 |
| 1660 |
"sort is now a reserved word" |
| 1661 |
.Sp |
| 1662 |
This error used to occur when \f(CW\*(C`sort\*(C'\fR was called without arguments, |
| 1663 |
followed by \f(CW\*(C`;\*(C'\fR or \f(CW\*(C`)\*(C'\fR. (E.g., \f(CW\*(C`sort;\*(C'\fR would die, but \f(CW\*(C`{sort}\*(C'\fR was |
| 1664 |
OK.) This error message was added in Perl 3 to catch code like |
| 1665 |
\&\f(CWclose(sort)\fR which would no longer work. More than two decades later, |
| 1666 |
this message is no longer appropriate. Now \f(CW\*(C`sort\*(C'\fR without arguments is |
| 1667 |
always allowed, and returns an empty list, as it did in those cases |
| 1668 |
where it was already allowed [perl #90030]. |
| 1669 |
.SS "Changes to Existing Diagnostics" |
| 1670 |
.IX Subsection "Changes to Existing Diagnostics" |
| 1671 |
.IP \(bu 4 |
| 1672 |
The "Applying pattern match..." or similar warning produced when an |
| 1673 |
array or hash is on the left-hand side of the \f(CW\*(C`=~\*(C'\fR operator now |
| 1674 |
mentions the name of the variable. |
| 1675 |
.IP \(bu 4 |
| 1676 |
The "Attempt to free non-existent shared string" has had the spelling |
| 1677 |
of "non-existent" corrected to "nonexistent". It was already listed |
| 1678 |
with the correct spelling in perldiag. |
| 1679 |
.IP \(bu 4 |
| 1680 |
The error messages for using \f(CW\*(C`default\*(C'\fR and \f(CW\*(C`when\*(C'\fR outside a |
| 1681 |
topicalizer have been standardized to match the messages for \f(CW\*(C`continue\*(C'\fR |
| 1682 |
and loop controls. They now read 'Can't "default" outside a |
| 1683 |
topicalizer' and 'Can't "when" outside a topicalizer'. They both used |
| 1684 |
to be 'Can't use \fBwhen()\fR outside a topicalizer' [perl #91514]. |
| 1685 |
.IP \(bu 4 |
| 1686 |
The message, "Code point 0x%X is not Unicode, no properties match it; |
| 1687 |
all inverse properties do" has been changed to "Code point 0x%X is not |
| 1688 |
Unicode, all \ep{} matches fail; all \eP{} matches succeed". |
| 1689 |
.IP \(bu 4 |
| 1690 |
Redefinition warnings for constant subroutines used to be mandatory, |
| 1691 |
even occurring under \f(CW\*(C`no warnings\*(C'\fR. Now they respect the warnings |
| 1692 |
pragma. |
| 1693 |
.IP \(bu 4 |
| 1694 |
The "glob failed" warning message is now suppressible via \f(CW\*(C`no warnings\*(C'\fR |
| 1695 |
[perl #111656]. |
| 1696 |
.IP \(bu 4 |
| 1697 |
The Invalid version format |
| 1698 |
error message now says "negative version number" within the parentheses, |
| 1699 |
rather than "non-numeric data", for negative numbers. |
| 1700 |
.IP \(bu 4 |
| 1701 |
The two warnings |
| 1702 |
Possible attempt to put comments in \fBqw()\fR list |
| 1703 |
and |
| 1704 |
Possible attempt to separate words with commas |
| 1705 |
are no longer mutually exclusive: the same \f(CW\*(C`qw\*(C'\fR construct may produce |
| 1706 |
both. |
| 1707 |
.IP \(bu 4 |
| 1708 |
The uninitialized warning for \f(CW\*(C`y///r\*(C'\fR when \f(CW$_\fR is implicit and |
| 1709 |
undefined now mentions the variable name, just like the non\-/r variation |
| 1710 |
of the operator. |
| 1711 |
.IP \(bu 4 |
| 1712 |
The 'Use of "foo" without parentheses is ambiguous' warning has been |
| 1713 |
extended to apply also to user-defined subroutines with a (;$) |
| 1714 |
prototype, and not just to built-in functions. |
| 1715 |
.IP \(bu 4 |
| 1716 |
Warnings that mention the names of lexical (\f(CW\*(C`my\*(C'\fR) variables with |
| 1717 |
Unicode characters in them now respect the presence or absence of the |
| 1718 |
\&\f(CW\*(C`:utf8\*(C'\fR layer on the output handle, instead of outputting UTF8 |
| 1719 |
regardless. Also, the correct names are included in the strings passed |
| 1720 |
to \f(CW$SIG{_\|_WARN_\|_}\fR handlers, rather than the raw UTF8 bytes. |
| 1721 |
.SH "Utility Changes" |
| 1722 |
.IX Header "Utility Changes" |
| 1723 |
\fIh2ph\fR |
| 1724 |
.IX Subsection "h2ph" |
| 1725 |
.IP \(bu 4 |
| 1726 |
h2ph used to generate code of the form |
| 1727 |
.Sp |
| 1728 |
.Vb 3 |
| 1729 |
\& unless(defined(&FOO)) { |
| 1730 |
\& sub FOO () {42;} |
| 1731 |
\& } |
| 1732 |
.Ve |
| 1733 |
.Sp |
| 1734 |
But the subroutine is a compile-time declaration, and is hence unaffected |
| 1735 |
by the condition. It has now been corrected to emit a string \f(CW\*(C`eval\*(C'\fR |
| 1736 |
around the subroutine [perl #99368]. |
| 1737 |
.PP |
| 1738 |
\fIsplain\fR |
| 1739 |
.IX Subsection "splain" |
| 1740 |
.IP \(bu 4 |
| 1741 |
\&\fIsplain\fR no longer emits backtraces with the first line number repeated. |
| 1742 |
.Sp |
| 1743 |
This: |
| 1744 |
.Sp |
| 1745 |
.Vb 6 |
| 1746 |
\& Uncaught exception from user code: |
| 1747 |
\& Cannot fwiddle the fwuddle at \-e line 1. |
| 1748 |
\& at \-e line 1 |
| 1749 |
\& main::baz() called at \-e line 1 |
| 1750 |
\& main::bar() called at \-e line 1 |
| 1751 |
\& main::foo() called at \-e line 1 |
| 1752 |
.Ve |
| 1753 |
.Sp |
| 1754 |
has become this: |
| 1755 |
.Sp |
| 1756 |
.Vb 5 |
| 1757 |
\& Uncaught exception from user code: |
| 1758 |
\& Cannot fwiddle the fwuddle at \-e line 1. |
| 1759 |
\& main::baz() called at \-e line 1 |
| 1760 |
\& main::bar() called at \-e line 1 |
| 1761 |
\& main::foo() called at \-e line 1 |
| 1762 |
.Ve |
| 1763 |
.IP \(bu 4 |
| 1764 |
Some error messages consist of multiple lines that are listed as separate |
| 1765 |
entries in perldiag. splain has been taught to find the separate |
| 1766 |
entries in these cases, instead of simply failing to find the message. |
| 1767 |
.PP |
| 1768 |
\fIzipdetails\fR |
| 1769 |
.IX Subsection "zipdetails" |
| 1770 |
.IP \(bu 4 |
| 1771 |
This is a new utility, included as part of an |
| 1772 |
IO::Compress::Base upgrade. |
| 1773 |
.Sp |
| 1774 |
zipdetails displays information about the internal record structure |
| 1775 |
of the zip file. It is not concerned with displaying any details of |
| 1776 |
the compressed data stored in the zip file. |
| 1777 |
.SH "Configuration and Compilation" |
| 1778 |
.IX Header "Configuration and Compilation" |
| 1779 |
.IP \(bu 4 |
| 1780 |
\&\fIregexp.h\fR has been modified for compatibility with GCC's \fB\-Werror\fR |
| 1781 |
option, as used by some projects that include perl's header files (5.14.1). |
| 1782 |
.IP \(bu 4 |
| 1783 |
\&\f(CW\*(C`USE_LOCALE{,_COLLATE,_CTYPE,_NUMERIC}\*(C'\fR have been added the output of perl \-V |
| 1784 |
as they have affect the behavior of the interpreter binary (albeit |
| 1785 |
in only a small area). |
| 1786 |
.IP \(bu 4 |
| 1787 |
The code and tests for IPC::Open2 have been moved from \fIext/IPC\-Open2\fR |
| 1788 |
into \fIext/IPC\-Open3\fR, as \f(CWIPC::Open2::open2()\fR is implemented as a thin |
| 1789 |
wrapper around \f(CWIPC::Open3::_open3()\fR, and hence is very tightly coupled to |
| 1790 |
it. |
| 1791 |
.IP \(bu 4 |
| 1792 |
The magic types and magic vtables are now generated from data in a new script |
| 1793 |
\&\fIregen/mg_vtable.pl\fR, instead of being maintained by hand. As different |
| 1794 |
EBCDIC variants can't agree on the code point for '~', the character to code |
| 1795 |
point conversion is done at build time by \fIgenerate_uudmap\fR to a new generated |
| 1796 |
header \fImg_data.h\fR. \f(CW\*(C`PL_vtbl_bm\*(C'\fR and \f(CW\*(C`PL_vtbl_fm\*(C'\fR are now defined by the |
| 1797 |
pre-processor as \f(CW\*(C`PL_vtbl_regexp\*(C'\fR, instead of being distinct C variables. |
| 1798 |
\&\f(CW\*(C`PL_vtbl_sig\*(C'\fR has been removed. |
| 1799 |
.IP \(bu 4 |
| 1800 |
Building with \f(CW\*(C`\-DPERL_GLOBAL_STRUCT\*(C'\fR works again. This configuration is not |
| 1801 |
generally used. |
| 1802 |
.IP \(bu 4 |
| 1803 |
Perl configured with \fIMAD\fR now correctly frees \f(CW\*(C`MADPROP\*(C'\fR structures when |
| 1804 |
OPs are freed. \f(CW\*(C`MADPROP\*(C'\fRs are now allocated with \f(CWPerlMemShared_malloc()\fR |
| 1805 |
.IP \(bu 4 |
| 1806 |
\&\fImakedef.pl\fR has been refactored. This should have no noticeable affect on |
| 1807 |
any of the platforms that use it as part of their build (AIX, VMS, Win32). |
| 1808 |
.IP \(bu 4 |
| 1809 |
\&\f(CW\*(C`useperlio\*(C'\fR can no longer be disabled. |
| 1810 |
.IP \(bu 4 |
| 1811 |
The file \fIglobal.sym\fR is no longer needed, and has been removed. It |
| 1812 |
contained a list of all exported functions, one of the files generated by |
| 1813 |
\&\fIregen/embed.pl\fR from data in \fIembed.fnc\fR and \fIregen/opcodes\fR. The code |
| 1814 |
has been refactored so that the only user of \fIglobal.sym\fR, \fImakedef.pl\fR, |
| 1815 |
now reads \fIembed.fnc\fR and \fIregen/opcodes\fR directly, removing the need to |
| 1816 |
store the list of exported functions in an intermediate file. |
| 1817 |
.Sp |
| 1818 |
As \fIglobal.sym\fR was never installed, this change should not be visible |
| 1819 |
outside the build process. |
| 1820 |
.IP \(bu 4 |
| 1821 |
\&\fIpod/buildtoc\fR, used by the build process to build perltoc, has been |
| 1822 |
refactored and simplified. It now contains only code to build perltoc; |
| 1823 |
the code to regenerate Makefiles has been moved to \fIPorting/pod_rules.pl\fR. |
| 1824 |
It's a bug if this change has any material effect on the build process. |
| 1825 |
.IP \(bu 4 |
| 1826 |
\&\fIpod/roffitall\fR is now built by \fIpod/buildtoc\fR, instead of being |
| 1827 |
shipped with the distribution. Its list of manpages is now generated |
| 1828 |
(and therefore current). See also RT #103202 for an unresolved related |
| 1829 |
issue. |
| 1830 |
.IP \(bu 4 |
| 1831 |
The man page for \f(CW\*(C`XS::Typemap\*(C'\fR is no longer installed. \f(CW\*(C`XS::Typemap\*(C'\fR |
| 1832 |
is a test module which is not installed, hence installing its |
| 1833 |
documentation makes no sense. |
| 1834 |
.IP \(bu 4 |
| 1835 |
The \-Dusesitecustomize and \-Duserelocatableinc options now work |
| 1836 |
together properly. |
| 1837 |
.SH "Platform Support" |
| 1838 |
.IX Header "Platform Support" |
| 1839 |
.SS "Platform-Specific Notes" |
| 1840 |
.IX Subsection "Platform-Specific Notes" |
| 1841 |
\fICygwin\fR |
| 1842 |
.IX Subsection "Cygwin" |
| 1843 |
.IP \(bu 4 |
| 1844 |
Since version 1.7, Cygwin supports native UTF\-8 paths. If Perl is built |
| 1845 |
under that environment, directory and filenames will be UTF\-8 encoded. |
| 1846 |
.IP \(bu 4 |
| 1847 |
Cygwin does not initialize all original Win32 environment variables. See |
| 1848 |
\&\fIREADME.cygwin\fR for a discussion of the newly-added |
| 1849 |
\&\f(CWCygwin::sync_winenv()\fR function [perl #110190] and for |
| 1850 |
further links. |
| 1851 |
.PP |
| 1852 |
\fIHP-UX\fR |
| 1853 |
.IX Subsection "HP-UX" |
| 1854 |
.IP \(bu 4 |
| 1855 |
HP-UX PA\-RISC/64 now supports gcc\-4.x |
| 1856 |
.Sp |
| 1857 |
A fix to correct the socketsize now makes the test suite pass on HP-UX |
| 1858 |
PA-RISC for 64bitall builds. (5.14.2) |
| 1859 |
.PP |
| 1860 |
\fIVMS\fR |
| 1861 |
.IX Subsection "VMS" |
| 1862 |
.IP \(bu 4 |
| 1863 |
Remove unnecessary includes, fix miscellaneous compiler warnings and |
| 1864 |
close some unclosed comments on \fIvms/vms.c\fR. |
| 1865 |
.IP \(bu 4 |
| 1866 |
Remove sockadapt layer from the VMS build. |
| 1867 |
.IP \(bu 4 |
| 1868 |
Explicit support for VMS versions before v7.0 and DEC C versions |
| 1869 |
before v6.0 has been removed. |
| 1870 |
.IP \(bu 4 |
| 1871 |
Since Perl 5.10.1, the home-grown \f(CW\*(C`stat\*(C'\fR wrapper has been unable to |
| 1872 |
distinguish between a directory name containing an underscore and an |
| 1873 |
otherwise-identical filename containing a dot in the same position |
| 1874 |
(e.g., t/test_pl as a directory and t/test.pl as a file). This problem |
| 1875 |
has been corrected. |
| 1876 |
.IP \(bu 4 |
| 1877 |
The build on VMS now permits names of the resulting symbols in C code for |
| 1878 |
Perl longer than 31 characters. Symbols like |
| 1879 |
\&\f(CW\*(C`Perl_\|_it_was_the_best_of_times_it_was_the_worst_of_times\*(C'\fR can now be |
| 1880 |
created freely without causing the VMS linker to seize up. |
| 1881 |
.PP |
| 1882 |
\fIGNU/Hurd\fR |
| 1883 |
.IX Subsection "GNU/Hurd" |
| 1884 |
.IP \(bu 4 |
| 1885 |
Numerous build and test failures on GNU/Hurd have been resolved with hints |
| 1886 |
for building DBM modules, detection of the library search path, and enabling |
| 1887 |
of large file support. |
| 1888 |
.PP |
| 1889 |
\fIOpenVOS\fR |
| 1890 |
.IX Subsection "OpenVOS" |
| 1891 |
.IP \(bu 4 |
| 1892 |
