1PERL58DELTA(1) Perl Programmers Reference Guide PERL58DELTA(1)
2
6 perl58delta - what is new for perl v5.8.0
7
9 This document describes differences between the 5.6.0 release and the
10 5.8.0 release.
11 Many of the bug fixes in 5.8.0 were already seen in the 5.6.1
13 maintenance release since the two releases were kept closely
14 coordinated (while 5.8.0 was still called 5.7.something).
15 Changes that were integrated into the 5.6.1 release are marked "[561]".
17 Many of these changes have been further developed since 5.6.1 was
18 released, those are marked "[561+]".
19 You can see the list of changes in the 5.6.1 release (both from the
21 5.005_03 release and the 5.6.0 release) by reading perl561delta.
22
24 • Better Unicode support
25 • New IO Implementation
27 • New Thread Implementation
29 • Better Numeric Accuracy
31 • Safe Signals
33 • Many New Modules
35 • More Extensive Regression Testing
37
39 Binary Incompatibility
40 Perl 5.8 is not binary compatible with earlier releases of Perl.
41 You have to recompile your XS modules.
43 (Pure Perl modules should continue to work.)
45 The major reason for the discontinuity is the new IO architecture
47 called PerlIO. PerlIO is the default configuration because without it
48 many new features of Perl 5.8 cannot be used. In other words: you just
49 have to recompile your modules containing XS code, sorry about that.
50 In future releases of Perl, non-PerlIO aware XS modules may become
52 completely unsupported. This shouldn't be too difficult for module
53 authors, however: PerlIO has been designed as a drop-in replacement (at
54 the source code level) for the stdio interface.
55 Depending on your platform, there are also other reasons why we decided
57 to break binary compatibility, please read on.
58 64-bit platforms and malloc
60 If your pointers are 64 bits wide, the Perl malloc is no longer being
61 used because it does not work well with 8-byte pointers. Also, usually
62 the system mallocs on such platforms are much better optimized for such
63 large memory models than the Perl malloc. Some memory-hungry Perl
64 applications like the PDL don't work well with Perl's malloc. Finally,
65 other applications than Perl (such as mod_perl) tend to prefer the
66 system malloc. Such platforms include Alpha and 64-bit HPPA, MIPS,
67 PPC, and Sparc.
68 AIX Dynaloading
70 The AIX dynaloading now uses in AIX releases 4.3 and newer the native
71 dlopen interface of AIX instead of the old emulated interface. This
72 change will probably break backward compatibility with compiled
73 modules. The change was made to make Perl more compliant with other
74 applications like mod_perl which are using the AIX native interface.
75 Attributes for "my" variables now handled at run-time
77 The "my EXPR : ATTRS" syntax now applies variable attributes at run-
78 time. (Subroutine and "our" variables still get attributes applied at
79 compile-time.) See attributes for additional details. In particular,
80 however, this allows variable attributes to be useful for "tie"
81 interfaces, which was a deficiency of earlier releases. Note that the
82 new semantics doesn't work with the Attribute::Handlers module (as of
83 version 0.76).
84 Socket Extension Dynamic in VMS
86 The Socket extension is now dynamically loaded instead of being
87 statically built in. This may or may not be a problem with ancient
88 TCP/IP stacks of VMS: we do not know since we weren't able to test Perl
89 in such configurations.
90 IEEE-format Floating Point Default on OpenVMS Alpha
92 Perl now uses IEEE format (T_FLOAT) as the default internal floating
93 point format on OpenVMS Alpha, potentially breaking binary
94 compatibility with external libraries or existing data. G_FLOAT is
95 still available as a configuration option. The default on VAX
96 (D_FLOAT) has not changed.
97 New Unicode Semantics (no more "use utf8", almost)
99 Previously in Perl 5.6 to use Unicode one would say "use utf8" and then
100 the operations (like string concatenation) were Unicode-aware in that
101 lexical scope.
102 This was found to be an inconvenient interface, and in Perl 5.8 the
104 Unicode model has completely changed: now the "Unicodeness" is bound to
105 the data itself, and for most of the time "use utf8" is not needed at
106 all. The only remaining use of "use utf8" is when the Perl script
107 itself has been written in the UTF-8 encoding of Unicode. (UTF-8 has
108 not been made the default since there are many Perl scripts out there
109 that are using various national eight-bit character sets, which would
110 be illegal in UTF-8.)
111 See perluniintro for the explanation of the current model, and utf8 for
113 the current use of the utf8 pragma.
114 New Unicode Properties
116 Unicode scripts are now supported. Scripts are similar to (and superior
117 to) Unicode blocks. The difference between scripts and blocks is that
118 scripts are the glyphs used by a language or a group of languages,
119 while the blocks are more artificial groupings of (mostly) 256
120 characters based on the Unicode numbering.
121 In general, scripts are more inclusive, but not universally so. For
123 example, while the script "Latin" includes all the Latin characters and
124 their various diacritic-adorned versions, it does not include the
125 various punctuation or digits (since they are not solely "Latin").
126 A number of other properties are now supported, including "\p{L&}",
128 "\p{Any}" "\p{Assigned}", "\p{Unassigned}", "\p{Blank}" [561] and
129 "\p{SpacePerl}" [561] (along with their "\P{...}" versions, of course).
130 See perlunicode for details, and more additions.
131 The "In" or "Is" prefix to names used with the "\p{...}" and "\P{...}"
133 are now almost always optional. The only exception is that a "In"
134 prefix is required to signify a Unicode block when a block name
135 conflicts with a script name. For example, "\p{Tibetan}" refers to the
136 script, while "\p{InTibetan}" refers to the block. When there is no
137 name conflict, you can omit the "In" from the block name (e.g.
138 "\p{BraillePatterns}"), but to be safe, it's probably best to always
139 use the "In").
140 REF(...) Instead Of SCALAR(...)
142 A reference to a reference now stringifies as "REF(0x81485ec)" instead
143 of "SCALAR(0x81485ec)" in order to be more consistent with the return
144 value of ref().
145 pack/unpack D/F recycled
147 The undocumented pack/unpack template letters D/F have been recycled
148 for better use: now they stand for long double (if supported by the
149 platform) and NV (Perl internal floating point type). (They used to be
150 aliases for d/f, but you never knew that.)
151 glob() now returns filenames in alphabetical order
153 The list of filenames from glob() (or <...>) is now by default sorted
154 alphabetically to be csh-compliant (which is what happened before in
155 most Unix platforms). (bsd_glob() does still sort platform natively,
156 ASCII or EBCDIC, unless GLOB_ALPHASORT is specified.) [561]
157 Deprecations
159 • The semantics of bless(REF, REF) were unclear and until someone
160 proves it to make some sense, it is forbidden.
161 • The obsolete chat2 library that should never have been allowed to
163 escape the laboratory has been decommissioned.
164 • Using chdir("") or chdir(undef) instead of explicit chdir() is
166 doubtful. A failure (think chdir(some_function()) can lead into
167 unintended chdir() to the home directory, therefore this behaviour
168 is deprecated.
169 • The builtin dump() function has probably outlived most of its
171 usefulness. The core-dumping functionality will remain in future
172 available as an explicit call to "CORE::dump()", but in future
173 releases the behaviour of an unqualified "dump()" call may change.
174 • The very dusty examples in the eg/ directory have been removed.
176 Suggestions for new shiny examples welcome but the main issue is
177 that the examples need to be documented, tested and (most
178 importantly) maintained.
