Commit | Line | Data |
---|---|---|
9021a1cf A |
1 | =head1 NAME |
2 | ||
3 | perldeprecation - list Perl deprecations | |
4 | ||
5 | =head1 DESCRIPTION | |
6 | ||
7 | The purpose of this document is to document what has been deprecated | |
8 | in Perl, and by which version the deprecated feature will disappear, | |
9 | or, for already removed features, when it was removed. | |
10 | ||
11 | This document will try to discuss what alternatives for the deprecated | |
12 | features are available. | |
13 | ||
14 | The deprecated features will be grouped by the version of Perl in | |
15 | which they will be removed. | |
16 | ||
9840d1d6 A |
17 | =head2 Perl 5.32 |
18 | ||
19 | =head3 Constants from lexical variables potentially modified elsewhere | |
20 | ||
21 | You wrote something like | |
22 | ||
23 | my $var; | |
24 | $sub = sub () { $var }; | |
25 | ||
26 | but $var is referenced elsewhere and could be modified after the C<sub> | |
27 | expression is evaluated. Either it is explicitly modified elsewhere | |
28 | (C<$var = 3>) or it is passed to a subroutine or to an operator like | |
29 | C<printf> or C<map>, which may or may not modify the variable. | |
30 | ||
31 | Traditionally, Perl has captured the value of the variable at that | |
32 | point and turned the subroutine into a constant eligible for inlining. | |
33 | In those cases where the variable can be modified elsewhere, this | |
34 | breaks the behavior of closures, in which the subroutine captures | |
35 | the variable itself, rather than its value, so future changes to the | |
36 | variable are reflected in the subroutine's return value. | |
37 | ||
38 | If you intended for the subroutine to be eligible for inlining, then | |
39 | make sure the variable is not referenced elsewhere, possibly by | |
40 | copying it: | |
41 | ||
42 | my $var2 = $var; | |
43 | $sub = sub () { $var2 }; | |
44 | ||
45 | If you do want this subroutine to be a closure that reflects future | |
46 | changes to the variable that it closes over, add an explicit C<return>: | |
47 | ||
48 | my $var; | |
49 | $sub = sub () { return $var }; | |
50 | ||
51 | This usage has been deprecated, and will no longer be allowed in Perl 5.32. | |
52 | ||
fada8285 | 53 | =head3 Use of strings with code points over 0xFF as arguments to C<vec> |
76aae383 KW |
54 | |
55 | C<vec> views its string argument as a sequence of bits. A string | |
56 | containing a code point over 0xFF is nonsensical. This usage is | |
57 | deprecated in Perl 5.28, and will be removed in Perl 5.32. | |
58 | ||
ba52ce15 KW |
59 | =head3 Use of code points over 0xFF in string bitwise operators |
60 | ||
61 | The string bitwise operators, C<&>, C<|>, C<^>, and C<~>, treat their | |
62 | operands as strings of bytes. As such, values above 0xFF are | |
63 | nonsensical. Some instances of these have been deprecated since Perl | |
64 | 5.24, and were made fatal in 5.28, but it turns out that in cases where | |
65 | the wide characters did not affect the end result, no deprecation | |
66 | notice was raised, and so remain legal. Now, all occurrences either are | |
67 | fatal or raise a deprecation warning, so that the remaining legal | |
68 | occurrences will be fatal in 5.32. | |
69 | ||
70 | An example of this is | |
71 | ||
72 | "" & "\x{100}" | |
73 | ||
74 | The wide character is not used in the C<&> operation because the left | |
75 | operand is shorter. This now warns anyway. | |
76 | ||
0c9c439d Z |
77 | =head3 hostname() doesn't accept any arguments |
78 | ||
79 | The function C<hostname()> in the L<Sys::Hostname> module has always | |
80 | been documented to be called with no arguments. Historically it has not | |
81 | enforced this, and has actually accepted and ignored any arguments. As a | |