Perl is now built with dynamic linking on OpenVOS, the minimum supported |
| 1893 |
version of which is now Release 17.1.0. |
| 1894 |
.PP |
| 1895 |
\fISunOS\fR |
| 1896 |
.IX Subsection "SunOS" |
| 1897 |
.PP |
| 1898 |
The CC workshop C++ compiler is now detected and used on systems that ship |
| 1899 |
without cc. |
| 1900 |
.SH "Internal Changes" |
| 1901 |
.IX Header "Internal Changes" |
| 1902 |
.IP \(bu 4 |
| 1903 |
The compiled representation of formats is now stored via the \f(CW\*(C`mg_ptr\*(C'\fR of |
| 1904 |
their \f(CW\*(C`PERL_MAGIC_fm\*(C'\fR. Previously it was stored in the string buffer, |
| 1905 |
beyond \f(CWSvLEN()\fR, the regular end of the string. \f(CWSvCOMPILED()\fR and |
| 1906 |
\&\f(CW\*(C`SvCOMPILED_{on,off}()\*(C'\fR now exist solely for compatibility for XS code. |
| 1907 |
The first is always 0, the other two now no-ops. (5.14.1) |
| 1908 |
.IP \(bu 4 |
| 1909 |
Some global variables have been marked \f(CW\*(C`const\*(C'\fR, members in the interpreter |
| 1910 |
structure have been re-ordered, and the opcodes have been re-ordered. The |
| 1911 |
op \f(CW\*(C`OP_AELEMFAST\*(C'\fR has been split into \f(CW\*(C`OP_AELEMFAST\*(C'\fR and \f(CW\*(C`OP_AELEMFAST_LEX\*(C'\fR. |
| 1912 |
.IP \(bu 4 |
| 1913 |
When empting a hash of its elements (e.g., via undef(%h), or \f(CW%h\fR=()), HvARRAY |
| 1914 |
field is no longer temporarily zeroed. Any destructors called on the freed |
| 1915 |
elements see the remaining elements. Thus, \f(CW%h\fR=() becomes more like |
| 1916 |
\&\f(CW\*(C`delete $h{$_} for keys %h\*(C'\fR. |
| 1917 |
.IP \(bu 4 |
| 1918 |
Boyer-Moore compiled scalars are now PVMGs, and the Boyer-Moore tables are now |
| 1919 |
stored via the mg_ptr of their \f(CW\*(C`PERL_MAGIC_bm\*(C'\fR. |
| 1920 |
Previously they were PVGVs, with the tables stored in |
| 1921 |
the string buffer, beyond \f(CWSvLEN()\fR. This eliminates |
| 1922 |
the last place where the core stores data beyond \f(CWSvLEN()\fR. |
| 1923 |
.IP \(bu 4 |
| 1924 |
Simplified logic in \f(CWPerl_sv_magic()\fR introduces a small change of |
| 1925 |
behavior for error cases involving unknown magic types. Previously, if |
| 1926 |
\&\f(CWPerl_sv_magic()\fR was passed a magic type unknown to it, it would |
| 1927 |
.RS 4 |
| 1928 |
.IP 1. 4 |
| 1929 |
Croak "Modification of a read-only value attempted" if read only |
| 1930 |
.IP 2. 4 |
| 1931 |
Return without error if the SV happened to already have this magic |
| 1932 |
.IP 3. 4 |
| 1933 |
otherwise croak "Don't know how to handle magic of type \e\e%o" |
| 1934 |
.RE |
| 1935 |
.RS 4 |
| 1936 |
.Sp |
| 1937 |
Now it will always croak "Don't know how to handle magic of type \e\e%o", even |
| 1938 |
on read-only values, or SVs which already have the unknown magic type. |
| 1939 |
.RE |
| 1940 |
.IP \(bu 4 |
| 1941 |
The experimental \f(CW\*(C`fetch_cop_label\*(C'\fR function has been renamed to |
| 1942 |
\&\f(CW\*(C`cop_fetch_label\*(C'\fR. |
| 1943 |
.IP \(bu 4 |
| 1944 |
The \f(CW\*(C`cop_store_label\*(C'\fR function has been added to the API, but is |
| 1945 |
experimental. |
| 1946 |
.IP \(bu 4 |
| 1947 |
\&\fIembedvar.h\fR has been simplified, and one level of macro indirection for |
| 1948 |
PL_* variables has been removed for the default (non-multiplicity) |
| 1949 |
configuration. PERLVAR*() macros now directly expand their arguments to |
| 1950 |
tokens such as \f(CW\*(C`PL_defgv\*(C'\fR, instead of expanding to \f(CW\*(C`PL_Idefgv\*(C'\fR, with |
| 1951 |
\&\fIembedvar.h\fR defining a macro to map \f(CW\*(C`PL_Idefgv\*(C'\fR to \f(CW\*(C`PL_defgv\*(C'\fR. XS code |
| 1952 |
which has unwarranted chumminess with the implementation may need updating. |
| 1953 |
.IP \(bu 4 |
| 1954 |
An API has been added to explicitly choose whether to export XSUB |
| 1955 |
symbols. More detail can be found in the comments for commit e64345f8. |
| 1956 |
.IP \(bu 4 |
| 1957 |
The \f(CW\*(C`is_gv_magical_sv\*(C'\fR function has been eliminated and merged with |
| 1958 |
\&\f(CW\*(C`gv_fetchpvn_flags\*(C'\fR. It used to be called to determine whether a GV |
| 1959 |
should be autovivified in rvalue context. Now it has been replaced with a |
| 1960 |
new \f(CW\*(C`GV_ADDMG\*(C'\fR flag (not part of the API). |
| 1961 |
.IP \(bu 4 |
| 1962 |
The returned code point from the function \f(CWutf8n_to_uvuni()\fR |
| 1963 |
when the input is malformed UTF\-8, malformations are allowed, and |
| 1964 |
\&\f(CW\*(C`utf8\*(C'\fR warnings are off is now the Unicode REPLACEMENT CHARACTER |
| 1965 |
whenever the malformation is such that no well-defined code point can be |
| 1966 |
computed. Previously the returned value was essentially garbage. The |
| 1967 |
only malformations that have well-defined values are a zero-length |
| 1968 |
string (0 is the return), and overlong UTF\-8 sequences. |
| 1969 |
.IP \(bu 4 |
| 1970 |
Padlists are now marked \f(CW\*(C`AvREAL\*(C'\fR; i.e., reference-counted. They have |
| 1971 |
always been reference-counted, but were not marked real, because \fIpad.c\fR |
| 1972 |
did its own clean-up, instead of using the usual clean-up code in \fIsv.c\fR. |
| 1973 |
That caused problems in thread cloning, so now the \f(CW\*(C`AvREAL\*(C'\fR flag is on, |
| 1974 |
but is turned off in \fIpad.c\fR right before the padlist is freed (after |
| 1975 |
\&\fIpad.c\fR has done its custom freeing of the pads). |
| 1976 |
.IP \(bu 4 |
| 1977 |
All C files that make up the Perl core have been converted to UTF\-8. |
| 1978 |
.IP \(bu 4 |
| 1979 |
These new functions have been added as part of the work on Unicode symbols: |
| 1980 |
.Sp |
| 1981 |
.Vb 10 |
| 1982 |
\& HvNAMELEN |
| 1983 |
\& HvNAMEUTF8 |
| 1984 |
\& HvENAMELEN |
| 1985 |
\& HvENAMEUTF8 |
| 1986 |
\& gv_init_pv |
| 1987 |
\& gv_init_pvn |
| 1988 |
\& gv_init_pvsv |
| 1989 |
\& gv_fetchmeth_pv |
| 1990 |
\& gv_fetchmeth_pvn |
| 1991 |
\& gv_fetchmeth_sv |
| 1992 |
\& gv_fetchmeth_pv_autoload |
| 1993 |
\& gv_fetchmeth_pvn_autoload |
| 1994 |
\& gv_fetchmeth_sv_autoload |
| 1995 |
\& gv_fetchmethod_pv_flags |
| 1996 |
\& gv_fetchmethod_pvn_flags |
| 1997 |
\& gv_fetchmethod_sv_flags |
| 1998 |
\& gv_autoload_pv |
| 1999 |
\& gv_autoload_pvn |
| 2000 |
\& gv_autoload_sv |
| 2001 |
\& newGVgen_flags |
| 2002 |
\& sv_derived_from_pv |
| 2003 |
\& sv_derived_from_pvn |
| 2004 |
\& sv_derived_from_sv |
| 2005 |
\& sv_does_pv |
| 2006 |
\& sv_does_pvn |
| 2007 |
\& sv_does_sv |
| 2008 |
\& whichsig_pv |
| 2009 |
\& whichsig_pvn |
| 2010 |
\& whichsig_sv |
| 2011 |
\& newCONSTSUB_flags |
| 2012 |
.Ve |
| 2013 |
.Sp |
| 2014 |
The gv_fetchmethod_*_flags functions, like gv_fetchmethod_flags, are |
| 2015 |
experimental and may change in a future release. |
| 2016 |
.IP \(bu 4 |
| 2017 |
The following functions were added. These are \fInot\fR part of the API: |
| 2018 |
.Sp |
| 2019 |
.Vb 9 |
| 2020 |
\& GvNAMEUTF8 |
| 2021 |
\& GvENAMELEN |
| 2022 |
\& GvENAME_HEK |
| 2023 |
\& CopSTASH_flags |
| 2024 |
\& CopSTASH_flags_set |
| 2025 |
\& PmopSTASH_flags |
| 2026 |
\& PmopSTASH_flags_set |
| 2027 |
\& sv_sethek |
| 2028 |
\& HEKfARG |
| 2029 |
.Ve |
| 2030 |
.Sp |
| 2031 |
There is also a \f(CW\*(C`HEKf\*(C'\fR macro corresponding to \f(CW\*(C`SVf\*(C'\fR, for |
| 2032 |
interpolating HEKs in formatted strings. |
| 2033 |
.IP \(bu 4 |
| 2034 |
\&\f(CW\*(C`sv_catpvn_flags\*(C'\fR takes a couple of new internal-only flags, |
| 2035 |
\&\f(CW\*(C`SV_CATBYTES\*(C'\fR and \f(CW\*(C`SV_CATUTF8\*(C'\fR, which tell it whether the char array to |
| 2036 |
be concatenated is UTF8. This allows for more efficient concatenation than |
| 2037 |
creating temporary SVs to pass to \f(CW\*(C`sv_catsv\*(C'\fR. |
| 2038 |
.IP \(bu 4 |
| 2039 |
For XS AUTOLOAD subs, \f(CW$AUTOLOAD\fR is set once more, as it was in 5.6.0. This |
| 2040 |
is in addition to setting \f(CWSvPVX(cv)\fR, for compatibility with 5.8 to 5.14. |
| 2041 |
See "Autoloading with XSUBs" in perlguts. |
| 2042 |
.IP \(bu 4 |
| 2043 |
Perl now checks whether the array (the linearized isa) returned by a MRO |
| 2044 |
plugin begins with the name of the class itself, for which the array was |
| 2045 |
created, instead of assuming that it does. This prevents the first element |
| 2046 |
from being skipped during method lookup. It also means that |
| 2047 |
\&\f(CW\*(C`mro::get_linear_isa\*(C'\fR may return an array with one more element than the |
| 2048 |
MRO plugin provided [perl #94306]. |
| 2049 |
.IP \(bu 4 |
| 2050 |
\&\f(CW\*(C`PL_curstash\*(C'\fR is now reference-counted. |
| 2051 |
.IP \(bu 4 |
| 2052 |
There are now feature bundle hints in \f(CW\*(C`PL_hints\*(C'\fR (\f(CW$^H\fR) that version |
| 2053 |
declarations use, to avoid having to load \fIfeature.pm\fR. One setting of |
| 2054 |
the hint bits indicates a "custom" feature bundle, which means that the |
| 2055 |
entries in \f(CW\*(C`%^H\*(C'\fR still apply. \fIfeature.pm\fR uses that. |
| 2056 |
.Sp |
| 2057 |
The \f(CW\*(C`HINT_FEATURE_MASK\*(C'\fR macro is defined in \fIperl.h\fR along with other |
| 2058 |
hints. Other macros for setting and testing features and bundles are in |
| 2059 |
the new \fIfeature.h\fR. \f(CW\*(C`FEATURE_IS_ENABLED\*(C'\fR (which has moved to |
| 2060 |
\&\fIfeature.h\fR) is no longer used throughout the codebase, but more specific |
| 2061 |
macros, e.g., \f(CW\*(C`FEATURE_SAY_IS_ENABLED\*(C'\fR, that are defined in \fIfeature.h\fR. |
| 2062 |
.IP \(bu 4 |
| 2063 |
\&\fIlib/feature.pm\fR is now a generated file, created by the new |
| 2064 |
\&\fIregen/feature.pl\fR script, which also generates \fIfeature.h\fR. |
| 2065 |
.IP \(bu 4 |
| 2066 |
Tied arrays are now always \f(CW\*(C`AvREAL\*(C'\fR. If \f(CW@_\fR or \f(CW\*(C`DB::args\*(C'\fR is tied, it |
| 2067 |
is reified first, to make sure this is always the case. |
| 2068 |
.IP \(bu 4 |
| 2069 |
Two new functions \f(CWutf8_to_uvchr_buf()\fR and \f(CWutf8_to_uvuni_buf()\fR have |
| 2070 |
been added. These are the same as \f(CW\*(C`utf8_to_uvchr\*(C'\fR and |
| 2071 |
\&\f(CW\*(C`utf8_to_uvuni\*(C'\fR (which are now deprecated), but take an extra parameter |
| 2072 |
that is used to guard against reading beyond the end of the input |
| 2073 |
string. |
| 2074 |
See "utf8_to_uvchr_buf" in perlapi and "utf8_to_uvuni_buf" in perlapi. |
| 2075 |
.IP \(bu 4 |
| 2076 |
The regular expression engine now does TRIE case insensitive matches |
| 2077 |
under Unicode. This may change the output of \f(CW\*(C`use re \*(Aqdebug\*(Aq;\*(C'\fR, |
| 2078 |
and will speed up various things. |
| 2079 |
.IP \(bu 4 |
| 2080 |
There is a new \f(CWwrap_op_checker()\fR function, which provides a thread-safe |
| 2081 |
alternative to writing to \f(CW\*(C`PL_check\*(C'\fR directly. |
| 2082 |
.SH "Selected Bug Fixes" |
| 2083 |
.IX Header "Selected Bug Fixes" |
| 2084 |
.SS "Array and hash" |
| 2085 |
.IX Subsection "Array and hash" |
| 2086 |
.IP \(bu 4 |
| 2087 |
A bug has been fixed that would cause a "Use of freed value in iteration" |
| 2088 |
error if the next two hash elements that would be iterated over are |
| 2089 |
deleted [perl #85026]. (5.14.1) |
| 2090 |
.IP \(bu 4 |
| 2091 |
Deleting the current hash iterator (the hash element that would be returned |
| 2092 |
by the next call to \f(CW\*(C`each\*(C'\fR) in void context used not to free it |
| 2093 |
[perl #85026]. |
| 2094 |
.IP \(bu 4 |
| 2095 |
Deletion of methods via \f(CW\*(C`delete $Class::{method}\*(C'\fR syntax used to update |
| 2096 |
method caches if called in void context, but not scalar or list context. |
| 2097 |
.IP \(bu 4 |
| 2098 |
When hash elements are deleted in void context, the internal hash entry is |
| 2099 |
now freed before the value is freed, to prevent destructors called by that |
| 2100 |
latter freeing from seeing the hash in an inconsistent state. It was |
| 2101 |
possible to cause double-frees if the destructor freed the hash itself |
| 2102 |
[perl #100340]. |
| 2103 |
.IP \(bu 4 |
| 2104 |
A \f(CW\*(C`keys\*(C'\fR optimization in Perl 5.12.0 to make it faster on empty hashes |
| 2105 |
caused \f(CW\*(C`each\*(C'\fR not to reset the iterator if called after the last element |
| 2106 |
was deleted. |
| 2107 |
.IP \(bu 4 |
| 2108 |
Freeing deeply nested hashes no longer crashes [perl #44225]. |
| 2109 |
.IP \(bu 4 |
| 2110 |
It is possible from XS code to create hashes with elements that have no |
| 2111 |
values. The hash element and slice operators used to crash |
| 2112 |
when handling these in lvalue context. They now |
| 2113 |