179 • The (bogus) escape sequences \8 and \9 now give an optional warning
181 ("Unrecognized escape passed through"). There is no need to
182 \-escape any "\w" character.
183 • The *glob{FILEHANDLE} is deprecated, use *glob{IO} instead.
185 • The "package;" syntax ("package" without an argument) has been
187 deprecated. Its semantics were never that clear and its
188 implementation even less so. If you have used that feature to
189 disallow all but fully qualified variables, "use strict;" instead.
190 • The unimplemented POSIX regex features [[.cc.]] and [[=c=]] are
192 still recognised but now cause fatal errors. The previous
193 behaviour of ignoring them by default and warning if requested was
194 unacceptable since it, in a way, falsely promised that the features
195 could be used.
196 • In future releases, non-PerlIO aware XS modules may become
198 completely unsupported. Since PerlIO is a drop-in replacement for
199 stdio at the source code level, this shouldn't be that drastic a
200 change.
201 • Previous versions of perl and some readings of some sections of
203 Camel III implied that the ":raw" "discipline" was the inverse of
204 ":crlf". Turning off "clrfness" is no longer enough to make a
205 stream truly binary. So the PerlIO ":raw" layer (or "discipline",
206 to use the Camel book's older terminology) is now formally defined
207 as being equivalent to binmode(FH) - which is in turn defined as
208 doing whatever is necessary to pass each byte as-is without any
209 translation. In particular binmode(FH) - and hence ":raw" - will
210 now turn off both CRLF and UTF-8 translation and remove other
211 layers (e.g. :encoding()) which would modify byte stream.
212 • The current user-visible implementation of pseudo-hashes (the weird
214 use of the first array element) is deprecated starting from Perl
215 5.8.0 and will be removed in Perl 5.10.0, and the feature will be
216 implemented differently. Not only is the current interface rather
217 ugly, but the current implementation slows down normal array and
218 hash use quite noticeably. The "fields" pragma interface will
219 remain available. The restricted hashes interface is expected to
220 be the replacement interface (see Hash::Util). If your existing
221 programs depends on the underlying implementation, consider using
222 Class::PseudoHash from CPAN.
223 • The syntaxes "@a->[...]" and "%h->{...}" have now been deprecated.
225 • After years of trying, suidperl is considered to be too complex to
227 ever be considered truly secure. The suidperl functionality is
228 likely to be removed in a future release.
229 • The 5.005 threads model (module "Thread") is deprecated and
231 expected to be removed in Perl 5.10. Multithreaded code should be
232 migrated to the new ithreads model (see threads, threads::shared
233 and perlthrtut).
234 • The long deprecated uppercase aliases for the string comparison
236 operators (EQ, NE, LT, LE, GE, GT) have now been removed.
237 • The tr///C and tr///U features have been removed and will not
239 return; the interface was a mistake. Sorry about that. For
240 similar functionality, see pack('U0', ...) and pack('C0', ...).
241 [561]
242 • Earlier Perls treated "sub foo (@bar)" as equivalent to "sub foo
244 (@)". The prototypes are now checked better at compile-time for
245 invalid syntax. An optional warning is generated ("Illegal
246 character in prototype...") but this may be upgraded to a fatal
247 error in a future release.
248 • The "exec LIST" and "system LIST" operations now produce warnings
250 on tainted data and in some future release they will produce fatal
251 errors.
252 • The existing behaviour when localising tied arrays and hashes is
254 wrong, and will be changed in a future release, so do not rely on
255 the existing behaviour. See "Localising Tied Arrays and Hashes Is
256 Broken".
257
259 Unicode Overhaul
260 Unicode in general should be now much more usable than in Perl 5.6.0
261 (or even in 5.6.1). Unicode can be used in hash keys, Unicode in
262 regular expressions should work now, Unicode in tr/// should work now,
263 Unicode in I/O should work now. See perluniintro for introduction and
264 perlunicode for details.
265 • The Unicode Character Database coming with Perl has been upgraded
267 to Unicode 3.2.0. For more information, see
268 http://www.unicode.org/ . [561+] (5.6.1 has UCD 3.0.1.)
269 • For developers interested in enhancing Perl's Unicode capabilities:
271 almost all the UCD files are included with the Perl distribution in
272 the lib/unicore subdirectory. The most notable omission, for space
273 considerations, is the Unihan database.
274 • The properties \p{Blank} and \p{SpacePerl} have been added. "Blank"
276 is like C isblank(), that is, it contains only "horizontal
277 whitespace" (the space character is, the newline isn't), and the
278 "SpacePerl" is the Unicode equivalent of "\s" (\p{Space} isn't,
279 since that includes the vertical tabulator character, whereas "\s"
280 doesn't.)
281 See "New Unicode Properties" earlier in this document for
283 additional information on changes with Unicode properties.
284 PerlIO is Now The Default
286 • IO is now by default done via PerlIO rather than system's "stdio".
287 PerlIO allows "layers" to be "pushed" onto a file handle to alter
288 the handle's behaviour. Layers can be specified at open time via
289 3-arg form of open:
290 open($fh,'>:crlf :utf8', $path) || ...
292 or on already opened handles via extended "binmode":
294 binmode($fh,':encoding(iso-8859-7)');
296 The built-in layers are: unix (low level read/write), stdio (as in
298 previous Perls), perlio (re-implementation of stdio buffering in a
299 portable manner), crlf (does CRLF <=> "\n" translation as on Win32,
300 but available on any platform). A mmap layer may be available if
301 platform supports it (mostly Unixes).
302 Layers to be applied by default may be specified via the 'open'
304 pragma.
305 See "Installation and Configuration Improvements" for the effects
307 of PerlIO on your architecture name.
308 • If your platform supports fork(), you can use the list form of
310 "open" for pipes. For example:
311 open KID_PS, "-|", "ps", "aux" or die $!;
313 forks the ps(1) command (without spawning a shell, as there are
315 more than three arguments to open()), and reads its standard output
316 via the "KID_PS" filehandle. See perlipc.
317 • File handles can be marked as accepting Perl's internal encoding of
319 Unicode (UTF-8 or UTF-EBCDIC depending on platform) by a pseudo
320 layer ":utf8" :
321 open($fh,">:utf8","Uni.txt");
323 Note for EBCDIC users: the pseudo layer ":utf8" is erroneously
325 named for you since it's not UTF-8 what you will be getting but
326 instead UTF-EBCDIC. See perlunicode, utf8, and
327 http://www.unicode.org/reports/tr16/ for more information. In
328 future releases this naming may change. See perluniintro for more
329 information about UTF-8.
330 • If your environment variables (LC_ALL, LC_CTYPE, LANG) look like
332 you want to use UTF-8 (any of the variables match "/utf-?8/i"),
333 your STDIN, STDOUT, STDERR handles and the default open layer (see
334 open) are marked as UTF-8. (This feature, like other new features
335 that combine Unicode and I/O, work only if you are using PerlIO,
336 but that's the default.)
337 Note that after this Perl really does assume that everything is
339 UTF-8: for example if some input handle is not, Perl will probably
340 very soon complain about the input data like this "Malformed UTF-8
341 ..." since any old eight-bit data is not legal UTF-8.
342 Note for code authors: if you want to enable your users to use
344 UTF-8 as their default encoding but in your code still have eight-
345 bit I/O streams (such as images or zip files), you need to
346 explicitly open() or binmode() with ":bytes" (see "open" in
347 perlfunc and "binmode" in perlfunc), or you can just use
348 "binmode(FH)" (nice for pre-5.8.0 backward compatibility).