82 | result, some users have got the mistaken impression that an argument does | |
83 | something useful. To avoid these bugs, the function is being made strict. | |
84 | Passing arguments was deprecated in Perl 5.28, and will become fatal in | |
85 | Perl 5.32. | |
86 | ||
0367231c KW |
87 | =head3 Unescaped left braces in regular expressions |
88 | ||
89 | The simple rule to remember, if you want to match a literal C<{> | |
90 | character (U+007B C<LEFT CURLY BRACKET>) in a regular expression | |
91 | pattern, is to escape each literal instance of it in some way. | |
92 | Generally easiest is to precede it with a backslash, like C<\{> | |
93 | or enclose it in square brackets (C<[{]>). If the pattern | |
94 | delimiters are also braces, any matching right brace (C<}>) should | |
95 | also be escaped to avoid confusing the parser, for example, | |
96 | ||
97 | qr{abc\{def\}ghi} | |
98 | ||
99 | Forcing literal C<{> characters to be escaped will enable the Perl | |
100 | language to be extended in various ways in future releases. To avoid | |
101 | needlessly breaking existing code, the restriction is is not enforced in | |
102 | contexts where there are unlikely to ever be extensions that could | |
b2247a87 KW |
103 | conflict with the use there of C<{> as a literal. A non-deprecation |
104 | warning that the left brace is being taken literally is raised in | |
105 | contexts where there could be confusion about it. | |
0367231c KW |
106 | |
107 | Literal uses of C<{> were deprecated in Perl 5.20, and some uses of it | |
108 | started to give deprecation warnings since. These cases were made fatal | |
109 | in Perl 5.26. Due to an oversight, not all cases of a use of a literal | |
110 | C<{> got a deprecation warning. Some cases started warning in Perl 5.26, | |
c96bf7f6 | 111 | and were made fatal in Perl 5.30. Other cases started in Perl 5.28, |
0367231c KW |
112 | and will be made fatal in 5.32. |
113 | ||
a0e213fc A |
114 | =head2 Perl 5.30 |
115 | ||
37398dc1 A |
116 | =head3 C<< $* >> is no longer supported |
117 | ||
118 | Before Perl 5.10, setting C<< $* >> to a true value globally enabled | |
119 | multi-line matching within a string. This relique from the past lost | |
120 | its special meaning in 5.10. Use of this variable will be a fatal error | |
121 | in Perl 5.30, freeing the variable up for a future special meaning. | |
122 | ||
123 | To enable multiline matching one should use the C<< /m >> regexp | |
124 | modifier (possibly in combination with C<< /s >>). This can be set | |
125 | on a per match bases, or can be enabled per lexical scope (including | |
126 | a whole file) with C<< use re '/m' >>. | |
127 | ||
128 | =head3 C<< $# >> is no longer supported | |
129 | ||
130 | This variable used to have a special meaning -- it could be used | |
131 | to control how numbers were formatted when printed. This seldom | |
132 | used functionality was removed in Perl 5.10. In order to free up | |
133 | the variable for a future special meaning, its use will be a fatal | |
134 | error in Perl 5.30. | |
135 | ||
33f0d962 | 136 | To specify how numbers are formatted when printed, one is advised |
37398dc1 A |
137 | to use C<< printf >> or C<< sprintf >> instead. |
138 | ||
c22e17d0 | 139 | =head3 Assigning non-zero to C<< $[ >> is fatal |
8e796115 DIM |
140 | |
141 | This variable (and the corresponding C<array_base> feature and | |
c22e17d0 | 142 | L<arybase> module) allowed changing the base for array and string |
8e796115 DIM |
143 | indexing operations. |
144 | ||
145 | Setting this to a non-zero value has been deprecated since Perl 5.12 and | |
c22e17d0 | 146 | throws a fatal error as of Perl 5.30. |
8e796115 | 147 | |
a0e213fc A |
148 | =head3 C<< File::Glob::glob() >> will disappear |
149 | ||