produce a "Modification of non-creatable hash value attempted" error |
| 2114 |
message. |
| 2115 |
.IP \(bu 4 |
| 2116 |
If list assignment to a hash or array triggered destructors that freed the |
| 2117 |
hash or array itself, a crash would ensue. This is no longer the case |
| 2118 |
[perl #107440]. |
| 2119 |
.IP \(bu 4 |
| 2120 |
It used to be possible to free the typeglob of a localized array or hash |
| 2121 |
(e.g., \f(CW\*(C`local @{"x"}; delete $::{x}\*(C'\fR), resulting in a crash on scope exit. |
| 2122 |
.IP \(bu 4 |
| 2123 |
Some core bugs affecting Hash::Util have been fixed: locking a hash |
| 2124 |
element that is a glob copy no longer causes the next assignment to it to |
| 2125 |
corrupt the glob (5.14.2), and unlocking a hash element that holds a |
| 2126 |
copy-on-write scalar no longer causes modifications to that scalar to |
| 2127 |
modify other scalars that were sharing the same string buffer. |
| 2128 |
.SS "C API fixes" |
| 2129 |
.IX Subsection "C API fixes" |
| 2130 |
.IP \(bu 4 |
| 2131 |
The \f(CW\*(C`newHVhv\*(C'\fR XS function now works on tied hashes, instead of crashing or |
| 2132 |
returning an empty hash. |
| 2133 |
.IP \(bu 4 |
| 2134 |
The \f(CW\*(C`SvIsCOW\*(C'\fR C macro now returns false for read-only copies of typeglobs, |
| 2135 |
such as those created by: |
| 2136 |
.Sp |
| 2137 |
.Vb 2 |
| 2138 |
\& $hash{elem} = *foo; |
| 2139 |
\& Hash::Util::lock_value %hash, \*(Aqelem\*(Aq; |
| 2140 |
.Ve |
| 2141 |
.Sp |
| 2142 |
It used to return true. |
| 2143 |
.IP \(bu 4 |
| 2144 |
The \f(CW\*(C`SvPVutf8\*(C'\fR C function no longer tries to modify its argument, |
| 2145 |
resulting in errors [perl #108994]. |
| 2146 |
.IP \(bu 4 |
| 2147 |
\&\f(CW\*(C`SvPVutf8\*(C'\fR now works properly with magical variables. |
| 2148 |
.IP \(bu 4 |
| 2149 |
\&\f(CW\*(C`SvPVbyte\*(C'\fR now works properly non-PVs. |
| 2150 |
.IP \(bu 4 |
| 2151 |
When presented with malformed UTF\-8 input, the XS-callable functions |
| 2152 |
\&\f(CWis_utf8_string()\fR, \f(CWis_utf8_string_loc()\fR, and |
| 2153 |
\&\f(CWis_utf8_string_loclen()\fR could read beyond the end of the input |
| 2154 |
string by up to 12 bytes. This no longer happens. [perl #32080]. |
| 2155 |
However, currently, \f(CWis_utf8_char()\fR still has this defect, see |
| 2156 |
"\fBis_utf8_char()\fR" above. |
| 2157 |
.IP \(bu 4 |
| 2158 |
The C\-level \f(CW\*(C`pregcomp\*(C'\fR function could become confused about whether the |
| 2159 |
pattern was in UTF8 if the pattern was an overloaded, tied, or otherwise |
| 2160 |
magical scalar [perl #101940]. |
| 2161 |
.SS "Compile-time hints" |
| 2162 |
.IX Subsection "Compile-time hints" |
| 2163 |
.IP \(bu 4 |
| 2164 |
Tying \f(CW\*(C`%^H\*(C'\fR no longer causes perl to crash or ignore the contents of |
| 2165 |
\&\f(CW\*(C`%^H\*(C'\fR when entering a compilation scope [perl #106282]. |
| 2166 |
.IP \(bu 4 |
| 2167 |
\&\f(CW\*(C`eval $string\*(C'\fR and \f(CW\*(C`require\*(C'\fR used not to |
| 2168 |
localize \f(CW\*(C`%^H\*(C'\fR during compilation if it |
| 2169 |
was empty at the time the \f(CW\*(C`eval\*(C'\fR call itself was compiled. This could |
| 2170 |
lead to scary side effects, like \f(CW\*(C`use re "/m"\*(C'\fR enabling other flags that |
| 2171 |
the surrounding code was trying to enable for its caller [perl #68750]. |
| 2172 |
.IP \(bu 4 |
| 2173 |
\&\f(CW\*(C`eval $string\*(C'\fR and \f(CW\*(C`require\*(C'\fR no longer localize hints (\f(CW$^H\fR and \f(CW\*(C`%^H\*(C'\fR) |
| 2174 |
at run time, but only during compilation of the \f(CW$string\fR or required file. |
| 2175 |
This makes \f(CW\*(C`BEGIN { $^H{foo}=7 }\*(C'\fR equivalent to |
| 2176 |
\&\f(CW\*(C`BEGIN { eval \*(Aq$^H{foo}=7\*(Aq }\*(C'\fR [perl #70151]. |
| 2177 |
.IP \(bu 4 |
| 2178 |
Creating a BEGIN block from XS code (via \f(CW\*(C`newXS\*(C'\fR or \f(CW\*(C`newATTRSUB\*(C'\fR) would, |
| 2179 |
on completion, make the hints of the current compiling code the current |
| 2180 |
hints. This could cause warnings to occur in a non-warning scope. |
| 2181 |
.SS "Copy-on-write scalars" |
| 2182 |
.IX Subsection "Copy-on-write scalars" |
| 2183 |
Copy-on-write or shared hash key scalars |
| 2184 |
were introduced in 5.8.0, but most Perl code |
| 2185 |
did not encounter them (they were used mostly internally). Perl |
| 2186 |
5.10.0 extended them, such that assigning \f(CW\*(C`_\|_PACKAGE_\|_\*(C'\fR or a |
| 2187 |
hash key to a scalar would make it copy-on-write. Several parts |
| 2188 |
of Perl were not updated to account for them, but have now been fixed. |
| 2189 |
.IP \(bu 4 |
| 2190 |
\&\f(CW\*(C`utf8::decode\*(C'\fR had a nasty bug that would modify copy-on-write scalars' |
| 2191 |
string buffers in place (i.e., skipping the copy). This could result in |
| 2192 |
hashes having two elements with the same key [perl #91834]. (5.14.2) |
| 2193 |
.IP \(bu 4 |
| 2194 |
Lvalue subroutines were not allowing COW scalars to be returned. This was |
| 2195 |
fixed for lvalue scalar context in Perl 5.12.3 and 5.14.0, but list context |
| 2196 |
was not fixed until this release. |
| 2197 |
.IP \(bu 4 |
| 2198 |
Elements of restricted hashes (see the fields pragma) containing |
| 2199 |
copy-on-write values couldn't be deleted, nor could such hashes be cleared |
| 2200 |
(\f(CW\*(C`%hash = ()\*(C'\fR). (5.14.2) |
| 2201 |
.IP \(bu 4 |
| 2202 |
Localizing a tied variable used to make it read-only if it contained a |
| 2203 |
copy-on-write string. (5.14.2) |
| 2204 |
.IP \(bu 4 |
| 2205 |
Assigning a copy-on-write string to a stash |
| 2206 |
element no longer causes a double free. Regardless of this change, the |
| 2207 |
results of such assignments are still undefined. |
| 2208 |
.IP \(bu 4 |
| 2209 |
Assigning a copy-on-write string to a tied variable no longer stops that |
| 2210 |
variable from being tied if it happens to be a PVMG or PVLV internally. |
| 2211 |
.IP \(bu 4 |
| 2212 |
Doing a substitution on a tied variable returning a copy-on-write |
| 2213 |
scalar used to cause an assertion failure or an "Attempt to free |
| 2214 |
nonexistent shared string" warning. |
| 2215 |
.IP \(bu 4 |
| 2216 |
This one is a regression from 5.12: In 5.14.0, the bitwise assignment |
| 2217 |
operators \f(CW\*(C`|=\*(C'\fR, \f(CW\*(C`^=\*(C'\fR and \f(CW\*(C`&=\*(C'\fR started leaving the left-hand side |
| 2218 |
undefined if it happened to be a copy-on-write string [perl #108480]. |
| 2219 |
.IP \(bu 4 |
| 2220 |
Storable, Devel::Peek and PerlIO::scalar had similar problems. |
| 2221 |
See "Updated Modules and Pragmata", above. |
| 2222 |
.SS "The debugger" |
| 2223 |
.IX Subsection "The debugger" |
| 2224 |
.IP \(bu 4 |
| 2225 |
\&\fIdumpvar.pl\fR, and therefore the \f(CW\*(C`x\*(C'\fR command in the debugger, have been |
| 2226 |
fixed to handle objects blessed into classes whose names contain "=". The |
| 2227 |
contents of such objects used not to be dumped [perl #101814]. |
| 2228 |
.IP \(bu 4 |
| 2229 |
The "R" command for restarting a debugger session has been fixed to work on |
| 2230 |
Windows, or any other system lacking a \f(CW\*(C`POSIX::_SC_OPEN_MAX\*(C'\fR constant |
| 2231 |
[perl #87740]. |
| 2232 |
.IP \(bu 4 |
| 2233 |
The \f(CW\*(C`#line 42 foo\*(C'\fR directive used not to update the arrays of lines used |
| 2234 |
by the debugger if it occurred in a string eval. This was partially fixed |
| 2235 |
in 5.14, but it worked only for a single \f(CW\*(C`#line 42 foo\*(C'\fR in each eval. Now |
| 2236 |
it works for multiple. |
| 2237 |
.IP \(bu 4 |
| 2238 |
When subroutine calls are intercepted by the debugger, the name of the |
| 2239 |
subroutine or a reference to it is stored in \f(CW$DB::sub\fR, for the debugger |
| 2240 |
to access. Sometimes (such as \f(CW\*(C`$foo = *bar; undef *bar; &$foo\*(C'\fR) |
| 2241 |
\&\f(CW$DB::sub\fR would be set to a name that could not be used to find the |
| 2242 |
subroutine, and so the debugger's attempt to call it would fail. Now the |
| 2243 |
check to see whether a reference is needed is more robust, so those |
| 2244 |
problems should not happen anymore [rt.cpan.org #69862]. |
| 2245 |
.IP \(bu 4 |
| 2246 |
Every subroutine has a filename associated with it that the debugger uses. |
| 2247 |
The one associated with constant subroutines used to be misallocated when |
| 2248 |
cloned under threads. Consequently, debugging threaded applications could |
| 2249 |
result in memory corruption [perl #96126]. |
| 2250 |
.SS "Dereferencing operators" |
| 2251 |
.IX Subsection "Dereferencing operators" |
| 2252 |
.IP \(bu 4 |
| 2253 |
\&\f(CWdefined(${"..."})\fR, \f(CWdefined(*{"..."})\fR, etc., used to |
| 2254 |
return true for most, but not all built-in variables, if |
| 2255 |
they had not been used yet. This bug affected \f(CW\*(C`${^GLOBAL_PHASE}\*(C'\fR and |
| 2256 |
\&\f(CW\*(C`${^UTF8CACHE}\*(C'\fR, among others. It also used to return false if the |
| 2257 |
package name was given as well (\f(CW\*(C`${"::!"}\*(C'\fR) [perl #97978, #97492]. |
| 2258 |
.IP \(bu 4 |
| 2259 |
Perl 5.10.0 introduced a similar bug: \f(CWdefined(*{"foo"})\fR where "foo" |
| 2260 |
represents the name of a built-in global variable used to return false if |
| 2261 |
the variable had never been used before, but only on the \fIfirst\fR call. |
| 2262 |
This, too, has been fixed. |
| 2263 |
.IP \(bu 4 |
| 2264 |
Since 5.6.0, \f(CW\*(C`*{ ... }\*(C'\fR has been inconsistent in how it treats undefined |
| 2265 |
values. It would die in strict mode or lvalue context for most undefined |
| 2266 |
values, but would be treated as the empty string (with a warning) for the |
| 2267 |
specific scalar return by \f(CWundef()\fR (\f(CW&PL_sv_undef\fR internally). This |
| 2268 |
has been corrected. \f(CWundef()\fR is now treated like other undefined |
| 2269 |
scalars, as in Perl 5.005. |
| 2270 |
.SS "Filehandle, last-accessed" |
| 2271 |
.IX Subsection "Filehandle, last-accessed" |
| 2272 |
Perl has an internal variable that stores the last filehandle to be |
| 2273 |
accessed. It is used by \f(CW$.\fR and by \f(CW\*(C`tell\*(C'\fR and \f(CW\*(C`eof\*(C'\fR without |
| 2274 |
arguments. |
| 2275 |
.IP \(bu 4 |
| 2276 |
It used to be possible to set this internal variable to a glob copy and |
| 2277 |
then modify that glob copy to be something other than a glob, and still |
| 2278 |
have the last-accessed filehandle associated with the variable after |
| 2279 |
assigning a glob to it again: |
| 2280 |
.Sp |
| 2281 |
.Vb 4 |
| 2282 |
\& my $foo = *STDOUT; # $foo is a glob copy |
| 2283 |
\& <$foo>; # $foo is now the last\-accessed handle |
| 2284 |
\& $foo = 3; # no longer a glob |
| 2285 |
\& $foo = *STDERR; # still the last\-accessed handle |
| 2286 |
.Ve |
| 2287 |
.Sp |
| 2288 |
Now the \f(CW\*(C`$foo = 3\*(C'\fR assignment unsets that internal variable, so there |
| 2289 |
is no last-accessed filehandle, just as if \f(CW\*(C`<$foo>\*(C'\fR had never |
| 2290 |
happened. |
| 2291 |
.Sp |
| 2292 |
This also prevents some unrelated handle from becoming the last-accessed |
| 2293 |
handle if \f(CW$foo\fR falls out of scope and the same internal SV gets used for |
| 2294 |
another handle [perl #97988]. |
| 2295 |
.IP \(bu 4 |
| 2296 |
A regression in 5.14 caused these statements not to set that internal |
| 2297 |
variable: |
| 2298 |
.Sp |
| 2299 |
.Vb 8 |
| 2300 |
\& my $fh = *STDOUT; |
| 2301 |
\& tell $fh; |
| 2302 |
\& eof $fh; |
| 2303 |
\& seek $fh, 0,0; |
| 2304 |
\& tell *$fh; |
| 2305 |
\& eof *$fh; |
| 2306 |
\& seek *$fh, 0,0; |
| 2307 |
\& readline *$fh; |
| 2308 |
.Ve |
| 2309 |
.Sp |
| 2310 |
This is now fixed, but \f(CW\*(C`tell *{ *$fh }\*(C'\fR still has the problem, and it |
| 2311 |
is not clear how to fix it [perl #106536]. |
| 2312 |
.ie n .SS "Filetests and ""stat""" |
| 2313 |
.el .SS "Filetests and \f(CWstat\fP" |
| 2314 |
.IX Subsection "Filetests and stat" |
| 2315 |
The term "filetests" refers to the operators that consist of a hyphen |
| 2316 |
followed by a single letter: \f(CW\*(C`\-r\*(C'\fR, \f(CW\*(C`\-x\*(C'\fR, \f(CW\*(C`\-M\*(C'\fR, etc. The term "stacked" |
| 2317 |
when applied to filetests means followed by another filetest operator |
| 2318 |
sharing the same operand, as in \f(CW\*(C`\-r \-x \-w $fooo\*(C'\fR. |
| 2319 |
.IP \(bu 4 |
| 2320 |
\&\f(CW\*(C`stat\*(C'\fR produces more consistent warnings. It no longer warns for "_" |
| 2321 |
[perl #71002] and no longer skips the warning at times for other unopened |
| 2322 |
handles. It no longer warns about an unopened handle when the operating |