349 • File handles can translate character encodings from/to Perl's
351 internal Unicode form on read/write via the ":encoding()" layer.
352 • File handles can be opened to "in memory" files held in Perl
354 scalars via:
355 open($fh,'>', \$variable) || ...
357 • Anonymous temporary files are available without need to 'use
359 FileHandle' or other module via
360 open($fh,"+>", undef) || ...
362 That is a literal undef, not an undefined value.
364 ithreads
366 The new interpreter threads ("ithreads" for short) implementation of
367 multithreading, by Arthur Bergman, replaces the old "5.005 threads"
368 implementation. In the ithreads model any data sharing between threads
369 must be explicit, as opposed to the model where data sharing was
370 implicit. See threads and threads::shared, and perlthrtut.
371 As a part of the ithreads implementation Perl will also use any
373 necessary and detectable reentrant libc interfaces.
374 Restricted Hashes
376 A restricted hash is restricted to a certain set of keys, no keys
377 outside the set can be added. Also individual keys can be restricted
378 so that the key cannot be deleted and the value cannot be changed. No
379 new syntax is involved: the Hash::Util module is the interface.
380 Safe Signals
382 Perl used to be fragile in that signals arriving at inopportune moments
383 could corrupt Perl's internal state. Now Perl postpones handling of
384 signals until it's safe (between opcodes).
385 This change may have surprising side effects because signals no longer
387 interrupt Perl instantly. Perl will now first finish whatever it was
388 doing, like finishing an internal operation (like sort()) or an
389 external operation (like an I/O operation), and only then look at any
390 arrived signals (and before starting the next operation). No more
391 corrupt internal state since the current operation is always finished
392 first, but the signal may take more time to get heard. Note that
393 breaking out from potentially blocking operations should still work,
394 though.
395 Understanding of Numbers
397 In general a lot of fixing has happened in the area of Perl's
398 understanding of numbers, both integer and floating point. Since in
399 many systems the standard number parsing functions like "strtoul()" and
400 "atof()" seem to have bugs, Perl tries to work around their
401 deficiencies. This results hopefully in more accurate numbers.
402 Perl now tries internally to use integer values in numeric conversions
404 and basic arithmetics (+ - * /) if the arguments are integers, and
405 tries also to keep the results stored internally as integers. This
406 change leads to often slightly faster and always less lossy
407 arithmetics. (Previously Perl always preferred floating point numbers
408 in its math.)
409 Arrays now always interpolate into double-quoted strings [561]
411 In double-quoted strings, arrays now interpolate, no matter what. The
412 behavior in earlier versions of perl 5 was that arrays would
413 interpolate into strings if the array had been mentioned before the
414 string was compiled, and otherwise Perl would raise a fatal compile-
415 time error. In versions 5.000 through 5.003, the error was
416 Literal @example now requires backslash
418 In versions 5.004_01 through 5.6.0, the error was
420 In string, @example now must be written as \@example
422 The idea here was to get people into the habit of writing
424 "fred\@example.com" when they wanted a literal "@" sign, just as they
425 have always written "Give me back my \$5" when they wanted a literal
426 "$" sign.
427 Starting with 5.6.1, when Perl now sees an "@" sign in a double-quoted
429 string, it always attempts to interpolate an array, regardless of
430 whether or not the array has been used or declared already. The fatal
431 error has been downgraded to an optional warning:
432 Possible unintended interpolation of @example in string
434 This warns you that "fred@example.com" is going to turn into "fred.com"
436 if you don't backslash the "@". See
437 http://perl.plover.com/at-error.html for more details about the history
438 here.
439 Miscellaneous Changes
441 • AUTOLOAD is now lvaluable, meaning that you can add the :lvalue
442 attribute to AUTOLOAD subroutines and you can assign to the
443 AUTOLOAD return value.
444 • The $Config{byteorder} (and corresponding BYTEORDER in config.h)
446 was previously wrong in platforms if sizeof(long) was 4, but
447 sizeof(IV) was 8. The byteorder was only sizeof(long) bytes long
448 (1234 or 4321), but now it is correctly sizeof(IV) bytes long,
449 (12345678 or 87654321). (This problem didn't affect Windows
450 platforms.)
451 Also, $Config{byteorder} is now computed dynamically--this is more
453 robust with "fat binaries" where an executable image contains
454 binaries for more than one binary platform, and when cross-
455 compiling.
456 • "perl -d:Module=arg,arg,arg" now works (previously one couldn't
458 pass in multiple arguments.)
459 • "do" followed by a bareword now ensures that this bareword isn't a
461 keyword (to avoid a bug where "do q(foo.pl)" tried to call a
462 subroutine called "q"). This means that for example instead of "do
463 format()" you must write "do &format()".
464 • The builtin dump() now gives an optional warning "dump() better
466 written as CORE::dump()", meaning that by default "dump(...)" is
467 resolved as the builtin dump() which dumps core and aborts, not as
468 (possibly) user-defined "sub dump". To call the latter, qualify
469 the call as "&dump(...)". (The whole dump() feature is to
470 considered deprecated, and possibly removed/changed in future
471 releases.)
472 • chomp() and chop() are now overridable. Note, however, that their
474 prototype (as given by "prototype("CORE::chomp")" is undefined,
475 because it cannot be expressed and therefore one cannot really
476 write replacements to override these builtins.
477 • END blocks are now run even if you exit/die in a BEGIN block.
479 Internally, the execution of END blocks is now controlled by
480 PL_exit_flags & PERL_EXIT_DESTRUCT_END. This enables the new
481 behaviour for Perl embedders. This will default in 5.10. See
482 perlembed.
483 • Formats now support zero-padded decimal fields.
485 • Although "you shouldn't do that", it was possible to write code
487 that depends on Perl's hashed key order (Data::Dumper does this).
488 The new algorithm "One-at-a-Time" produces a different hashed key
489 order. More details are in "Performance Enhancements".
490 • lstat(FILEHANDLE) now gives a warning because the operation makes
492 no sense. In future releases this may become a fatal error.
493 • Spurious syntax errors generated in certain situations, when glob()
495 caused File::Glob to be loaded for the first time, have been fixed.
496 [561]
497 • Lvalue subroutines can now return "undef" in list context.
499 However, the lvalue subroutine feature still remains experimental.
500 [561+]
501 • A lost warning "Can't declare ... dereference in my" has been
503 restored (Perl had it earlier but it became lost in later
504 releases.)
505 • A new special regular expression variable has been introduced: $^N,
507 which contains the most-recently closed group (submatch).
508 • "no Module;" does not produce an error even if Module does not have
510 an unimport() method. This parallels the behavior of "use" vis-a-
511 vis "import". [561]
512 • The numerical comparison operators return "undef" if either operand
514 is a NaN. Previously the behaviour was unspecified.
515 • "our" can now have an experimental optional attribute "unique" that
517 affects how global variables are shared among multiple
518 interpreters, see "our" in perlfunc.
519 • The following builtin functions are now overridable: each(),
521 keys(), pop(), push(), shift(), splice(), unshift(). [561]
522 • "pack() / unpack()" can now group template letters with "()" and
524 then apply repetition/count modifiers on the groups.
525 • "pack() / unpack()" can now process the Perl internal numeric
527 types: IVs, UVs, NVs-- and also long doubles, if supported by the
528 platform. The template letters are "j", "J", "F", and "D".