150 | C<< File::Glob >> has a function called C<< glob >>, which just calls | |
151 | C<< bsd_glob >>. However, its prototype is different from the prototype | |
152 | of C<< CORE::glob >>, and hence, C<< File::Glob::glob >> should not | |
153 | be used. | |
154 | ||
d1be68f6 A |
155 | C<< File::Glob::glob() >> was deprecated in Perl 5.8. A deprecation |
156 | message was issued from Perl 5.26 onwards, and the function will | |
157 | disappear in Perl 5.30. | |
a0e213fc A |
158 | |
159 | Code using C<< File::Glob::glob() >> should call | |
160 | C<< File::Glob::bsd_glob() >> instead. | |
161 | ||
0367231c | 162 | =head3 Unescaped left braces in regular expressions (for 5.30) |
286c9456 | 163 | |
0367231c | 164 | See L</Unescaped left braces in regular expressions> above. |
286c9456 | 165 | |
30b17cc1 A |
166 | =head3 Unqualified C<dump()> |
167 | ||
168 | Use of C<dump()> instead of C<CORE::dump()> was deprecated in Perl 5.8, | |
169 | and an unqualified C<dump()> will no longer be available in Perl 5.30. | |
170 | ||
171 | See L<perlfunc/dump>. | |
172 | ||
286c9456 | 173 | |
afb5c82e | 174 | =head3 Using my() in false conditional. |
c437f7ac A |
175 | |
176 | There has been a long-standing bug in Perl that causes a lexical variable | |
177 | not to be cleared at scope exit when its declaration includes a false | |
178 | conditional. Some people have exploited this bug to achieve a kind of | |
179 | static variable. Since we intend to fix this bug, we don't want people | |
180 | relying on this behavior. | |
181 | ||
182 | Instead, it's recommended one uses C<state> variables to achieve the | |
183 | same effect: | |
184 | ||
185 | use 5.10.0; | |
186 | sub count {state $counter; return ++ $counter} | |
187 | say count (); # Prints 1 | |
188 | say count (); # Prints 2 | |
189 | ||
190 | C<state> variables were introduced in Perl 5.10. | |
191 | ||
192 | Alternatively, you can achieve a similar static effect by | |
193 | declaring the variable in a separate block outside the function, eg | |
194 | ||
195 | sub f { my $x if 0; return $x++ } | |
196 | ||
197 | becomes | |
198 | ||
199 | { my $x; sub f { return $x++ } } | |
200 | ||
201 | The use of C<my()> in a false conditional has been deprecated in | |
202 | Perl 5.10, and it will become a fatal error in Perl 5.30. | |
203 | ||
1972ac5c A |
204 | |
205 | =head3 Reading/writing bytes from/to :utf8 handles. | |
206 | ||
207 | The sysread(), recv(), syswrite() and send() operators are | |
208 | deprecated on handles that have the C<:utf8> layer, either explicitly, or | |
209 | implicitly, eg., with the C<:encoding(UTF-16LE)> layer. | |
210 | ||
211 | Both sysread() and recv() currently use only the C<:utf8> flag for the stream, | |
212 | ignoring the actual layers. Since sysread() and recv() do no UTF-8 | |
213 | validation they can end up creating invalidly encoded scalars. | |
214 | ||
215 | Similarly, syswrite() and send() use only the C<:utf8> flag, otherwise ignoring | |
216 | any layers. If the flag is set, both write the value UTF-8 encoded, even if | |
217 | the layer is some different encoding, such as the example above. | |
218 | ||
219 | Ideally, all of these operators would completely ignore the C<:utf8> state, | |
220 | working only with bytes, but this would result in silently breaking existing | |
221 | code. To avoid this a future version of perl will throw an exception when | |
222 | any of sysread(), recv(), syswrite() or send() are called on handle with the | |
223 | C<:utf8> layer. | |
224 | ||
225 | In Perl 5.30, it will no longer be possible to use sysread(), recv(), | |
226 | syswrite() or send() to read or send bytes from/to :utf8 handles. | |
227 | ||
30573109 A |
228 | |
229 | =head3 Use of unassigned code point or non-standalone grapheme for a delimiter. | |