| 2323 |
system's \f(CW\*(C`fstat\*(C'\fR function fails. |
| 2324 |
.IP \(bu 4 |
| 2325 |
\&\f(CW\*(C`stat\*(C'\fR would sometimes return negative numbers for large inode numbers, |
| 2326 |
because it was using the wrong internal C type. [perl #84590] |
| 2327 |
.IP \(bu 4 |
| 2328 |
\&\f(CW\*(C`lstat\*(C'\fR is documented to fall back to \f(CW\*(C`stat\*(C'\fR (with a warning) when given |
| 2329 |
a filehandle. When passed an IO reference, it was actually doing the |
| 2330 |
equivalent of \f(CW\*(C`stat\ _\*(C'\fR and ignoring the handle. |
| 2331 |
.IP \(bu 4 |
| 2332 |
\&\f(CW\*(C`\-T _\*(C'\fR with no preceding \f(CW\*(C`stat\*(C'\fR used to produce a |
| 2333 |
confusing "uninitialized" warning, even though there |
| 2334 |
is no visible uninitialized value to speak of. |
| 2335 |
.IP \(bu 4 |
| 2336 |
\&\f(CW\*(C`\-T\*(C'\fR, \f(CW\*(C`\-B\*(C'\fR, \f(CW\*(C`\-l\*(C'\fR and \f(CW\*(C`\-t\*(C'\fR now work |
| 2337 |
when stacked with other filetest operators |
| 2338 |
[perl #77388]. |
| 2339 |
.IP \(bu 4 |
| 2340 |
In 5.14.0, filetest ops (\f(CW\*(C`\-r\*(C'\fR, \f(CW\*(C`\-x\*(C'\fR, etc.) started calling FETCH on a |
| 2341 |
tied argument belonging to the previous argument to a list operator, if |
| 2342 |
called with a bareword argument or no argument at all. This has been |
| 2343 |
fixed, so \f(CW\*(C`push @foo, $tied, \-r\*(C'\fR no longer calls FETCH on \f(CW$tied\fR. |
| 2344 |
.IP \(bu 4 |
| 2345 |
In Perl 5.6, \f(CW\*(C`\-l\*(C'\fR followed by anything other than a bareword would treat |
| 2346 |
its argument as a file name. That was changed in 5.8 for glob references |
| 2347 |
(\f(CW\*(C`\e*foo\*(C'\fR), but not for globs themselves (\f(CW*foo\fR). \f(CW\*(C`\-l\*(C'\fR started |
| 2348 |
returning \f(CW\*(C`undef\*(C'\fR for glob references without setting the last |
| 2349 |
stat buffer that the "_" handle uses, but only if warnings |
| 2350 |
were turned on. With warnings off, it was the same as 5.6. |
| 2351 |
In other words, it was simply buggy and inconsistent. Now the 5.6 |
| 2352 |
behavior has been restored. |
| 2353 |
.IP \(bu 4 |
| 2354 |
\&\f(CW\*(C`\-l\*(C'\fR followed by a bareword no longer "eats" the previous argument to |
| 2355 |
the list operator in whose argument list it resides. Hence, |
| 2356 |
\&\f(CW\*(C`print "bar", \-l foo\*(C'\fR now actually prints "bar", because \f(CW\*(C`\-l\*(C'\fR |
| 2357 |
on longer eats it. |
| 2358 |
.IP \(bu 4 |
| 2359 |
Perl keeps several internal variables to keep track of the last stat |
| 2360 |
buffer, from which file(handle) it originated, what type it was, and |
| 2361 |
whether the last stat succeeded. |
| 2362 |
.Sp |
| 2363 |
There were various cases where these could get out of synch, resulting in |
| 2364 |
inconsistent or erratic behavior in edge cases (every mention of \f(CW\*(C`\-T\*(C'\fR |
| 2365 |
applies to \f(CW\*(C`\-B\*(C'\fR as well): |
| 2366 |
.RS 4 |
| 2367 |
.IP \(bu 4 |
| 2368 |
\&\f(CW\*(C`\-T \fR\f(CIHANDLE\fR\f(CW\*(C'\fR, even though it does a \f(CW\*(C`stat\*(C'\fR, was not resetting the last |
| 2369 |
stat type, so an \f(CW\*(C`lstat _\*(C'\fR following it would merrily return the wrong |
| 2370 |
results. Also, it was not setting the success status. |
| 2371 |
.IP \(bu 4 |
| 2372 |
Freeing the handle last used by \f(CW\*(C`stat\*(C'\fR or a filetest could result in |
| 2373 |
\&\f(CW\*(C`\-T\ _\*(C'\fR using an unrelated handle. |
| 2374 |
.IP \(bu 4 |
| 2375 |
\&\f(CW\*(C`stat\*(C'\fR with an IO reference would not reset the stat type or record the |
| 2376 |
filehandle for \f(CW\*(C`\-T\ _\*(C'\fR to use. |
| 2377 |
.IP \(bu 4 |
| 2378 |
Fatal warnings could cause the stat buffer not to be reset |
| 2379 |
for a filetest operator on an unopened filehandle or \f(CW\*(C`\-l\*(C'\fR on any handle. |
| 2380 |
Fatal warnings also stopped \f(CW\*(C`\-T\*(C'\fR from setting \f(CW$!\fR. |
| 2381 |
.IP \(bu 4 |
| 2382 |
When the last stat was on an unreadable file, \f(CW\*(C`\-T _\*(C'\fR is supposed to |
| 2383 |
return \f(CW\*(C`undef\*(C'\fR, leaving the last stat buffer unchanged. But it was |
| 2384 |
setting the stat type, causing \f(CW\*(C`lstat _\*(C'\fR to stop working. |
| 2385 |
.IP \(bu 4 |
| 2386 |
\&\f(CW\*(C`\-T \fR\f(CIFILENAME\fR\f(CW\*(C'\fR was not resetting the internal stat buffers for |
| 2387 |
unreadable files. |
| 2388 |
.RE |
| 2389 |
.RS 4 |
| 2390 |
.Sp |
| 2391 |
These have all been fixed. |
| 2392 |
.RE |
| 2393 |
.SS Formats |
| 2394 |
.IX Subsection "Formats" |
| 2395 |
.IP \(bu 4 |
| 2396 |
Several edge cases have been fixed with formats and \f(CW\*(C`formline\*(C'\fR; |
| 2397 |
in particular, where the format itself is potentially variable (such as |
| 2398 |
with ties and overloading), and where the format and data differ in their |
| 2399 |
encoding. In both these cases, it used to possible for the output to be |
| 2400 |
corrupted [perl #91032]. |
| 2401 |
.IP \(bu 4 |
| 2402 |
\&\f(CW\*(C`formline\*(C'\fR no longer converts its argument into a string in-place. So |
| 2403 |
passing a reference to \f(CW\*(C`formline\*(C'\fR no longer destroys the reference |
| 2404 |
[perl #79532]. |
| 2405 |
.IP \(bu 4 |
| 2406 |
Assignment to \f(CW$^A\fR (the format output accumulator) now recalculates |
| 2407 |
the number of lines output. |
| 2408 |
.ie n .SS """given"" and ""when""" |
| 2409 |
.el .SS "\f(CWgiven\fP and \f(CWwhen\fP" |
| 2410 |
.IX Subsection "given and when" |
| 2411 |
.IP \(bu 4 |
| 2412 |
\&\f(CW\*(C`given\*(C'\fR was not scoping its implicit \f(CW$_\fR properly, resulting in memory |
| 2413 |
leaks or "Variable is not available" warnings [perl #94682]. |
| 2414 |
.IP \(bu 4 |
| 2415 |
\&\f(CW\*(C`given\*(C'\fR was not calling set-magic on the implicit lexical \f(CW$_\fR that it |
| 2416 |
uses. This meant, for example, that \f(CW\*(C`pos\*(C'\fR would be remembered from one |
| 2417 |
execution of the same \f(CW\*(C`given\*(C'\fR block to the next, even if the input were a |
| 2418 |
different variable [perl #84526]. |
| 2419 |
.IP \(bu 4 |
| 2420 |
\&\f(CW\*(C`when\*(C'\fR blocks are now capable of returning variables declared inside the |
| 2421 |
enclosing \f(CW\*(C`given\*(C'\fR block [perl #93548]. |
| 2422 |
.ie n .SS "The ""glob"" operator" |
| 2423 |
.el .SS "The \f(CWglob\fP operator" |
| 2424 |
.IX Subsection "The glob operator" |
| 2425 |
.IP \(bu 4 |
| 2426 |
On OSes other than VMS, Perl's \f(CW\*(C`glob\*(C'\fR operator (and the \f(CW\*(C`<...>\*(C'\fR form) |
| 2427 |
use File::Glob underneath. File::Glob splits the pattern into words, |
| 2428 |
before feeding each word to its \f(CW\*(C`bsd_glob\*(C'\fR function. |
| 2429 |
.Sp |
| 2430 |
There were several inconsistencies in the way the split was done. Now |
| 2431 |
quotation marks (' and ") are always treated as shell-style word delimiters |
| 2432 |
(that allow whitespace as part of a word) and backslashes are always |
| 2433 |
preserved, unless they exist to escape quotation marks. Before, those |
| 2434 |
would only sometimes be the case, depending on whether the pattern |
| 2435 |
contained whitespace. Also, escaped whitespace at the end of the pattern |
| 2436 |
is no longer stripped [perl #40470]. |
| 2437 |
.IP \(bu 4 |
| 2438 |
\&\f(CW\*(C`CORE::glob\*(C'\fR now works as a way to call the default globbing function. It |
| 2439 |
used to respect overrides, despite the \f(CW\*(C`CORE::\*(C'\fR prefix. |
| 2440 |
.IP \(bu 4 |
| 2441 |
Under miniperl (used to configure modules when perl itself is built), |
| 2442 |
\&\f(CW\*(C`glob\*(C'\fR now clears \f(CW%ENV\fR before calling csh, since the latter croaks on some |
| 2443 |
systems if it does not like the contents of the LS_COLORS environment |
| 2444 |
variable [perl #98662]. |
| 2445 |
.SS "Lvalue subroutines" |
| 2446 |
.IX Subsection "Lvalue subroutines" |
| 2447 |
.IP \(bu 4 |
| 2448 |
Explicit return now returns the actual argument passed to return, instead |
| 2449 |
of copying it [perl #72724, #72706]. |
| 2450 |
.IP \(bu 4 |
| 2451 |
Lvalue subroutines used to enforce lvalue syntax (i.e., whatever can go on |
| 2452 |
the left-hand side of \f(CW\*(C`=\*(C'\fR) for the last statement and the arguments to |
| 2453 |
return. Since lvalue subroutines are not always called in lvalue context, |
| 2454 |
this restriction has been lifted. |
| 2455 |
.IP \(bu 4 |
| 2456 |
Lvalue subroutines are less restrictive about what values can be returned. |
| 2457 |
It used to croak on values returned by \f(CW\*(C`shift\*(C'\fR and \f(CW\*(C`delete\*(C'\fR and from |
| 2458 |
other subroutines, but no longer does so [perl #71172]. |
| 2459 |
.IP \(bu 4 |
| 2460 |
Empty lvalue subroutines (\f(CW\*(C`sub :lvalue {}\*(C'\fR) used to return \f(CW@_\fR in list |
| 2461 |
context. All subroutines used to do this, but regular subs were fixed in |
| 2462 |
Perl 5.8.2. Now lvalue subroutines have been likewise fixed. |
| 2463 |
.IP \(bu 4 |
| 2464 |
Autovivification now works on values returned from lvalue subroutines |
| 2465 |
[perl #7946], as does returning \f(CW\*(C`keys\*(C'\fR in lvalue context. |
| 2466 |
.IP \(bu 4 |
| 2467 |
Lvalue subroutines used to copy their return values in rvalue context. Not |
| 2468 |
only was this a waste of CPU cycles, but it also caused bugs. A \f(CW\*(C`($)\*(C'\fR |
| 2469 |
prototype would cause an lvalue sub to copy its return value [perl #51408], |
| 2470 |
and \f(CW\*(C`while(lvalue_sub() =~ m/.../g) { ... }\*(C'\fR would loop endlessly |
| 2471 |
[perl #78680]. |
| 2472 |
.IP \(bu 4 |
| 2473 |
When called in potential lvalue context |
| 2474 |
(e.g., subroutine arguments or a list |
| 2475 |
passed to \f(CW\*(C`for\*(C'\fR), lvalue subroutines used to copy |
| 2476 |
any read-only value that was returned. E.g., \f(CW\*(C` sub :lvalue { $] } \*(C'\fR |
| 2477 |
would not return \f(CW$]\fR, but a copy of it. |
| 2478 |
.IP \(bu 4 |
| 2479 |
When called in potential lvalue context, an lvalue subroutine returning |
| 2480 |
arrays or hashes used to bind the arrays or hashes to scalar variables, |
| 2481 |
resulting in bugs. This was fixed in 5.14.0 if an array were the first |
| 2482 |
thing returned from the subroutine (but not for \f(CW\*(C`$scalar, @array\*(C'\fR or |
| 2483 |
hashes being returned). Now a more general fix has been applied |
| 2484 |
[perl #23790]. |
| 2485 |
.IP \(bu 4 |
| 2486 |
Method calls whose arguments were all surrounded with \f(CWmy()\fR or \f(CWour()\fR |
| 2487 |
(as in \f(CW\*(C`$object\->method(my($a,$b))\*(C'\fR) used to force lvalue context on |
| 2488 |
the subroutine. This would prevent lvalue methods from returning certain |
| 2489 |
values. |
| 2490 |
.IP \(bu 4 |
| 2491 |
Lvalue sub calls that are not determined to be such at compile time |
| 2492 |
(\f(CW&$name\fR or &{"name"}) are no longer exempt from strict refs if they |
| 2493 |
occur in the last statement of an lvalue subroutine [perl #102486]. |
| 2494 |
.IP \(bu 4 |
| 2495 |
Sub calls whose subs are not visible at compile time, if |
| 2496 |
they occurred in the last statement of an lvalue subroutine, |
| 2497 |
would reject non-lvalue subroutines and die with "Can't modify non-lvalue |
| 2498 |
subroutine call" [perl #102486]. |
| 2499 |
.Sp |
| 2500 |
Non-lvalue sub calls whose subs \fIare\fR visible at compile time exhibited |
| 2501 |
the opposite bug. If the call occurred in the last statement of an lvalue |
| 2502 |
subroutine, there would be no error when the lvalue sub was called in |
| 2503 |
lvalue context. Perl would blindly assign to the temporary value returned |
| 2504 |
by the non-lvalue subroutine. |
| 2505 |
.IP \(bu 4 |
| 2506 |
\&\f(CW\*(C`AUTOLOAD\*(C'\fR routines used to take precedence over the actual sub being |
| 2507 |
called (i.e., when autoloading wasn't needed), for sub calls in lvalue or |
| 2508 |
potential lvalue context, if the subroutine was not visible at compile |
| 2509 |
time. |
| 2510 |
.IP \(bu 4 |
| 2511 |
Applying the \f(CW\*(C`:lvalue\*(C'\fR attribute to an XSUB or to an aliased subroutine |
| 2512 |
stub with \f(CW\*(C`sub foo :lvalue;\*(C'\fR syntax stopped working in Perl 5.12. |
| 2513 |
This has been fixed. |
| 2514 |
.IP \(bu 4 |
| 2515 |
Applying the :lvalue attribute to subroutine that is already defined does |
| 2516 |
not work properly, as the attribute changes the way the sub is compiled. |
| 2517 |
Hence, Perl 5.12 began warning when an attempt is made to apply the |
| 2518 |