529 • "pack('U0a*', ...)" can now be used to force a string to UTF-8.
531 • my __PACKAGE__ $obj now works. [561]
533 • POSIX::sleep() now returns the number of unslept seconds (as the
535 POSIX standard says), as opposed to CORE::sleep() which returns the
536 number of slept seconds.
537 • printf() and sprintf() now support parameter reordering using the
539 "%\d+\$" and "*\d+\$" syntaxes. For example
540 printf "%2\$s %1\$s\n", "foo", "bar";
542 will print "bar foo\n". This feature helps in writing
544 internationalised software, and in general when the order of the
545 parameters can vary.
546 • The (\&) prototype now works properly. [561]
548 • prototype(\[$@%&]) is now available to implicitly create references
550 (useful for example if you want to emulate the tie() interface).
551 • A new command-line option, "-t" is available. It is the little
553 brother of "-T": instead of dying on taint violations, lexical
554 warnings are given. This is only meant as a temporary debugging
555 aid while securing the code of old legacy applications. This is
556 not a substitute for -T.
557 • In other taint news, the "exec LIST" and "system LIST" have now
559 been considered too risky (think "exec @ARGV": it can start any
560 program with any arguments), and now the said forms cause a warning
561 under lexical warnings. You should carefully launder the arguments
562 to guarantee their validity. In future releases of Perl the forms
563 will become fatal errors so consider starting laundering now.
564 • Tied hash interfaces are now required to have the EXISTS and DELETE
566 methods (either own or inherited).
567 • If tr/// is just counting characters, it doesn't attempt to modify
569 its target.
570 • untie() will now call an UNTIE() hook if it exists. See perltie
572 for details. [561]
573 • "utime" in perlfunc now supports "utime undef, undef, @files" to
575 change the file timestamps to the current time.
576 • The rules for allowing underscores (underbars) in numeric constants
578 have been relaxed and simplified: now you can have an underscore
579 simply between digits.
580 • Rather than relying on C's argv[0] (which may not contain a full
582 pathname) where possible $^X is now set by asking the operating
583 system. (eg by reading /proc/self/exe on Linux, /proc/curproc/file
584 on FreeBSD)
585 • A new variable, "${^TAINT}", indicates whether taint mode is
587 enabled.
588 • You can now override the readline() builtin, and this overrides
590 also the <FILEHANDLE> angle bracket operator.
591 • The command-line options -s and -F are now recognized on the
593 shebang (#!) line.
594 • Use of the "/c" match modifier without an accompanying "/g"
596 modifier elicits a new warning: "Use of /c modifier is meaningless
597 without /g".
598 Use of "/c" in substitutions, even with "/g", elicits "Use of /c
600 modifier is meaningless in s///".
601 Use of "/g" with "split" elicits "Use of /g modifier is meaningless
603 in split".
604 • Support for the "CLONE" special subroutine had been added. With
606 ithreads, when a new thread is created, all Perl data is cloned,
607 however non-Perl data cannot be cloned automatically. In "CLONE"
608 you can do whatever you need to do, like for example handle the
609 cloning of non-Perl data, if necessary. "CLONE" will be executed
610 once for every package that has it defined or inherited. It will
611 be called in the context of the new thread, so all modifications
612 are made in the new area.
613 See perlmod
615
617 New Modules and Pragmata
618 • "Attribute::Handlers", originally by Damian Conway and now
619 maintained by Arthur Bergman, allows a class to define attribute
620 handlers.
621 package MyPack;
623 use Attribute::Handlers;
624 sub Wolf :ATTR(SCALAR) { print "howl!\n" }
625 # later, in some package using or inheriting from MyPack...
627 my MyPack $Fluffy : Wolf; # the attribute handler Wolf will be called
629 Both variables and routines can have attribute handlers. Handlers
631 can be specific to type (SCALAR, ARRAY, HASH, or CODE), or specific
632 to the exact compilation phase (BEGIN, CHECK, INIT, or END). See
633 Attribute::Handlers.
634 • "B::Concise", by Stephen McCamant, is a new compiler backend for
636 walking the Perl syntax tree, printing concise info about ops. The
637 output is highly customisable. See B::Concise. [561+]
638 • The new bignum, bigint, and bigrat pragmas, by Tels, implement
640 transparent bignum support (using the Math::BigInt, Math::BigFloat,
641 and Math::BigRat backends).
642 • "Class::ISA", by Sean Burke, is a module for reporting the search
644 path for a class's ISA tree. See Class::ISA.
645 • "Cwd" now has a split personality: if possible, an XS extension is
647 used, (this will hopefully be faster, more secure, and more robust)
648 but if not possible, the familiar Perl implementation is used.
649 • "Devel::PPPort", originally by Kenneth Albanowski and now
651 maintained by Paul Marquess, has been added. It is primarily used
652 by "h2xs" to enhance portability of XS modules between different
653 versions of Perl. See Devel::PPPort.
654 • "Digest", frontend module for calculating digests (checksums), from
656 Gisle Aas, has been added. See Digest.
657 • "Digest::MD5" for calculating MD5 digests (checksums) as defined in
659 RFC 1321, from Gisle Aas, has been added. See Digest::MD5.
660 use Digest::MD5 'md5_hex';
662 $digest = md5_hex("Thirsty Camel");
664 print $digest, "\n"; # 01d19d9d2045e005c3f1b80e8b164de1
666 NOTE: the "MD5" backward compatibility module is deliberately not
668 included since its further use is discouraged.
669 See also PerlIO::via::QuotedPrint.
671 • "Encode", originally by Nick Ing-Simmons and now maintained by Dan
673 Kogai, provides a mechanism to translate between different
674 character encodings. Support for Unicode, ISO-8859-1, and ASCII
675 are compiled in to the module. Several other encodings (like the
676 rest of the ISO-8859, CP*/Win*, Mac, KOI8-R, three variants EBCDIC,
677 Chinese, Japanese, and Korean encodings) are included and can be
678 loaded at runtime. (For space considerations, the largest Chinese
679 encodings have been separated into their own CPAN module,
680 Encode::HanExtra, which Encode will use if available). See Encode.
681 Any encoding supported by Encode module is also available to the
683 ":encoding()" layer if PerlIO is used.
684 • "Hash::Util" is the interface to the new restricted hashes feature.
686 (Implemented by Jeffrey Friedl, Nick Ing-Simmons, and Michael
687 Schwern.) See Hash::Util.
688 • "I18N::Langinfo" can be used to query locale information. See
690 I18N::Langinfo.
691 • "I18N::LangTags", by Sean Burke, has functions for dealing with
693 RFC3066-style language tags. See I18N::LangTags.
694 • "ExtUtils::Constant", by Nicholas Clark, is a new tool for
696 extension writers for generating XS code to import C header
697 constants. See ExtUtils::Constant.
698 • "Filter::Simple", by Damian Conway, is an easy-to-use frontend to
700 Filter::Util::Call. See Filter::Simple.
701 # in MyFilter.pm:
703 package MyFilter;
705 use Filter::Simple sub {
707 while (my ($from, $to) = splice @_, 0, 2) {
708 s/$from/$to/g;
709 }
710 };
711 1;
713 # in user's code:
715 use MyFilter qr/red/ => 'green';
717 print "red\n"; # this code is filtered, will print "green\n"
719 print "bored\n"; # this code is filtered, will print "bogreen\n"
720 no MyFilter;
722 print "red\n"; # this code is not filtered, will print "red\n"
724 • "File::Temp", by Tim Jenness, allows one to create temporary files
726 and directories in an easy, portable, and secure way. See
727 File::Temp. [561+]
728 • "Filter::Util::Call", by Paul Marquess, provides you with the
730 framework to write source filters in Perl. For most uses, the
731 frontend Filter::Simple is to be preferred. See
732 Filter::Util::Call.
733 • "if", by Ilya Zakharevich, is a new pragma for conditional
735 inclusion of modules.