230 | ||
231 | A grapheme is what appears to a native-speaker of a language to be a | |
232 | character. In Unicode (and hence Perl) a grapheme may actually be | |
233 | several adjacent characters that together form a complete grapheme. For | |
234 | example, there can be a base character, like "R" and an accent, like a | |
4c821bda KW |
235 | circumflex "^", that appear to be a single character when displayed, |
236 | with the circumflex hovering over the "R". | |
237 | ||
238 | As of Perl 5.30, use of delimiters which are non-standalone graphemes is | |
239 | fatal, in order to move the language to be able to accept | |
240 | multi-character graphemes as delimiters. | |
241 | ||
c96bf7f6 | 242 | Also, as of Perl 5.30, delimiters which are unassigned code points |
4c821bda KW |
243 | but that may someday become assigned are prohibited. Otherwise, code |
244 | that works today would fail to compile if the currently unassigned | |
245 | delimiter ends up being something that isn't a stand-alone grapheme. | |
246 | Because Unicode is never going to assign L<non-character code | |
247 | points|perlunicode/Noncharacter code points>, nor L<code points that are | |
c96bf7f6 | 248 | above the legal Unicode maximum|perlunicode/Beyond Unicode code |
4c821bda | 249 | points>, those can be delimiters. |
30573109 | 250 | |
36d3e805 KW |
251 | =head3 In XS code, use of various macros dealing with UTF-8. |
252 | ||
253 | These macros will require an extra parameter in Perl 5.30: | |
254 | C<isALPHANUMERIC_utf8>, | |
255 | C<isASCII_utf8>, | |
256 | C<isBLANK_utf8>, | |
257 | C<isCNTRL_utf8>, | |
258 | C<isDIGIT_utf8>, | |
259 | C<isIDFIRST_utf8>, | |
260 | C<isPSXSPC_utf8>, | |
261 | C<isSPACE_utf8>, | |
262 | C<isVERTWS_utf8>, | |
263 | C<isWORDCHAR_utf8>, | |
264 | C<isXDIGIT_utf8>, | |
265 | C<isALPHANUMERIC_LC_utf8>, | |
266 | C<isALPHA_LC_utf8>, | |
267 | C<isASCII_LC_utf8>, | |
268 | C<isBLANK_LC_utf8>, | |
269 | C<isCNTRL_LC_utf8>, | |
270 | C<isDIGIT_LC_utf8>, | |
271 | C<isGRAPH_LC_utf8>, | |
272 | C<isIDCONT_LC_utf8>, | |
273 | C<isIDFIRST_LC_utf8>, | |
274 | C<isLOWER_LC_utf8>, | |
275 | C<isPRINT_LC_utf8>, | |
276 | C<isPSXSPC_LC_utf8>, | |
277 | C<isPUNCT_LC_utf8>, | |
278 | C<isSPACE_LC_utf8>, | |
279 | C<isUPPER_LC_utf8>, | |
280 | C<isWORDCHAR_LC_utf8>, | |
281 | C<isXDIGIT_LC_utf8>, | |
282 | C<toFOLD_utf8>, | |
283 | C<toLOWER_utf8>, | |
284 | C<toTITLE_utf8>, | |
285 | and | |
286 | C<toUPPER_utf8>. | |
287 | ||
288 | There is now a macro that corresponds to each one of these, simply by | |
289 | appending C<_safe> to the name. It takes the extra parameter. | |
290 | For example, C<isDIGIT_utf8_safe> corresponds to C<isDIGIT_utf8>, but | |
291 | takes the extra parameter, and its use doesn't generate a deprecation | |
292 | warning. All are documented in L<perlapi/Character case changing> and | |
293 | L<perlapi/Character classification>. | |
294 | ||
295 | You can change to use these versions at any time, or, if you can live | |
296 | with the deprecation messages, wait until 5.30 and add the parameter to | |
297 | the existing calls, without changing the names. | |
30573109 | 298 | |
c9680906 A |
299 | =head2 Perl 5.28 |
300 | ||
dcc013e3 | 301 | =head3 Attributes C<< :locked >> and C<< :unique >> |
c9680906 A |
302 | |
303 | The attributes C<< :locked >> (on code references) and C<< :unique >> | |
304 | (on array, hash and scalar references) have had no effect since | |
305 | Perl 5.005 and Perl 5.8.8 respectively. Their use has been deprecated | |
306 | since. | |
307 | ||
d1f1f359 | 308 | As of Perl 5.28, these attributes are syntax errors. Since the |
dcc013e3 A |
309 | attributes do not do anything, removing them from your code fixes |