attribute to an already defined sub. In such cases, the attribute is |
| 2519 |
discarded. |
| 2520 |
.Sp |
| 2521 |
But the change in 5.12 missed the case where custom attributes are also |
| 2522 |
present: that case still silently and ineffectively applied the attribute. |
| 2523 |
That omission has now been corrected. \f(CW\*(C`sub foo :lvalue :Whatever\*(C'\fR (when |
| 2524 |
\&\f(CW\*(C`foo\*(C'\fR is already defined) now warns about the :lvalue attribute, and does |
| 2525 |
not apply it. |
| 2526 |
.IP \(bu 4 |
| 2527 |
A bug affecting lvalue context propagation through nested lvalue subroutine |
| 2528 |
calls has been fixed. Previously, returning a value in nested rvalue |
| 2529 |
context would be treated as lvalue context by the inner subroutine call, |
| 2530 |
resulting in some values (such as read-only values) being rejected. |
| 2531 |
.SS Overloading |
| 2532 |
.IX Subsection "Overloading" |
| 2533 |
.IP \(bu 4 |
| 2534 |
Arithmetic assignment (\f(CW\*(C`$left += $right\*(C'\fR) involving overloaded objects |
| 2535 |
that rely on the 'nomethod' override no longer segfault when the left |
| 2536 |
operand is not overloaded. |
| 2537 |
.IP \(bu 4 |
| 2538 |
Errors that occur when methods cannot be found during overloading now |
| 2539 |
mention the correct package name, as they did in 5.8.x, instead of |
| 2540 |
erroneously mentioning the "overload" package, as they have since 5.10.0. |
| 2541 |
.IP \(bu 4 |
| 2542 |
Undefining \f(CW%overload::\fR no longer causes a crash. |
| 2543 |
.SS "Prototypes of built-in keywords" |
| 2544 |
.IX Subsection "Prototypes of built-in keywords" |
| 2545 |
.IP \(bu 4 |
| 2546 |
The \f(CW\*(C`prototype\*(C'\fR function no longer dies for the \f(CW\*(C`_\|_FILE_\|_\*(C'\fR, \f(CW\*(C`_\|_LINE_\|_\*(C'\fR |
| 2547 |
and \f(CW\*(C`_\|_PACKAGE_\|_\*(C'\fR directives. It now returns an empty-string prototype |
| 2548 |
for them, because they are syntactically indistinguishable from nullary |
| 2549 |
functions like \f(CW\*(C`time\*(C'\fR. |
| 2550 |
.IP \(bu 4 |
| 2551 |
\&\f(CW\*(C`prototype\*(C'\fR now returns \f(CW\*(C`undef\*(C'\fR for all overridable infix operators, |
| 2552 |
such as \f(CW\*(C`eq\*(C'\fR, which are not callable in any way resembling functions. |
| 2553 |
It used to return incorrect prototypes for some and die for others |
| 2554 |
[perl #94984]. |
| 2555 |
.IP \(bu 4 |
| 2556 |
The prototypes of several built-in functions\-\-\f(CW\*(C`getprotobynumber\*(C'\fR, \f(CW\*(C`lock\*(C'\fR, |
| 2557 |
\&\f(CW\*(C`not\*(C'\fR and \f(CW\*(C`select\*(C'\fR\-\-have been corrected, or at least are now closer to |
| 2558 |
reality than before. |
| 2559 |
.SS "Regular expressions" |
| 2560 |
.IX Subsection "Regular expressions" |
| 2561 |
.IP \(bu 4 |
| 2562 |
\&\f(CW\*(C`/[[:ascii:]]/\*(C'\fR and \f(CW\*(C`/[[:blank:]]/\*(C'\fR now use locale rules under |
| 2563 |
\&\f(CW\*(C`use locale\*(C'\fR when the platform supports that. Previously, they used |
| 2564 |
the platform's native character set. |
| 2565 |
.IP \(bu 4 |
| 2566 |
\&\f(CW\*(C`m/[[:ascii:]]/i\*(C'\fR and \f(CW\*(C`/\ep{ASCII}/i\*(C'\fR now match identically (when not |
| 2567 |
under a differing locale). This fixes a regression introduced in 5.14 |
| 2568 |
in which the first expression could match characters outside of ASCII, |
| 2569 |
such as the KELVIN SIGN. |
| 2570 |
.IP \(bu 4 |
| 2571 |
\&\f(CW\*(C`/.*/g\*(C'\fR would sometimes refuse to match at the end of a string that ends |
| 2572 |
with "\en". This has been fixed [perl #109206]. |
| 2573 |
.IP \(bu 4 |
| 2574 |
Starting with 5.12.0, Perl used to get its internal bookkeeping muddled up |
| 2575 |
after assigning \f(CW\*(C`${ qr// }\*(C'\fR to a hash element and locking it with |
| 2576 |
Hash::Util. This could result in double frees, crashes, or erratic |
| 2577 |
behavior. |
| 2578 |
.IP \(bu 4 |
| 2579 |
The new (in 5.14.0) regular expression modifier \f(CW\*(C`/a\*(C'\fR when repeated like |
| 2580 |
\&\f(CW\*(C`/aa\*(C'\fR forbids the characters outside the ASCII range that match |
| 2581 |
characters inside that range from matching under \f(CW\*(C`/i\*(C'\fR. This did not |
| 2582 |
work under some circumstances, all involving alternation, such as: |
| 2583 |
.Sp |
| 2584 |
.Vb 1 |
| 2585 |
\& "\eN{KELVIN SIGN}" =~ /k|foo/iaa; |
| 2586 |
.Ve |
| 2587 |
.Sp |
| 2588 |
succeeded inappropriately. This is now fixed. |
| 2589 |
.IP \(bu 4 |
| 2590 |
5.14.0 introduced some memory leaks in regular expression character |
| 2591 |
classes such as \f(CW\*(C`[\ew\es]\*(C'\fR, which have now been fixed. (5.14.1) |
| 2592 |
.IP \(bu 4 |
| 2593 |
An edge case in regular expression matching could potentially loop. |
| 2594 |
This happened only under \f(CW\*(C`/i\*(C'\fR in bracketed character classes that have |
| 2595 |
characters with multi-character folds, and the target string to match |
| 2596 |
against includes the first portion of the fold, followed by another |
| 2597 |
character that has a multi-character fold that begins with the remaining |
| 2598 |
portion of the fold, plus some more. |
| 2599 |
.Sp |
| 2600 |
.Vb 1 |
| 2601 |
\& "s\eN{U+DF}" =~ /[\ex{DF}foo]/i |
| 2602 |
.Ve |
| 2603 |
.Sp |
| 2604 |
is one such case. \f(CW\*(C`\exDF\*(C'\fR folds to \f(CW"ss"\fR. (5.14.1) |
| 2605 |
.IP \(bu 4 |
| 2606 |
A few characters in regular expression pattern matches did not |
| 2607 |
match correctly in some circumstances, all involving \f(CW\*(C`/i\*(C'\fR. The |
| 2608 |
affected characters are: |
| 2609 |
COMBINING GREEK YPOGEGRAMMENI, |
| 2610 |
GREEK CAPITAL LETTER IOTA, |
| 2611 |
GREEK CAPITAL LETTER UPSILON, |
| 2612 |
GREEK PROSGEGRAMMENI, |
| 2613 |
GREEK SMALL LETTER IOTA WITH DIALYTIKA AND OXIA, |
| 2614 |
GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS, |
| 2615 |
GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND OXIA, |
| 2616 |
GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS, |
| 2617 |
LATIN SMALL LETTER LONG S, |
| 2618 |
LATIN SMALL LIGATURE LONG S T, |
| 2619 |
and |
| 2620 |
LATIN SMALL LIGATURE ST. |
| 2621 |
.IP \(bu 4 |
| 2622 |
A memory leak regression in regular expression compilation |
| 2623 |
under threading has been fixed. |
| 2624 |
.IP \(bu 4 |
| 2625 |
A regression introduced in 5.14.0 has |
| 2626 |
been fixed. This involved an inverted |
| 2627 |
bracketed character class in a regular expression that consisted solely |
| 2628 |
of a Unicode property. That property wasn't getting inverted outside the |
| 2629 |
Latin1 range. |
| 2630 |
.IP \(bu 4 |
| 2631 |
Three problematic Unicode characters now work better in regex pattern matching under \f(CW\*(C`/i\*(C'\fR. |
| 2632 |
.Sp |
| 2633 |
In the past, three Unicode characters: |
| 2634 |
LATIN SMALL LETTER SHARP S, |
| 2635 |
GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS, |
| 2636 |
and |
| 2637 |
GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS, |
| 2638 |
along with the sequences that they fold to |
| 2639 |
(including "ss" for LATIN SMALL LETTER SHARP S), |
| 2640 |
did not properly match under \f(CW\*(C`/i\*(C'\fR. 5.14.0 fixed some of these cases, |
| 2641 |
but introduced others, including a panic when one of the characters or |
| 2642 |
sequences was used in the \f(CW\*(C`(?(DEFINE)\*(C'\fR regular expression predicate. |
| 2643 |
The known bugs that were introduced in 5.14 have now been fixed; as well |
| 2644 |
as some other edge cases that have never worked until now. These all |
| 2645 |
involve using the characters and sequences outside bracketed character |
| 2646 |
classes under \f(CW\*(C`/i\*(C'\fR. This closes [perl #98546]. |
| 2647 |
.Sp |
| 2648 |
There remain known problems when using certain characters with |
| 2649 |
multi-character folds inside bracketed character classes, including such |
| 2650 |
constructs as \f(CW\*(C`qr/[\eN{LATIN SMALL LETTER SHARP}a\-z]/i\*(C'\fR. These |
| 2651 |
remaining bugs are addressed in [perl #89774]. |
| 2652 |
.IP \(bu 4 |
| 2653 |
RT #78266: The regex engine has been leaking memory when accessing |
| 2654 |
named captures that weren't matched as part of a regex ever since 5.10 |
| 2655 |
when they were introduced; e.g., this would consume over a hundred MB of |
| 2656 |
memory: |
| 2657 |
.Sp |
| 2658 |
.Vb 6 |
| 2659 |
\& for (1..10_000_000) { |
| 2660 |
\& if ("foo" =~ /(foo|(?<capture>bar))?/) { |
| 2661 |
\& my $capture = $+{capture} |
| 2662 |
\& } |
| 2663 |
\& } |
| 2664 |
\& system "ps \-o rss $$"\*(Aq |
| 2665 |
.Ve |
| 2666 |
.IP \(bu 4 |
| 2667 |
In 5.14, \f(CW\*(C`/[[:lower:]]/i\*(C'\fR and \f(CW\*(C`/[[:upper:]]/i\*(C'\fR no longer matched the |
| 2668 |
opposite case. This has been fixed [perl #101970]. |
| 2669 |
.IP \(bu 4 |
| 2670 |
A regular expression match with an overloaded object on the right-hand side |
| 2671 |
would sometimes stringify the object too many times. |
| 2672 |
.IP \(bu 4 |
| 2673 |
A regression has been fixed that was introduced in 5.14, in \f(CW\*(C`/i\*(C'\fR |
| 2674 |
regular expression matching, in which a match improperly fails if the |
| 2675 |
pattern is in UTF\-8, the target string is not, and a Latin\-1 character |
| 2676 |
precedes a character in the string that should match the pattern. |
| 2677 |
[perl #101710] |
| 2678 |
.IP \(bu 4 |
| 2679 |
In case-insensitive regular expression pattern matching, no longer on |
| 2680 |
UTF\-8 encoded strings does the scan for the start of match look only at |
| 2681 |
the first possible position. This caused matches such as |
| 2682 |
\&\f(CW\*(C`"f\ex{FB00}" =~ /ff/i\*(C'\fR to fail. |
| 2683 |
.IP \(bu 4 |
| 2684 |
The regexp optimizer no longer crashes on debugging builds when merging |
| 2685 |
fixed-string nodes with inconvenient contents. |
| 2686 |
.IP \(bu 4 |
| 2687 |
A panic involving the combination of the regular expression modifiers |
| 2688 |
\&\f(CW\*(C`/aa\*(C'\fR and the \f(CW\*(C`\eb\*(C'\fR escape sequence introduced in 5.14.0 has been |
| 2689 |
fixed [perl #95964]. (5.14.2) |
| 2690 |
.IP \(bu 4 |
| 2691 |
The combination of the regular expression modifiers \f(CW\*(C`/aa\*(C'\fR and the \f(CW\*(C`\eb\*(C'\fR |
| 2692 |
and \f(CW\*(C`\eB\*(C'\fR escape sequences did not work properly on UTF\-8 encoded |
| 2693 |
strings. All non-ASCII characters under \f(CW\*(C`/aa\*(C'\fR should be treated as |
| 2694 |
non-word characters, but what was happening was that Unicode rules were |
| 2695 |
used to determine wordness/non\-wordness for non-ASCII characters. This |
| 2696 |
is now fixed [perl #95968]. |
| 2697 |
.IP \(bu 4 |
| 2698 |
\&\f(CW\*(C`(?foo: ...)\*(C'\fR no longer loses passed in character set. |
| 2699 |
.IP \(bu 4 |
| 2700 |
The trie optimization used to have problems with alternations containing |
| 2701 |
an empty \f(CW\*(C`(?:)\*(C'\fR, causing \f(CW\*(C`"x" =~ /\eA(?>(?:(?:)A|B|C?x))\ez/\*(C'\fR not to |
| 2702 |
match, whereas it should [perl #111842]. |
| 2703 |
.IP \(bu 4 |
| 2704 |
Use of lexical (\f(CW\*(C`my\*(C'\fR) variables in code blocks embedded in regular |
| 2705 |
expressions will no longer result in memory corruption or crashes. |
| 2706 |
.Sp |
| 2707 |
Nevertheless, these code blocks are still experimental, as there are still |
| 2708 |
problems with the wrong variables being closed over (in loops for instance) |
| 2709 |
and with abnormal exiting (e.g., \f(CW\*(C`die\*(C'\fR) causing memory corruption. |
| 2710 |
.IP \(bu 4 |
| 2711 |
The \f(CW\*(C`\eh\*(C'\fR, \f(CW\*(C`\eH\*(C'\fR, \f(CW\*(C`\ev\*(C'\fR and \f(CW\*(C`\eV\*(C'\fR regular expression metacharacters used to |
| 2712 |
cause a panic error message when trying to match at the end of the |
| 2713 |
string [perl #96354]. |
| 2714 |
.IP \(bu 4 |
| 2715 |
The abbreviations for four C1 control characters \f(CW\*(C`MW\*(C'\fR \f(CW\*(C`PM\*(C'\fR, \f(CW\*(C`RI\*(C'\fR, and |
| 2716 |
\&\f(CW\*(C`ST\*(C'\fR were previously unrecognized by \f(CW\*(C`\eN{}\*(C'\fR, \fBvianame()\fR, and |
| 2717 |
\&\fBstring_vianame()\fR. |
| 2718 |
.IP \(bu 4 |
| 2719 |
Mentioning a variable named "&" other than \f(CW$&\fR (i.e., \f(CW\*(C`@&\*(C'\fR or \f(CW\*(C`%&\*(C'\fR) no |
| 2720 |
longer stops \f(CW$&\fR from working. The same applies to variables named "'" |
| 2721 |
and "`" [perl #24237]. |
| 2722 |
.IP \(bu 4 |
| 2723 |
Creating a \f(CW\*(C`UNIVERSAL::AUTOLOAD\*(C'\fR sub no longer stops \f(CW\*(C`%+\*(C'\fR, \f(CW\*(C`%\-\*(C'\fR and |
| 2724 |
\&\f(CW\*(C`%!\*(C'\fR from working some of the time [perl #105024]. |
| 2725 |
.SS Smartmatching |
| 2726 |
.IX Subsection "Smartmatching" |