736 • libnet, by Graham Barr, is a collection of perl5 modules related to
738 network programming. See Net::FTP, Net::NNTP, Net::Ping (not part
739 of libnet, but related), Net::POP3, Net::SMTP, and Net::Time.
740 Perl installation leaves libnet unconfigured; use libnetcfg to
742 configure it.
743 • "List::Util", by Graham Barr, is a selection of general-utility
745 list subroutines, such as sum(), min(), first(), and shuffle().
746 See List::Util.
747 • "Locale::Constants", "Locale::Country", "Locale::Currency"
749 "Locale::Language", and Locale::Script, by Neil Bowers, have been
750 added. They provide the codes for various locale standards, such
751 as "fr" for France, "usd" for US Dollar, and "ja" for Japanese.
752 use Locale::Country;
754 $country = code2country('jp'); # $country gets 'Japan'
756 $code = country2code('Norway'); # $code gets 'no'
757 See Locale::Constants, Locale::Country, Locale::Currency, and
759 Locale::Language.
760 • "Locale::Maketext", by Sean Burke, is a localization framework.
762 See Locale::Maketext, and Locale::Maketext::TPJ13. The latter is
763 an article about software localization, originally published in The
764 Perl Journal #13, and republished here with kind permission.
765 • "Math::BigRat" for big rational numbers, to accompany Math::BigInt
767 and Math::BigFloat, from Tels. See Math::BigRat.
768 • "Memoize" can make your functions faster by trading space for time,
770 from Mark-Jason Dominus. See Memoize.
771 • "MIME::Base64", by Gisle Aas, allows you to encode data in base64,
773 as defined in RFC 2045 - MIME (Multipurpose Internet Mail
774 Extensions).
775 use MIME::Base64;
777 $encoded = encode_base64('Aladdin:open sesame');
779 $decoded = decode_base64($encoded);
780 print $encoded, "\n"; # "QWxhZGRpbjpvcGVuIHNlc2FtZQ=="
782 See MIME::Base64.
784 • "MIME::QuotedPrint", by Gisle Aas, allows you to encode data in
786 quoted-printable encoding, as defined in RFC 2045 - MIME
787 (Multipurpose Internet Mail Extensions).
788 use MIME::QuotedPrint;
790 $encoded = encode_qp("\xDE\xAD\xBE\xEF");
792 $decoded = decode_qp($encoded);
793 print $encoded, "\n"; # "=DE=AD=BE=EF\n"
795 print $decoded, "\n"; # "\xDE\xAD\xBE\xEF\n"
796 See also PerlIO::via::QuotedPrint.
798 • "NEXT", by Damian Conway, is a pseudo-class for method redispatch.
800 See NEXT.
801 • "open" is a new pragma for setting the default I/O layers for
803 open().
804 • "PerlIO::scalar", by Nick Ing-Simmons, provides the implementation
806 of IO to "in memory" Perl scalars as discussed above. It also
807 serves as an example of a loadable PerlIO layer. Other future
808 possibilities include PerlIO::Array and PerlIO::Code. See
809 PerlIO::scalar.
810 • "PerlIO::via", by Nick Ing-Simmons, acts as a PerlIO layer and
812 wraps PerlIO layer functionality provided by a class (typically
813 implemented in Perl code).
814 • "PerlIO::via::QuotedPrint", by Elizabeth Mattijsen, is an example
816 of a "PerlIO::via" class:
817 use PerlIO::via::QuotedPrint;
819 open($fh,">:via(QuotedPrint)",$path);
820 This will automatically convert everything output to $fh to Quoted-
822 Printable. See PerlIO::via and PerlIO::via::QuotedPrint.
823 • "Pod::ParseLink", by Russ Allbery, has been added, to parse L<>
825 links in pods as described in the new perlpodspec.
826 • "Pod::Text::Overstrike", by Joe Smith, has been added. It converts
828 POD data to formatted overstrike text. See Pod::Text::Overstrike.
829 [561+]
830 • "Scalar::Util" is a selection of general-utility scalar
832 subroutines, such as blessed(), reftype(), and tainted(). See
833 Scalar::Util.
834 • "sort" is a new pragma for controlling the behaviour of sort().
836 • "Storable" gives persistence to Perl data structures by allowing
838 the storage and retrieval of Perl data to and from files in a fast
839 and compact binary format. Because in effect Storable does
840 serialisation of Perl data structures, with it you can also clone
841 deep, hierarchical datastructures. Storable was originally created
842 by Raphael Manfredi, but it is now maintained by Abhijit Menon-Sen.
843 Storable has been enhanced to understand the two new hash features,
844 Unicode keys and restricted hashes. See Storable.
845 • "Switch", by Damian Conway, has been added. Just by saying
847 use Switch;
849 you have "switch" and "case" available in Perl.
851 use Switch;
853 switch ($val) {
855 case 1 { print "number 1" }
857 case "a" { print "string a" }
858 case [1..10,42] { print "number in list" }
859 case (@array) { print "number in list" }
860 case /\w+/ { print "pattern" }
861 case qr/\w+/ { print "pattern" }
862 case (%hash) { print "entry in hash" }
863 case (\%hash) { print "entry in hash" }
864 case (\&sub) { print "arg to subroutine" }
865 else { print "previous case not true" }
866 }
867 See Switch.
869 • "Test::More", by Michael Schwern, is yet another framework for
871 writing test scripts, more extensive than Test::Simple. See
872 Test::More.
873 • "Test::Simple", by Michael Schwern, has basic utilities for writing
875 tests. See Test::Simple.
876 • "Text::Balanced", by Damian Conway, has been added, for extracting
878 delimited text sequences from strings.
879 use Text::Balanced 'extract_delimited';
881 ($a, $b) = extract_delimited("'never say never', he never said", "'", '');
883 $a will be "'never say never'", $b will be ', he never said'.
885 In addition to extract_delimited(), there are also
887 extract_bracketed(), extract_quotelike(), extract_codeblock(),
888 extract_variable(), extract_tagged(), extract_multiple(),
889 gen_delimited_pat(), and gen_extract_tagged(). With these, you can
890 implement rather advanced parsing algorithms. See Text::Balanced.
891 • "threads", by Arthur Bergman, is an interface to interpreter
893 threads. Interpreter threads (ithreads) is the new thread model
894 introduced in Perl 5.6 but only available as an internal interface
895 for extension writers (and for Win32 Perl for "fork()" emulation).
896 See threads, threads::shared, and perlthrtut.
897 • "threads::shared", by Arthur Bergman, allows data sharing for
899 interpreter threads. See threads::shared.
900 • "Tie::File", by Mark-Jason Dominus, associates a Perl array with
902 the lines of a file. See Tie::File.
903 • "Tie::Memoize", by Ilya Zakharevich, provides on-demand loaded
905 hashes. See Tie::Memoize.
906 • "Tie::RefHash::Nestable", by Edward Avis, allows storing hash
908 references (unlike the standard Tie::RefHash) The module is
909 contained within Tie::RefHash. See Tie::RefHash.