310 | the syntax error; and removing them will not influence the behaviour | |
311 | of your code. | |
c9680906 | 312 | |
ac641426 | 313 | |
e5aa3f0b A |
314 | =head3 Bare here-document terminators |
315 | ||
316 | Perl has allowed you to use a bare here-document terminator to have the | |
317 | here-document end at the first empty line. This practise was deprecated | |
d1f1f359 | 318 | in Perl 5.000; as of Perl 5.28, using a bare here-document terminator |
dcc013e3 | 319 | throws a fatal error. |
e5aa3f0b | 320 | |
33f0d962 | 321 | You are encouraged to use the explicitly quoted form if you wish to |
e5aa3f0b A |
322 | use an empty line as the terminator of the here-document: |
323 | ||
324 | print <<""; | |
325 | Print this line. | |
326 | ||
327 | # Previous blank line ends the here-document. | |
328 | ||
329 | ||
d8940893 A |
330 | =head3 Setting $/ to a reference to a non-positive integer |
331 | ||
332 | You assigned a reference to a scalar to C<$/> where the | |
333 | referenced item is not a positive integer. In older perls this B<appeared> | |
334 | to work the same as setting it to C<undef> but was in fact internally | |
335 | different, less efficient and with very bad luck could have resulted in | |
336 | your file being split by a stringified form of the reference. | |
337 | ||
338 | In Perl 5.20.0 this was changed so that it would be B<exactly> the same as | |
339 | setting C<$/> to undef, with the exception that this warning would be | |
340 | thrown. | |
341 | ||
dcc013e3 A |
342 | As of Perl 5.28, setting C<$/> to a reference of a non-positive |
343 | integer throws a fatal error. | |
d8940893 A |
344 | |
345 | You are recommended to change your code to set C<$/> to C<undef> explicitly | |
346 | if you wish to slurp the file. | |
347 | ||
348 | ||
fcdb3ac1 A |
349 | =head3 Limit on the value of Unicode code points. |
350 | ||
dcc013e3 A |
351 | Unicode only allows code points up to 0x10FFFF, but Perl allows |
352 | much larger ones. Up till Perl 5.28, it was allowed to use code | |
353 | points exceeding the maximum value of an integer (C<IV_MAX>). | |
354 | However, that did break the perl interpreter in some constructs, | |
355 | including causing it to hang in a few cases. The known problem | |
356 | areas were in C<tr///>, regular expression pattern matching using | |
357 | quantifiers, as quote delimiters in C<qI<X>...I<X>> (where I<X> is | |
358 | the C<chr()> of a large code point), and as the upper limits in | |
359 | loops. | |
fcdb3ac1 | 360 | |
d1f1f359 | 361 | The use of out of range code points was deprecated in Perl 5.24; as of |
dcc013e3 | 362 | Perl 5.28 using a code point exceeding C<IV_MAX> throws a fatal error. |
fcdb3ac1 A |
363 | |
364 | If your code is to run on various platforms, keep in mind that the upper | |
dcc013e3 A |
365 | limit depends on the platform. It is much larger on 64-bit word sizes |
366 | than 32-bit ones. For 32-bit integers, C<IV_MAX> equals C<0x7FFFFFFF>, | |
367 | for 64-bit integers, C<IV_MAX> equals C<0x7FFFFFFFFFFFFFFF>. | |
fcdb3ac1 | 368 | |
db99d38d | 369 | |
6ef4f8b7 A |
370 | =head3 Use of comma-less variable list in formats. |
371 | ||
dcc013e3 | 372 | It was allowed to use a list of variables in a format, without |
6ef4f8b7 | 373 | separating them with commas. This usage has been deprecated |
d1f1f359 | 374 | for a long time, and as of Perl 5.28, this throws a fatal error. |
6ef4f8b7 | 375 | |
db99d38d A |
376 | =head3 Use of C<\N{}> |
377 | ||
378 | Use of C<\N{}> with nothing between the braces was deprecated in | |
be332ba0 | 379 | Perl 5.24, and throws a fatal error as of Perl 5.28. |
db99d38d A |
380 | |
381 | Since such a construct is equivalent to using an empty string, | |
382 | you are recommended to remove such C<\N{}> constructs. | |
383 | ||