| 2727 |
.IP \(bu 4 |
| 2728 |
\&\f(CW\*(C`~~\*(C'\fR now correctly handles the precedence of Any~~Object, and is not tricked |
| 2729 |
by an overloaded object on the left-hand side. |
| 2730 |
.IP \(bu 4 |
| 2731 |
In Perl 5.14.0, \f(CW\*(C`$tainted ~~ @array\*(C'\fR stopped working properly. Sometimes |
| 2732 |
it would erroneously fail (when \f(CW$tainted\fR contained a string that occurs |
| 2733 |
in the array \fIafter\fR the first element) or erroneously succeed (when |
| 2734 |
\&\f(CW\*(C`undef\*(C'\fR occurred after the first element) [perl #93590]. |
| 2735 |
.ie n .SS "The ""sort"" operator" |
| 2736 |
.el .SS "The \f(CWsort\fP operator" |
| 2737 |
.IX Subsection "The sort operator" |
| 2738 |
.IP \(bu 4 |
| 2739 |
\&\f(CW\*(C`sort\*(C'\fR was not treating \f(CW\*(C`sub {}\*(C'\fR and \f(CW\*(C`sub {()}\*(C'\fR as equivalent when |
| 2740 |
such a sub was provided as the comparison routine. It used to croak on |
| 2741 |
\&\f(CW\*(C`sub {()}\*(C'\fR. |
| 2742 |
.IP \(bu 4 |
| 2743 |
\&\f(CW\*(C`sort\*(C'\fR now works once more with custom sort routines that are XSUBs. It |
| 2744 |
stopped working in 5.10.0. |
| 2745 |
.IP \(bu 4 |
| 2746 |
\&\f(CW\*(C`sort\*(C'\fR with a constant for a custom sort routine, although it produces |
| 2747 |
unsorted results, no longer crashes. It started crashing in 5.10.0. |
| 2748 |
.IP \(bu 4 |
| 2749 |
Warnings emitted by \f(CW\*(C`sort\*(C'\fR when a custom comparison routine returns a |
| 2750 |
non-numeric value now contain "in sort" and show the line number of the |
| 2751 |
\&\f(CW\*(C`sort\*(C'\fR operator, rather than the last line of the comparison routine. The |
| 2752 |
warnings also now occur only if warnings are enabled in the scope where |
| 2753 |
\&\f(CW\*(C`sort\*(C'\fR occurs. Previously the warnings would occur if enabled in the |
| 2754 |
comparison routine's scope. |
| 2755 |
.IP \(bu 4 |
| 2756 |
\&\f(CW\*(C`sort { $a <=> $b }\*(C'\fR, which is optimized internally, now produces |
| 2757 |
"uninitialized" warnings for NaNs (not-a-number values), since \f(CW\*(C`<=>\*(C'\fR |
| 2758 |
returns \f(CW\*(C`undef\*(C'\fR for those. This brings it in line with |
| 2759 |
\&\f(CW\*(C`sort\ {\ 1;\ $a\ <=>\ $b\ }\*(C'\fR and other more complex cases, which are not |
| 2760 |
optimized [perl #94390]. |
| 2761 |
.ie n .SS "The ""substr"" operator" |
| 2762 |
.el .SS "The \f(CWsubstr\fP operator" |
| 2763 |
.IX Subsection "The substr operator" |
| 2764 |
.IP \(bu 4 |
| 2765 |
Tied (and otherwise magical) variables are no longer exempt from the |
| 2766 |
"Attempt to use reference as lvalue in substr" warning. |
| 2767 |
.IP \(bu 4 |
| 2768 |
That warning now occurs when the returned lvalue is assigned to, not |
| 2769 |
when \f(CW\*(C`substr\*(C'\fR itself is called. This makes a difference only if the |
| 2770 |
return value of \f(CW\*(C`substr\*(C'\fR is referenced and later assigned to. |
| 2771 |
.IP \(bu 4 |
| 2772 |
Passing a substring of a read-only value or a typeglob to a function |
| 2773 |
(potential lvalue context) no longer causes an immediate "Can't coerce" |
| 2774 |
or "Modification of a read-only value" error. That error occurs only |
| 2775 |
if the passed value is assigned to. |
| 2776 |
.Sp |
| 2777 |
The same thing happens with the "substr outside of string" error. If |
| 2778 |
the lvalue is only read from, not written to, it is now just a warning, as |
| 2779 |
with rvalue \f(CW\*(C`substr\*(C'\fR. |
| 2780 |
.IP \(bu 4 |
| 2781 |
\&\f(CW\*(C`substr\*(C'\fR assignments no longer call FETCH twice if the first argument |
| 2782 |
is a tied variable, just once. |
| 2783 |
.SS "Support for embedded nulls" |
| 2784 |
.IX Subsection "Support for embedded nulls" |
| 2785 |
Some parts of Perl did not work correctly with nulls (\f(CW\*(C`chr 0\*(C'\fR) embedded in |
| 2786 |
strings. That meant that, for instance, \f(CW\*(C`$m = "a\e0b"; foo\->$m\*(C'\fR would |
| 2787 |
call the "a" method, instead of the actual method name contained in \f(CW$m\fR. |
| 2788 |
These parts of perl have been fixed to support nulls: |
| 2789 |
.IP \(bu 4 |
| 2790 |
Method names |
| 2791 |
.IP \(bu 4 |
| 2792 |
Typeglob names (including filehandle and subroutine names) |
| 2793 |
.IP \(bu 4 |
| 2794 |
Package names, including the return value of \f(CWref()\fR |
| 2795 |
.IP \(bu 4 |
| 2796 |
Typeglob elements (\f(CW*foo{"THING\e0stuff"}\fR) |
| 2797 |
.IP \(bu 4 |
| 2798 |
Signal names |
| 2799 |
.IP \(bu 4 |
| 2800 |
Various warnings and error messages that mention variable names or values, |
| 2801 |
methods, etc. |
| 2802 |
.PP |
| 2803 |
One side effect of these changes is that blessing into "\e0" no longer |
| 2804 |
causes \f(CWref()\fR to return false. |
| 2805 |
.SS "Threading bugs" |
| 2806 |
.IX Subsection "Threading bugs" |
| 2807 |
.IP \(bu 4 |
| 2808 |
Typeglobs returned from threads are no longer cloned if the parent thread |
| 2809 |
already has a glob with the same name. This means that returned |
| 2810 |
subroutines will now assign to the right package variables [perl #107366]. |
| 2811 |
.IP \(bu 4 |
| 2812 |
Some cases of threads crashing due to memory allocation during cloning have |
| 2813 |
been fixed [perl #90006]. |
| 2814 |
.IP \(bu 4 |
| 2815 |
Thread joining would sometimes emit "Attempt to free unreferenced scalar" |
| 2816 |
warnings if \f(CW\*(C`caller\*(C'\fR had been used from the \f(CW\*(C`DB\*(C'\fR package before thread |
| 2817 |
creation [perl #98092]. |
| 2818 |
.IP \(bu 4 |
| 2819 |
Locking a subroutine (via \f(CW\*(C`lock &sub\*(C'\fR) is no longer a compile-time error |
| 2820 |
for regular subs. For lvalue subroutines, it no longer tries to return the |
| 2821 |
sub as a scalar, resulting in strange side effects like \f(CW\*(C`ref \e$_\*(C'\fR |
| 2822 |
returning "CODE" in some instances. |
| 2823 |
.Sp |
| 2824 |
\&\f(CW\*(C`lock &sub\*(C'\fR is now a run-time error if threads::shared is loaded (a |
| 2825 |
no-op otherwise), but that may be rectified in a future version. |
| 2826 |
.SS "Tied variables" |
| 2827 |
.IX Subsection "Tied variables" |
| 2828 |
.IP \(bu 4 |
| 2829 |
Various cases in which FETCH was being ignored or called too many times |
| 2830 |
have been fixed: |
| 2831 |
.RS 4 |
| 2832 |
.IP \(bu 4 |
| 2833 |
\&\f(CW\*(C`PerlIO::get_layers\*(C'\fR [perl #97956] |
| 2834 |
.IP \(bu 4 |
| 2835 |
\&\f(CW\*(C`$tied =~ y/a/b/\*(C'\fR, \f(CW\*(C`chop $tied\*(C'\fR and \f(CW\*(C`chomp $tied\*(C'\fR when \f(CW$tied\fR holds a |
| 2836 |
reference. |
| 2837 |
.IP \(bu 4 |
| 2838 |
When calling \f(CW\*(C`local $_\*(C'\fR [perl #105912] |
| 2839 |
.IP \(bu 4 |
| 2840 |
Four-argument \f(CW\*(C`select\*(C'\fR |
| 2841 |
.IP \(bu 4 |
| 2842 |
A tied buffer passed to \f(CW\*(C`sysread\*(C'\fR |
| 2843 |
.IP \(bu 4 |
| 2844 |
\&\f(CW\*(C`$tied .= <>\*(C'\fR |
| 2845 |
.IP \(bu 4 |
| 2846 |
Three-argument \f(CW\*(C`open\*(C'\fR, the third being a tied file handle |
| 2847 |
(as in \f(CW\*(C`open $fh, ">&", $tied\*(C'\fR) |
| 2848 |
.IP \(bu 4 |
| 2849 |
\&\f(CW\*(C`sort\*(C'\fR with a reference to a tied glob for the comparison routine. |
| 2850 |
.IP \(bu 4 |
| 2851 |
\&\f(CW\*(C`..\*(C'\fR and \f(CW\*(C`...\*(C'\fR in list context [perl #53554]. |
| 2852 |
.IP \(bu 4 |
| 2853 |
\&\f(CW\*(C`${$tied}\*(C'\fR, \f(CW\*(C`@{$tied}\*(C'\fR, \f(CW\*(C`%{$tied}\*(C'\fR and \f(CW\*(C`*{$tied}\*(C'\fR where the tied |
| 2854 |
variable returns a string (\f(CW\*(C`&{}\*(C'\fR was unaffected) |
| 2855 |
.IP \(bu 4 |
| 2856 |
\&\f(CW\*(C`defined ${ $tied_variable }\*(C'\fR |
| 2857 |
.IP \(bu 4 |
| 2858 |
Various functions that take a filehandle argument in rvalue context |
| 2859 |
(\f(CW\*(C`close\*(C'\fR, \f(CW\*(C`readline\*(C'\fR, etc.) [perl #97482] |
| 2860 |
.IP \(bu 4 |
| 2861 |
Some cases of dereferencing a complex expression, such as |
| 2862 |
\&\f(CW\*(C`${ (), $tied } = 1\*(C'\fR, used to call \f(CW\*(C`FETCH\*(C'\fR multiple times, but now call |
| 2863 |
it once. |
| 2864 |
.IP \(bu 4 |
| 2865 |
\&\f(CW\*(C`$tied\->method\*(C'\fR where \f(CW$tied\fR returns a package name\-\-even resulting in |
| 2866 |
a failure to call the method, due to memory corruption |
| 2867 |
.IP \(bu 4 |
| 2868 |
Assignments like \f(CW\*(C`*$tied = \e&{"..."}\*(C'\fR and \f(CW\*(C`*glob = $tied\*(C'\fR |
| 2869 |
.IP \(bu 4 |
| 2870 |
\&\f(CW\*(C`chdir\*(C'\fR, \f(CW\*(C`chmod\*(C'\fR, \f(CW\*(C`chown\*(C'\fR, \f(CW\*(C`utime\*(C'\fR, \f(CW\*(C`truncate\*(C'\fR, \f(CW\*(C`stat\*(C'\fR, \f(CW\*(C`lstat\*(C'\fR and |
| 2871 |
the filetest ops (\f(CW\*(C`\-r\*(C'\fR, \f(CW\*(C`\-x\*(C'\fR, etc.) |
| 2872 |
.RE |
| 2873 |
.RS 4 |
| 2874 |
.RE |
| 2875 |
.IP \(bu 4 |
| 2876 |
\&\f(CW\*(C`caller\*(C'\fR sets \f(CW@DB::args\fR to the subroutine arguments when called from |
| 2877 |
the DB package. It used to crash when doing so if \f(CW@DB::args\fR happened to |
| 2878 |
be tied. Now it croaks instead. |
| 2879 |
.IP \(bu 4 |
| 2880 |
Tying an element of \f(CW%ENV\fR or \f(CW\*(C`%^H\*(C'\fR and then deleting that element would |
| 2881 |
result in a call to the tie object's DELETE method, even though tying the |
| 2882 |
element itself is supposed to be equivalent to tying a scalar (the element |
| 2883 |
is, of course, a scalar) [perl #67490]. |
| 2884 |
.IP \(bu 4 |
| 2885 |
When Perl autovivifies an element of a tied array or hash (which entails |
| 2886 |
calling STORE with a new reference), it now calls FETCH immediately after |
| 2887 |
the STORE, instead of assuming that FETCH would have returned the same |
| 2888 |
reference. This can make it easier to implement tied objects [perl #35865, #43011]. |
| 2889 |
.IP \(bu 4 |
| 2890 |
Four-argument \f(CW\*(C`select\*(C'\fR no longer produces its "Non-string passed as |
| 2891 |
bitmask" warning on tied or tainted variables that are strings. |
| 2892 |
.IP \(bu 4 |
| 2893 |
Localizing a tied scalar that returns a typeglob no longer stops it from |
| 2894 |
being tied till the end of the scope. |
| 2895 |
.IP \(bu 4 |
| 2896 |
Attempting to \f(CW\*(C`goto\*(C'\fR out of a tied handle method used to cause memory |
| 2897 |
corruption or crashes. Now it produces an error message instead |
| 2898 |
[perl #8611]. |
| 2899 |
.IP \(bu 4 |
| 2900 |
A bug has been fixed that occurs when a tied variable is used as a |
| 2901 |
subroutine reference: if the last thing assigned to or returned from the |
| 2902 |
variable was a reference or typeglob, the \f(CW\*(C`\e&$tied\*(C'\fR could either crash or |
| 2903 |
return the wrong subroutine. The reference case is a regression introduced |
| 2904 |
in Perl 5.10.0. For typeglobs, it has probably never worked till now. |
| 2905 |
.SS "Version objects and vstrings" |
| 2906 |
.IX Subsection "Version objects and vstrings" |
| 2907 |
.IP \(bu 4 |
| 2908 |
The bitwise complement operator (and possibly other operators, too) when |
| 2909 |
passed a vstring would leave vstring magic attached to the return value, |
| 2910 |
even though the string had changed. This meant that |
| 2911 |
\&\f(CW\*(C`version\->new(~v1.2.3)\*(C'\fR would create a version looking like "v1.2.3" |
| 2912 |
even though the string passed to \f(CW\*(C`version\->new\*(C'\fR was actually |
| 2913 |
"\e376\e375\e374". This also caused B::Deparse to deparse \f(CW\*(C`~v1.2.3\*(C'\fR |
| 2914 |
incorrectly, without the \f(CW\*(C`~\*(C'\fR [perl #29070]. |
| 2915 |
.IP \(bu 4 |
| 2916 |
Assigning a vstring to a magic (e.g., tied, \f(CW$!\fR) variable and then |
| 2917 |
assigning something else used to blow away all magic. This meant that |
| 2918 |
tied variables would come undone, \f(CW$!\fR would stop getting updated on |
| 2919 |
failed system calls, \f(CW$|\fR would stop setting autoflush, and other |
| 2920 |
mischief would take place. This has been fixed. |
| 2921 |
.IP \(bu 4 |
| 2922 |
\&\f(CW\*(C`version\->new("version")\*(C'\fR and \f(CW\*(C`printf "%vd", "version"\*(C'\fR no longer |
| 2923 |
crash [perl #102586]. |
| 2924 |
.IP \(bu 4 |
| 2925 |
Version comparisons, such as those that happen implicitly with \f(CW\*(C`use |
| 2926 |
v5.43\*(C'\fR, no longer cause locale settings to change [perl #105784]. |
| 2927 |
.IP \(bu 4 |
| 2928 |
Version objects no longer cause memory leaks in boolean context |
| 2929 |
[perl #109762]. |
| 2930 |
.SS "Warnings, redefinition" |
| 2931 |
.IX Subsection "Warnings, redefinition" |
| 2932 |
.IP \(bu 4 |
| 2933 |
Subroutines from the \f(CW\*(C`autouse\*(C'\fR namespace are once more exempt from |
| 2934 |