910 • "Time::HiRes", by Douglas E. Wegscheid, provides high resolution
912 timing (ualarm, usleep, and gettimeofday). See Time::HiRes.
913 • "Unicode::UCD" offers a querying interface to the Unicode Character
915 Database. See Unicode::UCD.
916 • "Unicode::Collate", by SADAHIRO Tomoyuki, implements the UCA
918 (Unicode Collation Algorithm) for sorting Unicode strings. See
919 Unicode::Collate.
920 • "Unicode::Normalize", by SADAHIRO Tomoyuki, implements the various
922 Unicode normalization forms. See Unicode::Normalize.
923 • "XS::APItest", by Tim Jenness, is a test extension that exercises
925 XS APIs. Currently only "printf()" is tested: how to output
926 various basic data types from XS.
927 • "XS::Typemap", by Tim Jenness, is a test extension that exercises
929 XS typemaps. Nothing gets installed, but the code is worth
930 studying for extension writers.
931 Updated And Improved Modules and Pragmata
933 • The following independently supported modules have been updated to
934 the newest versions from CPAN: CGI, CPAN, DB_File, File::Spec,
935 File::Temp, Getopt::Long, Math::BigFloat, Math::BigInt, the
936 podlators bundle (Pod::Man, Pod::Text), Pod::LaTeX [561+],
937 Pod::Parser, Storable, Term::ANSIColor, Test, Text-Tabs+Wrap.
938 • attributes::reftype() now works on tied arguments.
940 • AutoLoader can now be disabled with "no AutoLoader;".
942 • B::Deparse has been significantly enhanced by Robin Houston. It
944 can now deparse almost all of the standard test suite (so that the
945 tests still succeed). There is a make target "test.deparse" for
946 trying this out.
947 • Carp now has better interface documentation, and the @CARP_NOT
949 interface has been added to get optional control over where errors
950 are reported independently of @ISA, by Ben Tilly.
951 • Class::Struct can now define the classes in compile time.
953 • Class::Struct now assigns the array/hash element if the accessor is
955 called with an array/hash element as the sole argument.
956 • The return value of Cwd::fastcwd() is now tainted.
958 • Data::Dumper now has an option to sort hashes.
960 • Data::Dumper now has an option to dump code references using
962 B::Deparse.
963 • DB_File now supports newer Berkeley DB versions, among other
965 improvements.
966 • Devel::Peek now has an interface for the Perl memory statistics
968 (this works only if you are using perl's malloc, and if you have
969 compiled with debugging).
970 • The English module can now be used without the infamous performance
972 hit by saying
973 use English '-no_match_vars';
975 (Assuming, of course, that you don't need the troublesome variables
977 "$`", $&, or "$'".) Also, introduced @LAST_MATCH_START and
978 @LAST_MATCH_END English aliases for "@-" and "@+".
979 • ExtUtils::MakeMaker has been significantly cleaned up and fixed.
981 The enhanced version has also been backported to earlier releases
982 of Perl and submitted to CPAN so that the earlier releases can
983 enjoy the fixes.
984 • The arguments of WriteMakefile() in Makefile.PL are now checked for
986 sanity much more carefully than before. This may cause new
987 warnings when modules are being installed. See ExtUtils::MakeMaker
988 for more details.
989 • ExtUtils::MakeMaker now uses File::Spec internally, which hopefully
991 leads to better portability.
992 • Fcntl, Socket, and Sys::Syslog have been rewritten by Nicholas
994 Clark to use the new-style constant dispatch section (see
995 ExtUtils::Constant). This means that they will be more robust and
996 hopefully faster.
997 • File::Find now chdir()s correctly when chasing symbolic links.
999 [561]
1000 • File::Find now has pre- and post-processing callbacks. It also
1002 correctly changes directories when chasing symbolic links.
1003 Callbacks (naughtily) exiting with "next;" instead of "return;" now
1004 work.
1005 • File::Find is now (again) reentrant. It also has been made more
1007 portable.
1008 • The warnings issued by File::Find now belong to their own category.
1010 You can enable/disable them with "use/no warnings 'File::Find';".
1011 • File::Glob::glob() has been renamed to File::Glob::bsd_glob()
1013 because the name clashes with the builtin glob(). The older name
1014 is still available for compatibility, but is deprecated. [561]
1015 • File::Glob now supports "GLOB_LIMIT" constant to limit the size of
1017 the returned list of filenames.
1018 • IPC::Open3 now allows the use of numeric file descriptors.
1020 • IO::Socket now has an atmark() method, which returns true if the
1022 socket is positioned at the out-of-band mark. The method is also
1023 exportable as a sockatmark() function.
1024 • IO::Socket::INET failed to open the specified port if the service
1026 name was not known. It now correctly uses the supplied port number
1027 as is. [561]
1028 • IO::Socket::INET has support for the ReusePort option (if your
1030 platform supports it). The Reuse option now has an alias,
1031 ReuseAddr. For clarity, you may want to prefer ReuseAddr.
1032 • IO::Socket::INET now supports a value of zero for "LocalPort"
1034 (usually meaning that the operating system will make one up.)
1035 • 'use lib' now works identically to @INC. Removing directories with
1037 'no lib' now works.
1038 • Math::BigFloat and Math::BigInt have undergone a full rewrite by
1040 Tels. They are now magnitudes faster, and they support various
1041 bignum libraries such as GMP and PARI as their backends.
1042 • Math::Complex handles inf, NaN etc., better.
1044 • Net::Ping has been considerably enhanced by Rob Brown: multihoming
1046 is now supported, Win32 functionality is better, there is now time
1047 measuring functionality (optionally high-resolution using
1048 Time::HiRes), and there is now "external" protocol which uses
1049 Net::Ping::External module which runs your external ping utility
1050 and parses the output. A version of Net::Ping::External is
1051 available in CPAN.
1052 Note that some of the Net::Ping tests are disabled when running
1054 under the Perl distribution since one cannot assume one or more of
1055 the following: enabled echo port at localhost, full Internet
1056 connectivity, or sympathetic firewalls. You can set the
1057 environment variable PERL_TEST_Net_Ping to "1" (one) before running
1058 the Perl test suite to enable all the Net::Ping tests.
1059 • POSIX::sigaction() is now much more flexible and robust. You can
1061 now install coderef handlers, 'DEFAULT', and 'IGNORE' handlers,
1062 installing new handlers was not atomic.
1063 • In Safe, %INC is now localised in a Safe compartment so that
1065 use/require work.
1066 • In SDBM_File on DOSish platforms, some keys went missing because of
1068 lack of support for files with "holes". A workaround for the
1069 problem has been added.
1070 • In Search::Dict one can now have a pre-processing hook for the
1072 lines being searched.
1073 • The Shell module now has an OO interface.
1075 • In Sys::Syslog there is now a failover mechanism that will go
1077 through alternative connection mechanisms until the message is
1078 successfully logged.
1079 • The Test module has been significantly enhanced.
1081 • Time::Local::timelocal() does not handle fractional seconds
1083 anymore. The rationale is that neither does localtime(), and
1084 timelocal() and localtime() are supposed to be inverses of each
1085 other.
1086 • The vars pragma now supports declaring fully qualified variables.
1088 (Something that "our()" does not and will not support.)
1089 • The "utf8::" name space (as in the pragma) provides various Perl-
1091 callable functions to provide low level access to Perl's internal
1092 Unicode representation. At the moment only length() has been
1093 implemented.