122d6c09 A |
384 | =head3 Using the same symbol to open a filehandle and a dirhandle |
385 | ||
386 | It used to be legal to use C<open()> to associate both a | |
387 | filehandle and a dirhandle to the same symbol (glob or scalar). | |
388 | This idiom is likely to be confusing, and it was deprecated in | |
389 | Perl 5.10. | |
390 | ||
391 | Using the same symbol to C<open()> a filehandle and a dirhandle | |
d1f1f359 | 392 | throws a fatal error as of Perl 5.28. |
122d6c09 A |
393 | |
394 | You should be using two different symbols instead. | |
395 | ||
ac641426 A |
396 | =head3 ${^ENCODING} is no longer supported. |
397 | ||
398 | The special variable C<${^ENCODING}> was used to implement | |
399 | the C<encoding> pragma. Setting this variable to anything other | |
400 | than C<undef> was deprecated in Perl 5.22. Full deprecation | |
401 | of the variable happened in Perl 5.25.3. | |
402 | ||
dcc013e3 | 403 | Setting this variable to anything other than an undefined value |
d1f1f359 | 404 | throws a fatal error as of Perl 5.28. |
ac641426 | 405 | |
d9d53e86 | 406 | |
838ba4df A |
407 | =head3 C<< B::OP::terse >> |
408 | ||
409 | This method, which just calls C<< B::Concise::b_terse >>, has been | |
dcc013e3 | 410 | deprecated, and disappeared in Perl 5.28. Please use |
838ba4df A |
411 | C<< B::Concise >> instead. |
412 | ||
413 | ||
d9d53e86 | 414 | |
dcc013e3 | 415 | =head3 Use of inherited AUTOLOAD for non-method %s::%s() is no longer allowed |
d9d53e86 | 416 | |
dcc013e3 | 417 | As an (ahem) accidental feature, C<AUTOLOAD> subroutines were looked |
d9d53e86 A |
418 | up as methods (using the C<@ISA> hierarchy) even when the subroutines |
419 | to be autoloaded were called as plain functions (e.g. C<Foo::bar()>), | |
420 | not as methods (e.g. C<< Foo->bar() >> or C<< $obj->bar() >>). | |
421 | ||
dcc013e3 A |
422 | This bug was deprecated in Perl 5.004, has been rectified in Perl 5.28 |
423 | by using method lookup only for methods' C<AUTOLOAD>s. | |
d9d53e86 A |
424 | |
425 | The simple rule is: Inheritance will not work when autoloading | |
426 | non-methods. The simple fix for old code is: In any module that used | |
427 | to depend on inheriting C<AUTOLOAD> for non-methods from a base class | |
428 | named C<BaseClass>, execute C<*AUTOLOAD = \&BaseClass::AUTOLOAD> during | |
429 | startup. | |
430 | ||
431 | In code that currently says C<use AutoLoader; @ISA = qw(AutoLoader);> | |
432 | you should remove AutoLoader from @ISA and change C<use AutoLoader;> to | |
433 | C<use AutoLoader 'AUTOLOAD';>. | |
434 | ||
d9d53e86 | 435 | |
ecbcbef0 A |
436 | =head3 Use of code points over 0xFF in string bitwise operators |
437 | ||
438 | The string bitwise operators, C<&>, C<|>, C<^>, and C<~>, treat | |
439 | their operands as strings of bytes. As such, values above 0xFF | |
440 | are nonsensical. Using such code points with these operators | |
d1f1f359 | 441 | was deprecated in Perl 5.24, and is fatal as of Perl 5.28. |
ecbcbef0 | 442 | |
36d3e805 KW |
443 | =head3 In XS code, use of C<to_utf8_case()> |
444 | ||
f566c7cf | 445 | This function has been removed as of Perl 5.28; instead convert to call |
36d3e805 KW |
446 | the appropriate one of: |
447 | L<C<toFOLD_utf8_safe>|perlapi/toFOLD_utf8_safe>. | |
448 | L<C<toLOWER_utf8_safe>|perlapi/toLOWER_utf8_safe>, | |
449 | L<C<toTITLE_utf8_safe>|perlapi/toTITLE_utf8_safe>, | |
450 | or | |
451 | L<C<toUPPER_utf8_safe>|perlapi/toUPPER_utf8_safe>. | |
bfdc8cd3 | 452 | |
856f8944 A |
453 | =head2 Perl 5.26 |
454 | ||
455 | =head3 C<< --libpods >> in C<< Pod::Html >> | |
456 | ||
457 | Since Perl 5.18, the option C<< --libpods >> has been deprecated, and | |
458 | using this option did not do anything other than producing a warning. | |
459 | ||