redefinition warnings. This used to work in 5.005, but was broken in |
| 2935 |
5.6 for most subroutines. For subs created via XS that redefine |
| 2936 |
subroutines from the \f(CW\*(C`autouse\*(C'\fR package, this stopped working in 5.10. |
| 2937 |
.IP \(bu 4 |
| 2938 |
New XSUBs now produce redefinition warnings if they overwrite existing |
| 2939 |
subs, as they did in 5.8.x. (The \f(CW\*(C`autouse\*(C'\fR logic was reversed in |
| 2940 |
5.10\-14. Only subroutines from the \f(CW\*(C`autouse\*(C'\fR namespace would warn |
| 2941 |
when clobbered.) |
| 2942 |
.IP \(bu 4 |
| 2943 |
\&\f(CW\*(C`newCONSTSUB\*(C'\fR used to use compile-time warning hints, instead of |
| 2944 |
run-time hints. The following code should never produce a redefinition |
| 2945 |
warning, but it used to, if \f(CW\*(C`newCONSTSUB\*(C'\fR redefined an existing |
| 2946 |
subroutine: |
| 2947 |
.Sp |
| 2948 |
.Vb 5 |
| 2949 |
\& use warnings; |
| 2950 |
\& BEGIN { |
| 2951 |
\& no warnings; |
| 2952 |
\& some_XS_function_that_calls_new_CONSTSUB(); |
| 2953 |
\& } |
| 2954 |
.Ve |
| 2955 |
.IP \(bu 4 |
| 2956 |
Redefinition warnings for constant subroutines are on by default (what |
| 2957 |
are known as severe warnings in perldiag). This occurred only |
| 2958 |
when it was a glob assignment or declaration of a Perl subroutine that |
| 2959 |
caused the warning. If the creation of XSUBs triggered the warning, it |
| 2960 |
was not a default warning. This has been corrected. |
| 2961 |
.IP \(bu 4 |
| 2962 |
The internal check to see whether a redefinition warning should occur |
| 2963 |
used to emit "uninitialized" warnings in cases like this: |
| 2964 |
.Sp |
| 2965 |
.Vb 4 |
| 2966 |
\& use warnings "uninitialized"; |
| 2967 |
\& use constant {u => undef, v => undef}; |
| 2968 |
\& sub foo(){u} |
| 2969 |
\& sub foo(){v} |
| 2970 |
.Ve |
| 2971 |
.SS "Warnings, ""Uninitialized""" |
| 2972 |
.IX Subsection "Warnings, ""Uninitialized""" |
| 2973 |
.IP \(bu 4 |
| 2974 |
Various functions that take a filehandle argument in rvalue context |
| 2975 |
(\f(CW\*(C`close\*(C'\fR, \f(CW\*(C`readline\*(C'\fR, etc.) used to warn twice for an undefined handle |
| 2976 |
[perl #97482]. |
| 2977 |
.IP \(bu 4 |
| 2978 |
\&\f(CW\*(C`dbmopen\*(C'\fR now only warns once, rather than three times, if the mode |
| 2979 |
argument is \f(CW\*(C`undef\*(C'\fR [perl #90064]. |
| 2980 |
.IP \(bu 4 |
| 2981 |
The \f(CW\*(C`+=\*(C'\fR operator does not usually warn when the left-hand side is |
| 2982 |
\&\f(CW\*(C`undef\*(C'\fR, but it was doing so for tied variables. This has been fixed |
| 2983 |
[perl #44895]. |
| 2984 |
.IP \(bu 4 |
| 2985 |
A bug fix in Perl 5.14 introduced a new bug, causing "uninitialized" |
| 2986 |
warnings to report the wrong variable if the operator in question had |
| 2987 |
two operands and one was \f(CW\*(C`%{...}\*(C'\fR or \f(CW\*(C`@{...}\*(C'\fR. This has been fixed |
| 2988 |
[perl #103766]. |
| 2989 |
.IP \(bu 4 |
| 2990 |
\&\f(CW\*(C`..\*(C'\fR and \f(CW\*(C`...\*(C'\fR in list context now mention the name of the variable in |
| 2991 |
"uninitialized" warnings for string (as opposed to numeric) ranges. |
| 2992 |
.SS "Weak references" |
| 2993 |
.IX Subsection "Weak references" |
| 2994 |
.IP \(bu 4 |
| 2995 |
Weakening the first argument to an automatically-invoked \f(CW\*(C`DESTROY\*(C'\fR method |
| 2996 |
could result in erroneous "DESTROY created new reference" errors or |
| 2997 |
crashes. Now it is an error to weaken a read-only reference. |
| 2998 |
.IP \(bu 4 |
| 2999 |
Weak references to lexical hashes going out of scope were not going stale |
| 3000 |
(becoming undefined), but continued to point to the hash. |
| 3001 |
.IP \(bu 4 |
| 3002 |
Weak references to lexical variables going out of scope are now broken |
| 3003 |
before any magical methods (e.g., DESTROY on a tie object) are called. |
| 3004 |
This prevents such methods from modifying the variable that will be seen |
| 3005 |
the next time the scope is entered. |
| 3006 |
.IP \(bu 4 |
| 3007 |
Creating a weak reference to an \f(CW@ISA\fR array or accessing the array index |
| 3008 |
(\f(CW$#ISA\fR) could result in confused internal bookkeeping for elements |
| 3009 |
later added to the \f(CW@ISA\fR array. For instance, creating a weak |
| 3010 |
reference to the element itself could push that weak reference on to \f(CW@ISA\fR; |
| 3011 |
and elements added after use of \f(CW$#ISA\fR would be ignored by method lookup |
| 3012 |
[perl #85670]. |
| 3013 |
.SS "Other notable fixes" |
| 3014 |
.IX Subsection "Other notable fixes" |
| 3015 |
.IP \(bu 4 |
| 3016 |
\&\f(CW\*(C`quotemeta\*(C'\fR now quotes consistently the same non-ASCII characters under |
| 3017 |
\&\f(CW\*(C`use feature \*(Aqunicode_strings\*(Aq\*(C'\fR, regardless of whether the string is |
| 3018 |
encoded in UTF\-8 or not, hence fixing the last vestiges (we hope) of the |
| 3019 |
notorious "The "Unicode Bug"" in perlunicode. [perl #77654]. |
| 3020 |
.Sp |
| 3021 |
Which of these code points is quoted has changed, based on Unicode's |
| 3022 |
recommendations. See "quotemeta" in perlfunc for details. |
| 3023 |
.IP \(bu 4 |
| 3024 |
\&\f(CW\*(C`study\*(C'\fR is now a no-op, presumably fixing all outstanding bugs related to |
| 3025 |
study causing regex matches to behave incorrectly! |
| 3026 |
.IP \(bu 4 |
| 3027 |
When one writes \f(CW\*(C`open foo || die\*(C'\fR, which used to work in Perl 4, a |
| 3028 |
"Precedence problem" warning is produced. This warning used erroneously to |
| 3029 |
apply to fully-qualified bareword handle names not followed by \f(CW\*(C`||\*(C'\fR. This |
| 3030 |
has been corrected. |
| 3031 |
.IP \(bu 4 |
| 3032 |
After package aliasing (\f(CW\*(C`*foo:: = *bar::\*(C'\fR), \f(CW\*(C`select\*(C'\fR with 0 or 1 argument |
| 3033 |
would sometimes return a name that could not be used to refer to the |
| 3034 |
filehandle, or sometimes it would return \f(CW\*(C`undef\*(C'\fR even when a filehandle |
| 3035 |
was selected. Now it returns a typeglob reference in such cases. |
| 3036 |
.IP \(bu 4 |
| 3037 |
\&\f(CW\*(C`PerlIO::get_layers\*(C'\fR no longer ignores some arguments that it thinks are |
| 3038 |
numeric, while treating others as filehandle names. It is now consistent |
| 3039 |
for flat scalars (i.e., not references). |
| 3040 |
.IP \(bu 4 |
| 3041 |
Unrecognized switches on \f(CW\*(C`#!\*(C'\fR line |
| 3042 |
.Sp |
| 3043 |
If a switch, such as \fB\-x\fR, that cannot occur on the \f(CW\*(C`#!\*(C'\fR line is used |
| 3044 |
there, perl dies with "Can't emulate...". |
| 3045 |
.Sp |
| 3046 |
It used to produce the same message for switches that perl did not |
| 3047 |
recognize at all, whether on the command line or the \f(CW\*(C`#!\*(C'\fR line. |
| 3048 |
.Sp |
| 3049 |
Now it produces the "Unrecognized switch" error message [perl #104288]. |
| 3050 |
.IP \(bu 4 |
| 3051 |
\&\f(CW\*(C`system\*(C'\fR now temporarily blocks the SIGCHLD signal handler, to prevent the |
| 3052 |
signal handler from stealing the exit status [perl #105700]. |
| 3053 |
.IP \(bu 4 |
| 3054 |
The \f(CW%n\fR formatting code for \f(CW\*(C`printf\*(C'\fR and \f(CW\*(C`sprintf\*(C'\fR, which causes the number |
| 3055 |
of characters to be assigned to the next argument, now actually |
| 3056 |
assigns the number of characters, instead of the number of bytes. |
| 3057 |
.Sp |
| 3058 |
It also works now with special lvalue functions like \f(CW\*(C`substr\*(C'\fR and with |
| 3059 |
nonexistent hash and array elements [perl #3471, #103492]. |
| 3060 |
.IP \(bu 4 |
| 3061 |
Perl skips copying values returned from a subroutine, for the sake of |
| 3062 |
speed, if doing so would make no observable difference. Because of faulty |
| 3063 |
logic, this would happen with the |
| 3064 |
result of \f(CW\*(C`delete\*(C'\fR, \f(CW\*(C`shift\*(C'\fR or \f(CW\*(C`splice\*(C'\fR, even if the result was |
| 3065 |
referenced elsewhere. It also did so with tied variables about to be freed |
| 3066 |
[perl #91844, #95548]. |
| 3067 |
.IP \(bu 4 |
| 3068 |
\&\f(CW\*(C`utf8::decode\*(C'\fR now refuses to modify read-only scalars [perl #91850]. |
| 3069 |
.IP \(bu 4 |
| 3070 |
Freeing \f(CW$_\fR inside a \f(CW\*(C`grep\*(C'\fR or \f(CW\*(C`map\*(C'\fR block, a code block embedded in a |
| 3071 |
regular expression, or an \f(CW@INC\fR filter (a subroutine returned by a |
| 3072 |
subroutine in \f(CW@INC\fR) used to result in double frees or crashes |
| 3073 |
[perl #91880, #92254, #92256]. |
| 3074 |
.IP \(bu 4 |
| 3075 |
\&\f(CW\*(C`eval\*(C'\fR returns \f(CW\*(C`undef\*(C'\fR in scalar context or an empty list in list |
| 3076 |
context when there is a run-time error. When \f(CW\*(C`eval\*(C'\fR was passed a |
| 3077 |
string in list context and a syntax error occurred, it used to return a |
| 3078 |
list containing a single undefined element. Now it returns an empty |
| 3079 |
list in list context for all errors [perl #80630]. |
| 3080 |
.IP \(bu 4 |
| 3081 |
\&\f(CW\*(C`goto &func\*(C'\fR no longer crashes, but produces an error message, when |
| 3082 |
the unwinding of the current subroutine's scope fires a destructor that |
| 3083 |
undefines the subroutine being "goneto" [perl #99850]. |
| 3084 |
.IP \(bu 4 |
| 3085 |
Perl now holds an extra reference count on the package that code is |
| 3086 |
currently compiling in. This means that the following code no longer |
| 3087 |
crashes [perl #101486]: |
| 3088 |
.Sp |
| 3089 |
.Vb 3 |
| 3090 |
\& package Foo; |
| 3091 |
\& BEGIN {*Foo:: = *Bar::} |
| 3092 |
\& sub foo; |
| 3093 |
.Ve |
| 3094 |
.IP \(bu 4 |
| 3095 |
The \f(CW\*(C`x\*(C'\fR repetition operator no longer crashes on 64\-bit builds with large |
| 3096 |
repeat counts [perl #94560]. |
| 3097 |
.IP \(bu 4 |
| 3098 |
Calling \f(CW\*(C`require\*(C'\fR on an implicit \f(CW$_\fR when \f(CW*CORE::GLOBAL::require\fR has |
| 3099 |
been overridden does not segfault anymore, and \f(CW$_\fR is now passed to the |
| 3100 |
overriding subroutine [perl #78260]. |
| 3101 |
.IP \(bu 4 |
| 3102 |
\&\f(CW\*(C`use\*(C'\fR and \f(CW\*(C`require\*(C'\fR are no longer affected by the I/O layers active in |
| 3103 |
the caller's scope (enabled by open.pm) [perl #96008]. |
| 3104 |
.IP \(bu 4 |
| 3105 |
\&\f(CW\*(C`our $::é; $é\*(C'\fR (which is invalid) no longer produces the "Compilation |
| 3106 |
error at lib/utf8_heavy.pl..." error message, which it started emitting in |
| 3107 |
5.10.0 [perl #99984]. |
| 3108 |
.IP \(bu 4 |
| 3109 |
On 64\-bit systems, \f(CWread()\fR now understands large string offsets beyond |
| 3110 |
the 32\-bit range. |
| 3111 |
.IP \(bu 4 |
| 3112 |
Errors that occur when processing subroutine attributes no longer cause the |
| 3113 |
subroutine's op tree to leak. |
| 3114 |
.IP \(bu 4 |
| 3115 |
Passing the same constant subroutine to both \f(CW\*(C`index\*(C'\fR and \f(CW\*(C`formline\*(C'\fR no |
| 3116 |
longer causes one or the other to fail [perl #89218]. (5.14.1) |
| 3117 |
.IP \(bu 4 |
| 3118 |
List assignment to lexical variables declared with attributes in the same |
| 3119 |
statement (\f(CW\*(C`my ($x,@y) : blimp = (72,94)\*(C'\fR) stopped working in Perl 5.8.0. |
| 3120 |
It has now been fixed. |
| 3121 |
.IP \(bu 4 |
| 3122 |
Perl 5.10.0 introduced some faulty logic that made "U*" in the middle of |
| 3123 |
a pack template equivalent to "U0" if the input string was empty. This has |
| 3124 |
been fixed [perl #90160]. (5.14.2) |
| 3125 |
.IP \(bu 4 |
| 3126 |
Destructors on objects were not called during global destruction on objects |
| 3127 |
that were not referenced by any scalars. This could happen if an array |
| 3128 |
element were blessed (e.g., \f(CW\*(C`bless \e$a[0]\*(C'\fR) or if a closure referenced a |
| 3129 |
blessed variable (\f(CW\*(C`bless \emy @a; sub foo { @a }\*(C'\fR). |
| 3130 |
.Sp |
| 3131 |
Now there is an extra pass during global destruction to fire destructors on |
| 3132 |
any objects that might be left after the usual passes that check for |
| 3133 |
objects referenced by scalars [perl #36347]. |
| 3134 |
.IP \(bu 4 |
| 3135 |
Fixed a case where it was possible that a freed buffer may have been read |
| 3136 |
from when parsing a here document [perl #90128]. (5.14.1) |
| 3137 |
.IP \(bu 4 |
| 3138 |
\&\f(CWeach(\fR\f(CIARRAY\fR\f(CW)\fR is now wrapped in \f(CWdefined(...)\fR, like \f(CWeach(\fR\f(CIHASH\fR\f(CW)\fR, |
| 3139 |
inside a \f(CW\*(C`while\*(C'\fR condition [perl #90888]. |
| 3140 |
.IP \(bu 4 |
| 3141 |