1094
1096 • Emacs perl mode (emacs/cperl-mode.el) has been updated to version
1097 4.31.
1098 • emacs/e2ctags.pl is now much faster.
1100 • "enc2xs" is a tool for people adding their own encodings to the
1102 Encode module.
1103 • "h2ph" now supports C trigraphs.
1105 • "h2xs" now produces a template README.
1107 • "h2xs" now uses "Devel::PPPort" for better portability between
1109 different versions of Perl.
1110 • "h2xs" uses the new ExtUtils::Constant module which will affect
1112 newly created extensions that define constants. Since the new code
1113 is more correct (if you have two constants where the first one is a
1114 prefix of the second one, the first constant never got defined),
1115 less lossy (it uses integers for integer constant, as opposed to
1116 the old code that used floating point numbers even for integer
1117 constants), and slightly faster, you might want to consider
1118 regenerating your extension code (the new scheme makes regenerating
1119 easy). h2xs now also supports C trigraphs.
1120 • "libnetcfg" has been added to configure libnet.
1122 • "perlbug" is now much more robust. It also sends the bug report to
1124 perl.org, not perl.com.
1125 • "perlcc" has been rewritten and its user interface (that is,
1127 command line) is much more like that of the Unix C compiler, cc.
1128 (The perlbc tools has been removed. Use "perlcc -B" instead.)
1129 Note that perlcc is still considered very experimental and
1130 unsupported. [561]
1131 • "perlivp" is a new Installation Verification Procedure utility for
1133 running any time after installing Perl.
1134 • "piconv" is an implementation of the character conversion utility
1136 "iconv", demonstrating the new Encode module.
1137 • "pod2html" now allows specifying a cache directory.
1139 • "pod2html" now produces XHTML 1.0.
1141 • "pod2html" now understands POD written using different line endings
1143 (PC-like CRLF versus Unix-like LF versus MacClassic-like CR).
1144 • "s2p" has been completely rewritten in Perl. (It is in fact a full
1146 implementation of sed in Perl: you can use the sed functionality by
1147 using the "psed" utility.)
1148 • "xsubpp" now understands POD documentation embedded in the *.xs
1150 files. [561]
1151 • "xsubpp" now supports the OUT keyword.
1153
1155 • perl56delta details the changes between the 5.005 release and the
1156 5.6.0 release.
1157 • perlclib documents the internal replacements for standard C library
1159 functions. (Interesting only for extension writers and Perl core
1160 hackers.) [561+]
1161 • perldebtut is a Perl debugging tutorial. [561+]
1163 • perlebcdic contains considerations for running Perl on EBCDIC
1165 platforms. [561+]
1166 • perlintro is a gentle introduction to Perl.
1168 • perliol documents the internals of PerlIO with layers.
1170 • perlmodstyle is a style guide for writing modules.
1172 • perlnewmod tells about writing and submitting a new module. [561+]
1174 • perlpacktut is a pack() tutorial.
1176 • perlpod has been rewritten to be clearer and to record the best
1178 practices gathered over the years.
1179 • perlpodspec is a more formal specification of the pod format,
1181 mainly of interest for writers of pod applications, not to people
1182 writing in pod.
1183 • perlretut is a regular expression tutorial. [561+]
1185 • perlrequick is a regular expressions quick-start guide. Yes, much
1187 quicker than perlretut. [561]
1188 • perltodo has been updated.
1190 • perltootc has been renamed as perltooc (to not to conflict with
1192 perltoot in filesystems restricted to "8.3" names).
1193 • perluniintro is an introduction to using Unicode in Perl.
1195 (perlunicode is more of a detailed reference and background
1196 information)
1197 • perlutil explains the command line utilities packaged with the Perl
1199 distribution. [561+]
1200 The following platform-specific documents are available before the
1202 installation as README.platform, and after the installation as
1203 perlplatform:
1204 perlaix perlamiga perlapollo perlbeos perlbs2000
1206 perlce perlcygwin perldgux perldos perlepoc perlfreebsd perlhpux
1207 perlhurd perlirix perlmachten perlmacos perlmint perlmpeix
1208 perlnetware perlos2 perlos390 perlplan9 perlqnx perlsolaris
1209 perltru64 perluts perlvmesa perlvms perlvos perlwin32
1210 These documents usually detail one or more of the following subjects:
1212 configuring, building, testing, installing, and sometimes also using
1213 Perl on the said platform.
1214 Eastern Asian Perl users are now welcomed in their own languages:
1216 README.jp (Japanese), README.ko (Korean), README.cn (simplified
1217 Chinese) and README.tw (traditional Chinese), which are written in
1218 normal pod but encoded in EUC-JP, EUC-KR, EUC-CN and Big5. These will
1219 get installed as
1220 perljp perlko perlcn perltw
1222 • The documentation for the POSIX-BC platform is called "BS2000", to
1224 avoid confusion with the Perl POSIX module.
1225 • The documentation for the WinCE platform is called perlce
1227 (README.ce in the source code kit), to avoid confusion with the
1228 perlwin32 documentation on 8.3-restricted filesystems.
1229
1231 • map() could get pathologically slow when the result list it
1232 generates is larger than the source list. The performance has been
1233 improved for common scenarios. [561]
1234 • sort() is also fully reentrant, in the sense that the sort function
1236 can itself call sort(). This did not work reliably in previous
1237 releases. [561]
1238 • sort() has been changed to use primarily mergesort internally as
1240 opposed to the earlier quicksort. For very small lists this may
1241 result in slightly slower sorting times, but in general the speedup
1242 should be at least 20%. Additional bonuses are that the worst case
1243 behaviour of sort() is now better (in computer science terms it now
1244 runs in time O(N log N), as opposed to quicksort's Theta(N**2)
1245 worst-case run time behaviour), and that sort() is now stable
1246 (meaning that elements with identical keys will stay ordered as
1247 they were before the sort). See the "sort" pragma for information.
1248 The story in more detail: suppose you want to serve yourself a
1250 little slice of Pi.
1251 @digits = ( 3,1,4,1,5,9 );
1253 A numerical sort of the digits will yield (1,1,3,4,5,9), as
1255 expected. Which 1 comes first is hard to know, since one 1 looks
1256 pretty much like any other. You can regard this as totally
1257 trivial, or somewhat profound. However, if you just want to sort
1258 the even digits ahead of the odd ones, then what will
1259 sort { ($a % 2) <=> ($b % 2) } @digits;
1261 yield? The only even digit, 4, will come first. But how about the
1263 odd numbers, which all compare equal? With the quicksort algorithm
1264 used to implement Perl 5.6 and earlier, the order of ties is left
1265 up to the sort. So, as you add more and more digits of Pi, the
1266 order in which the sorted even and odd digits appear will change.
1267 and, for sufficiently large slices of Pi, the quicksort algorithm
1268 in Perl 5.8 won't return the same results even if reinvoked with
1269 the same input. The justification for this rests with quicksort's
1270 worst case behavior. If you run
1271 sort { $a <=> $b } ( 1 .. $N , 1 .. $N );
1273 (something you might approximate if you wanted to merge two sorted
1275 arrays using sort), doubling $N doesn't just double the quicksort
1276 time, it quadruples it. Quicksort has a worst case run time that
1277 can grow like N**2, so-called quadratic behaviour, and it can
1278 happen on patterns that may well arise in normal use. You won't
1279 notice this for small arrays, but you will notice it with larger
1280 arrays, and you may not live long enough for the sort to complete
1281 on arrays of a million elements. So the 5.8 quicksort scrambles
1282 large arrays before sorting them, as a statistical defence against
1283 quadratic behaviour. But that means if you sort the same large
1284 array twice, ties may be broken in different ways.