d1f1f359 | 460 | The C<< --libpods >> option is no longer recognized as of Perl 5.26. |
856f8944 A |
461 | |
462 | ||
2560602c A |
463 | =head3 The utilities C<< c2ph >> and C<< pstruct >> |
464 | ||
465 | These old, perl3-era utilities have been deprecated in favour of | |
d1f1f359 | 466 | C<< h2xs >> for a long time. As of Perl 5.26, they have been removed. |
2560602c | 467 | |
d9d53e86 | 468 | |
4a29ab5e A |
469 | =head3 Trapping C<< $SIG {__DIE__} >> other than during program exit. |
470 | ||
471 | The C<$SIG{__DIE__}> hook is called even inside an C<eval()>. It was | |
472 | never intended to happen this way, but an implementation glitch made | |
473 | this possible. This used to be deprecated, as it allowed strange action | |
474 | at a distance like rewriting a pending exception in C<$@>. Plans to | |
475 | rectify this have been scrapped, as users found that rewriting a | |
476 | pending exception is actually a useful feature, and not a bug. | |
477 | ||
478 | Perl never issued a deprecation warning for this; the deprecation | |
479 | was by documentation policy only. But this deprecation has been | |
d1f1f359 | 480 | lifted as of Perl 5.26. |
4a29ab5e A |
481 | |
482 | ||
24ca4586 A |
483 | =head3 Malformed UTF-8 string in "%s" |
484 | ||
485 | This message indicates a bug either in the Perl core or in XS | |
486 | code. Such code was trying to find out if a character, allegedly | |
487 | stored internally encoded as UTF-8, was of a given type, such as | |
488 | being punctuation or a digit. But the character was not encoded | |
489 | in legal UTF-8. The C<%s> is replaced by a string that can be used | |
490 | by knowledgeable people to determine what the type being checked | |
491 | against was. | |
492 | ||
493 | Passing malformed strings was deprecated in Perl 5.18, and | |
494 | became fatal in Perl 5.26. | |
495 | ||
496 | ||
9021a1cf A |
497 | =head2 Perl 5.24 |
498 | ||
499 | =head3 Use of C<< *glob{FILEHANDLE} >> | |
500 | ||
d1be68f6 | 501 | The use of C<< *glob{FILEHANDLE} >> was deprecated in Perl 5.8. |
9021a1cf A |
502 | The intention was to use C<< *glob{IO} >> instead, for which |
503 | C<< *glob{FILEHANDLE} >> is an alias. | |
504 | ||
d1be68f6 | 505 | However, this feature was undeprecated in Perl 5.24. |
9021a1cf | 506 | |
46d7f3c1 A |
507 | =head3 Calling POSIX::%s() is deprecated |
508 | ||
509 | The following functions in the C<POSIX> module are no longer available: | |
510 | C<isalnum>, C<isalpha>, C<iscntrl>, C<isdigit>, C<isgraph>, C<islower>, | |
511 | C<isprint>, C<ispunct>, C<isspace>, C<isupper>, and C<isxdigit>. The | |
512 | functions are buggy and don't work on UTF-8 encoded strings. See their | |
513 | entries in L<POSIX> for more information. | |
514 | ||
d1be68f6 | 515 | The functions were deprecated in Perl 5.20, and removed in Perl 5.24. |
46d7f3c1 A |
516 | |
517 | ||
c4d8d6a2 A |
518 | =head2 Perl 5.16 |
519 | ||
520 | =head3 Use of %s on a handle without * is deprecated | |
521 | ||
522 | It used to be possible to use C<tie>, C<tied> or C<untie> on a scalar | |
523 | while the scalar holds a typeglob. This caused its filehandle to be | |
524 | tied. It left no way to tie the scalar itself when it held a typeglob, | |
525 | and no way to untie a scalar that had had a typeglob assigned to it. | |
526 | ||
d1be68f6 | 527 | This was deprecated in Perl 5.14, and the bug was fixed in Perl 5.16. |
c4d8d6a2 A |
528 | |
529 | So now C<tie $scalar> will always tie the scalar, not the handle it holds. | |
530 | To tie the handle, use C<tie *$scalar> (with an explicit asterisk). The same | |
531 | applies to C<tied *$scalar> and C<untie *$scalar>. | |
532 | ||
533 | ||
9021a1cf A |
534 | =head1 SEE ALSO |
535 | ||
536 | L<warnings>, L<diagnostics>. | |
537 | ||
538 | =cut |