A problem with context propagation when a \f(CW\*(C`do\*(C'\fR block is an argument to |
| 3142 |
\&\f(CW\*(C`return\*(C'\fR has been fixed. It used to cause \f(CW\*(C`undef\*(C'\fR to be returned in |
| 3143 |
certain cases of a \f(CW\*(C`return\*(C'\fR inside an \f(CW\*(C`if\*(C'\fR block which itself is followed by |
| 3144 |
another \f(CW\*(C`return\*(C'\fR. |
| 3145 |
.IP \(bu 4 |
| 3146 |
Calling \f(CW\*(C`index\*(C'\fR with a tainted constant no longer causes constants in |
| 3147 |
subsequently compiled code to become tainted [perl #64804]. |
| 3148 |
.IP \(bu 4 |
| 3149 |
Infinite loops like \f(CW\*(C`1 while 1\*(C'\fR used to stop \f(CW\*(C`strict \*(Aqsubs\*(Aq\*(C'\fR mode from |
| 3150 |
working for the rest of the block. |
| 3151 |
.IP \(bu 4 |
| 3152 |
For list assignments like \f(CW\*(C`($a,$b) = ($b,$a)\*(C'\fR, Perl has to make a copy of |
| 3153 |
the items on the right-hand side before assignment them to the left. For |
| 3154 |
efficiency's sake, it assigns the values on the right straight to the items |
| 3155 |
on the left if no one variable is mentioned on both sides, as in \f(CW\*(C`($a,$b) = |
| 3156 |
($c,$d)\*(C'\fR. The logic for determining when it can cheat was faulty, in that |
| 3157 |
\&\f(CW\*(C`&&\*(C'\fR and \f(CW\*(C`||\*(C'\fR on the right-hand side could fool it. So \f(CW\*(C`($a,$b) = |
| 3158 |
$some_true_value && ($b,$a)\*(C'\fR would end up assigning the value of \f(CW$b\fR to |
| 3159 |
both scalars. |
| 3160 |
.IP \(bu 4 |
| 3161 |
Perl no longer tries to apply lvalue context to the string in |
| 3162 |
\&\f(CW\*(C`("string", $variable) ||= 1\*(C'\fR (which used to be an error). Since the |
| 3163 |
left-hand side of \f(CW\*(C`||=\*(C'\fR is evaluated in scalar context, that's a scalar |
| 3164 |
comma operator, which gives all but the last item void context. There is |
| 3165 |
no such thing as void lvalue context, so it was a mistake for Perl to try |
| 3166 |
to force it [perl #96942]. |
| 3167 |
.IP \(bu 4 |
| 3168 |
\&\f(CW\*(C`caller\*(C'\fR no longer leaks memory when called from the DB package if |
| 3169 |
\&\f(CW@DB::args\fR was assigned to after the first call to \f(CW\*(C`caller\*(C'\fR. Carp |
| 3170 |
was triggering this bug [perl #97010]. (5.14.2) |
| 3171 |
.IP \(bu 4 |
| 3172 |
\&\f(CW\*(C`close\*(C'\fR and similar filehandle functions, when called on built-in global |
| 3173 |
variables (like \f(CW$+\fR), used to die if the variable happened to hold the |
| 3174 |
undefined value, instead of producing the usual "Use of uninitialized |
| 3175 |
value" warning. |
| 3176 |
.IP \(bu 4 |
| 3177 |
When autovivified file handles were introduced in Perl 5.6.0, \f(CW\*(C`readline\*(C'\fR |
| 3178 |
was inadvertently made to autovivify when called as \f(CWreadline($foo)\fR (but |
| 3179 |
not as \f(CW\*(C`<$foo>\*(C'\fR). It has now been fixed never to autovivify. |
| 3180 |
.IP \(bu 4 |
| 3181 |
Calling an undefined anonymous subroutine (e.g., what \f(CW$x\fR holds after |
| 3182 |
\&\f(CW\*(C`undef &{$x = sub{}}\*(C'\fR) used to cause a "Not a CODE reference" error, which |
| 3183 |
has been corrected to "Undefined subroutine called" [perl #71154]. |
| 3184 |
.IP \(bu 4 |
| 3185 |
Causing \f(CW@DB::args\fR to be freed between uses of \f(CW\*(C`caller\*(C'\fR no longer |
| 3186 |
results in a crash [perl #93320]. |
| 3187 |
.IP \(bu 4 |
| 3188 |
\&\f(CWsetpgrp($foo)\fR used to be equivalent to \f(CW\*(C`($foo, setpgrp)\*(C'\fR, because |
| 3189 |
\&\f(CW\*(C`setpgrp\*(C'\fR was ignoring its argument if there was just one. Now it is |
| 3190 |
equivalent to \f(CW\*(C`setpgrp($foo,0)\*(C'\fR. |
| 3191 |
.IP \(bu 4 |
| 3192 |
\&\f(CW\*(C`shmread\*(C'\fR was not setting the scalar flags correctly when reading from |
| 3193 |
shared memory, causing the existing cached numeric representation in the |
| 3194 |
scalar to persist [perl #98480]. |
| 3195 |
.IP \(bu 4 |
| 3196 |
\&\f(CW\*(C`++\*(C'\fR and \f(CW\*(C`\-\-\*(C'\fR now work on copies of globs, instead of dying. |
| 3197 |
.IP \(bu 4 |
| 3198 |
\&\f(CWsplice()\fR doesn't warn when truncating |
| 3199 |
.Sp |
| 3200 |
You can now limit the size of an array using \f(CW\*(C`splice(@a,MAX_LEN)\*(C'\fR without |
| 3201 |
worrying about warnings. |
| 3202 |
.IP \(bu 4 |
| 3203 |
\&\f(CW$$\fR is no longer tainted. Since this value comes directly from |
| 3204 |
\&\f(CWgetpid()\fR, it is always safe. |
| 3205 |
.IP \(bu 4 |
| 3206 |
The parser no longer leaks a filehandle if STDIN was closed before parsing |
| 3207 |
started [perl #37033]. |
| 3208 |
.IP \(bu 4 |
| 3209 |
\&\f(CW\*(C`die;\*(C'\fR with a non-reference, non-string, or magical (e.g., tainted) |
| 3210 |
value in $@ now properly propagates that value [perl #111654]. |
| 3211 |
.SH "Known Problems" |
| 3212 |
.IX Header "Known Problems" |
| 3213 |
.IP \(bu 4 |
| 3214 |
On Solaris, we have two kinds of failure. |
| 3215 |
.Sp |
| 3216 |
If \fImake\fR is Sun's \fImake\fR, we get an error about a badly formed macro |
| 3217 |
assignment in the \fIMakefile\fR. That happens when \fI./Configure\fR tries to |
| 3218 |
make depends. \fIConfigure\fR then exits 0, but further \fImake\fR\-ing fails. |
| 3219 |
.Sp |
| 3220 |
If \fImake\fR is \fIgmake\fR, \fIConfigure\fR completes, then we get errors related |
| 3221 |
to \fI/usr/include/stdbool.h\fR |
| 3222 |
.IP \(bu 4 |
| 3223 |
On Win32, a number of tests hang unless STDERR is redirected. The cause of |
| 3224 |
this is still under investigation. |
| 3225 |
.IP \(bu 4 |
| 3226 |
When building as root with a umask that prevents files from being |
| 3227 |
other-readable, \fIt/op/filetest.t\fR will fail. This is a test bug, not a |
| 3228 |
bug in perl's behavior. |
| 3229 |
.IP \(bu 4 |
| 3230 |
Configuring with a recent gcc and link-time-optimization, such as |
| 3231 |
\&\f(CW\*(C`Configure \-Doptimize=\*(Aq\-O2 \-flto\*(Aq\*(C'\fR fails |
| 3232 |
because the optimizer optimizes away some of Configure's tests. A |
| 3233 |
workaround is to omit the \f(CW\*(C`\-flto\*(C'\fR flag when running Configure, but add |
| 3234 |
it back in while actually building, something like |
| 3235 |
.Sp |
| 3236 |
.Vb 2 |
| 3237 |
\& sh Configure \-Doptimize=\-O2 |
| 3238 |
\& make OPTIMIZE=\*(Aq\-O2 \-flto\*(Aq |
| 3239 |
.Ve |
| 3240 |
.IP \(bu 4 |
| 3241 |
The following CPAN modules have test failures with perl 5.16. Patches have |
| 3242 |
been submitted for all of these, so hopefully there will be new releases |
| 3243 |
soon: |
| 3244 |
.RS 4 |
| 3245 |
.IP \(bu 4 |
| 3246 |
Date::Pcalc version 6.1 |
| 3247 |
.IP \(bu 4 |
| 3248 |
Module::CPANTS::Analyse version 0.85 |
| 3249 |
.Sp |
| 3250 |
This fails due to problems in Module::Find 0.10 and File::MMagic |
| 3251 |
1.27. |
| 3252 |
.IP \(bu 4 |
| 3253 |
PerlIO::Util version 0.72 |
| 3254 |
.RE |
| 3255 |
.RS 4 |
| 3256 |
.RE |
| 3257 |
.SH Acknowledgements |
| 3258 |
.IX Header "Acknowledgements" |
| 3259 |
Perl 5.16.0 represents approximately 12 months of development since Perl |
| 3260 |
5.14.0 and contains approximately 590,000 lines of changes across 2,500 |
| 3261 |
files from 139 authors. |
| 3262 |
.PP |
| 3263 |
Perl continues to flourish into its third decade thanks to a vibrant |
| 3264 |
community of users and developers. The following people are known to |
| 3265 |
have contributed the improvements that became Perl 5.16.0: |
| 3266 |
.PP |
| 3267 |
Aaron Crane, Abhijit Menon-Sen, Abigail, Alan Haggai Alavi, Alberto |
| 3268 |
Simões, Alexandr Ciornii, Andreas König, Andy Dougherty, Aristotle |
| 3269 |
Pagaltzis, Bo Johansson, Bo Lindbergh, Breno G. de Oliveira, brian d |
| 3270 |
foy, Brian Fraser, Brian Greenfield, Carl Hayter, Chas. Owens, |
| 3271 |
Chia-liang Kao, Chip Salzenberg, Chris 'BinGOs' Williams, Christian |
| 3272 |
Hansen, Christopher J. Madsen, chromatic, Claes Jacobsson, Claudio |
| 3273 |
Ramirez, Craig A. Berry, Damian Conway, Daniel Kahn Gillmor, Darin |
| 3274 |
McBride, Dave Rolsky, David Cantrell, David Golden, David Leadbeater, |
| 3275 |
David Mitchell, Dee Newcum, Dennis Kaarsemaker, Dominic Hargreaves, |
| 3276 |
Douglas Christopher Wilson, Eric Brine, Father Chrysostomos, Florian |
| 3277 |
Ragwitz, Frederic Briere, George Greer, Gerard Goossen, Gisle Aas, |
| 3278 |
H.Merijn Brand, Hojung Youn, Ian Goodacre, James E Keenan, Jan Dubois, |
| 3279 |
Jerry D. Hedden, Jesse Luehrs, Jesse Vincent, Jilles Tjoelker, Jim |
| 3280 |
Cromie, Jim Meyering, Joel Berger, Johan Vromans, Johannes Plunien, John |
| 3281 |
Hawkinson, John P. Linderman, John Peacock, Joshua ben Jore, Juerd |
| 3282 |
Waalboer, Karl Williamson, Karthik Rajagopalan, Keith Thompson, Kevin J. |
| 3283 |
Woolley, Kevin Ryde, Laurent Dami, Leo Lapworth, Leon Brocard, Leon |
| 3284 |
Timmermans, Louis Strous, Lukas Mai, Marc Green, Marcel Grünauer, Mark |
| 3285 |
A. Stratman, Mark Dootson, Mark Jason Dominus, Martin Hasch, Matthew |
| 3286 |
Horsfall, Max Maischein, Michael G Schwern, Michael Witten, Mike |
| 3287 |
Sheldrake, Moritz Lenz, Nicholas Clark, Niko Tyni, Nuno Carvalho, Pau |
| 3288 |
Amma, Paul Evans, Paul Green, Paul Johnson, Perlover, Peter John Acklam, |
| 3289 |
Peter Martini, Peter Scott, Phil Monsen, Pino Toscano, Rafael |
| 3290 |
Garcia-Suarez, Rainer Tammer, Reini Urban, Ricardo Signes, Robin Barker, |
| 3291 |
Rodolfo Carvalho, Salvador Fandiño, Sam Kimbrel, Samuel Thibault, Shawn |
| 3292 |
M Moore, Shigeya Suzuki, Shirakata Kentaro, Shlomi Fish, Sisyphus, |
| 3293 |
Slaven Rezic, Spiros Denaxas, Steffen Müller, Steffen Schwigon, Stephen |
| 3294 |
Bennett, Stephen Oberholtzer, Stevan Little, Steve Hay, Steve Peters, |
| 3295 |
Thomas Sibley, Thorsten Glaser, Timothe Litt, Todd Rinaldo, Tom |
| 3296 |
Christiansen, Tom Hukins, Tony Cook, Vadim Konovalov, Vincent Pit, |
| 3297 |
Vladimir Timofeev, Walt Mankowski, Yves Orton, Zefram, Zsbán Ambrus, |
| 3298 |
Ævar Arnfjörð Bjarmason. |
| 3299 |
.PP |
| 3300 |
The list above is almost certainly incomplete as it is automatically |
| 3301 |
generated from version control history. In particular, it does not |
| 3302 |
include the names of the (very much appreciated) contributors who |
| 3303 |
reported issues to the Perl bug tracker. |
| 3304 |
.PP |
| 3305 |
Many of the changes included in this version originated in the CPAN |
| 3306 |
modules included in Perl's core. We're grateful to the entire CPAN |
| 3307 |
community for helping Perl to flourish. |
| 3308 |
.PP |
| 3309 |
For a more complete list of all of Perl's historical contributors, |
| 3310 |
please see the \fIAUTHORS\fR file in the Perl source distribution. |
| 3311 |
.SH "Reporting Bugs" |
| 3312 |
.IX Header "Reporting Bugs" |
| 3313 |
If you find what you think is a bug, you might check the articles |
| 3314 |
recently posted to the comp.lang.perl.misc newsgroup and the perl |
| 3315 |
bug database at <http://rt.perl.org/perlbug/>. There may also be |
| 3316 |
information at <http://www.perl.org/>, the Perl Home Page. |
| 3317 |
.PP |
| 3318 |
If you believe you have an unreported bug, please run the perlbug |
| 3319 |
program included with your release. Be sure to trim your bug down |
| 3320 |
to a tiny but sufficient test case. Your bug report, along with the |
| 3321 |
output of \f(CW\*(C`perl \-V\*(C'\fR, will be sent off to [email protected] to be |
| 3322 |
analysed by the Perl porting team. |
| 3323 |
.PP |
| 3324 |
If the bug you are reporting has security implications, which make it |
| 3325 |
inappropriate to send to a publicly archived mailing list, then please |
| 3326 |
send it to perl5\-security\[email protected]. This points to a closed |
| 3327 |
subscription unarchived mailing list, which includes all core |
| 3328 |
committers, who will be able to help assess the impact of issues, figure |
| 3329 |
out a resolution, and help co-ordinate the release of patches to |
| 3330 |
mitigate or fix the problem across all platforms on which Perl is |
| 3331 |
supported. Please use this address only for security issues in the Perl |
| 3332 |
core, not for modules independently distributed on CPAN. |
| 3333 |
.SH "SEE ALSO" |
| 3334 |
.IX Header "SEE ALSO" |
| 3335 |
The \fIChanges\fR file for an explanation of how to view exhaustive details |
| 3336 |
on what changed. |
| 3337 |
.PP |
| 3338 |
The \fIINSTALL\fR file for how to build Perl. |
| 3339 |
.PP |
| 3340 |
The \fIREADME\fR file for general stuff. |
| 3341 |
.PP |
| 3342 |
The \fIArtistic\fR and \fICopying\fR files for copyright information. |