1285 Because of the unpredictability of tie-breaking order, and the
1287 quadratic worst-case behaviour, quicksort was almost replaced
1288 completely with a stable mergesort. Stable means that ties are
1289 broken to preserve the original order of appearance in the input
1290 array. So
1291 sort { ($a % 2) <=> ($b % 2) } (3,1,4,1,5,9);
1293 will yield (4,3,1,1,5,9), guaranteed. The even and odd numbers
1295 appear in the output in the same order they appeared in the input.
1296 Mergesort has worst case O(N log N) behaviour, the best value
1297 attainable. And, ironically, this mergesort does particularly well
1298 where quicksort goes quadratic: mergesort sorts (1..$N, 1..$N) in
1299 O(N) time. But quicksort was rescued at the last moment because it
1300 is faster than mergesort on certain inputs and platforms. For
1301 example, if you really don't care about the order of even and odd
1302 digits, quicksort will run in O(N) time; it's very good at sorting
1303 many repetitions of a small number of distinct elements. The
1304 quicksort divide and conquer strategy works well on platforms with
1305 relatively small, very fast, caches. Eventually, the problem gets
1306 whittled down to one that fits in the cache, from which point it
1307 benefits from the increased memory speed.
1308 Quicksort was rescued by implementing a sort pragma to control
1310 aspects of the sort. The stable subpragma forces stable behaviour,
1311 regardless of algorithm. The _quicksort and _mergesort subpragmas
1312 are heavy-handed ways to select the underlying implementation. The
1313 leading "_" is a reminder that these subpragmas may not survive
1314 beyond 5.8. More appropriate mechanisms for selecting the
1315 implementation exist, but they wouldn't have arrived in time to
1316 save quicksort.
1317 • Hashes now use Bob Jenkins "One-at-a-Time" hashing key algorithm (
1319 http://burtleburtle.net/bob/hash/doobs.html ). This algorithm is
1320 reasonably fast while producing a much better spread of values than
1321 the old hashing algorithm (originally by Chris Torek, later tweaked
1322 by Ilya Zakharevich). Hash values output from the algorithm on a
1323 hash of all 3-char printable ASCII keys comes much closer to
1324 passing the DIEHARD random number generation tests. According to
1325 perlbench, this change has not affected the overall speed of Perl.
1326 • unshift() should now be noticeably faster.
1328
1330 Generic Improvements
1331 • INSTALL now explains how you can configure Perl to use 64-bit
1332 integers even on non-64-bit platforms.
1333 • Policy.sh policy change: if you are reusing a Policy.sh file (see
1335 INSTALL) and you use Configure -Dprefix=/foo/bar and in the old
1336 Policy $prefix eq $siteprefix and $prefix eq $vendorprefix, all of
1337 them will now be changed to the new prefix, /foo/bar. (Previously
1338 only $prefix changed.) If you do not like this new behaviour,
1339 specify prefix, siteprefix, and vendorprefix explicitly.
1340 • A new optional location for Perl libraries, otherlibdirs, is
1342 available. It can be used for example for vendor add-ons without
1343 disturbing Perl's own library directories.
1344 • In many platforms, the vendor-supplied 'cc' is too stripped-down to
1346 build Perl (basically, 'cc' doesn't do ANSI C). If this seems to
1347 be the case and 'cc' does not seem to be the GNU C compiler 'gcc',
1348 an automatic attempt is made to find and use 'gcc' instead.
1349 • gcc needs to closely track the operating system release to avoid
1351 build problems. If Configure finds that gcc was built for a
1352 different operating system release than is running, it now gives a
1353 clearly visible warning that there may be trouble ahead.
1354 • Since Perl 5.8 is not binary-compatible with previous releases of
1356 Perl, Configure no longer suggests including the 5.005 modules in
1357 @INC.
1358 • Configure "-S" can now run non-interactively. [561]
1360 • Configure support for pdp11-style memory models has been removed
1362 due to obsolescence. [561]
1363 • configure.gnu now works with options with whitespace in them.
1365 • installperl now outputs everything to STDERR.
1367 • Because PerlIO is now the default on most platforms, "-perlio"
1369 doesn't get appended to the $Config{archname} (also known as $^O)
1370 anymore. Instead, if you explicitly choose not to use perlio
1371 (Configure command line option -Uuseperlio), you will get "-stdio"
1372 appended.
1373 • Another change related to the architecture name is that "-64all"
1375 (-Duse64bitall, or "maximally 64-bit") is appended only if your
1376 pointers are 64 bits wide. (To be exact, the use64bitall is
1377 ignored.)
1378 • In AFS installations, one can configure the root of the AFS to be
1380 somewhere else than the default /afs by using the Configure
1381 parameter "-Dafsroot=/some/where/else".
1382 • APPLLIB_EXP, a lesser-known configuration-time definition, has been
1384 documented. It can be used to prepend site-specific directories to
1385 Perl's default search path (@INC); see INSTALL for information.
1386 • The version of Berkeley DB used when the Perl (and, presumably, the
1388 DB_File extension) was built is now available as
1389 @Config{qw(db_version_major db_version_minor db_version_patch)}
1390 from Perl and as "DB_VERSION_MAJOR_CFG DB_VERSION_MINOR_CFG
1391 DB_VERSION_PATCH_CFG" from C.
1392 • Building Berkeley DB3 for compatibility modes for DB, NDBM, and
1394 ODBM has been documented in INSTALL.
1395 • If you have CPAN access (either network or a local copy such as a
1397 CD-ROM) you can during specify extra modules to Configure to build
1398 and install with Perl using the -Dextras=... option. See INSTALL
1399 for more details.
1400 • In addition to config.over, a new override file, config.arch, is
1402 available. This file is supposed to be used by hints file writers
1403 for architecture-wide changes (as opposed to config.over which is
1404 for site-wide changes).
1405 • If your file system supports symbolic links, you can build Perl
1407 outside of the source directory by
1408 mkdir perl/build/directory
1410 cd perl/build/directory
1411 sh /path/to/perl/source/Configure -Dmksymlinks ...
1412 This will create in perl/build/directory a tree of symbolic links
1414 pointing to files in /path/to/perl/source. The original files are
1415 left unaffected. After Configure has finished, you can just say
1416 make all test
1418 and Perl will be built and tested, all in perl/build/directory.
1420 [561]
1421 • For Perl developers, several new make targets for profiling and
1423 debugging have been added; see perlhack.
1424 • Use of the gprof tool to profile Perl has been documented
1426 in perlhack. There is a make target called "perl.gprof"
1427 for generating a gprofiled Perl executable.
1428 • If you have GCC 3, there is a make target called
1430 "perl.gcov" for creating a gcoved Perl executable for
1431 coverage analysis. See perlhack.
1432 • If you are on IRIX or Tru64 platforms, new
1434 profiling/debugging options have been added; see perlhack
1435 for more information about pixie and Third Degree.
1436 • Guidelines of how to construct minimal Perl installations have been
1438 added to INSTALL.
1439 • The Thread extension is now not built at all under ithreads
1441 ("Configure -Duseithreads") because it wouldn't work anyway (the
1442 Thread extension requires being Configured with
1443 "-Duse5005threads").
1444 Note that the 5.005 threads are unsupported and deprecated: if you
1446 have code written for the old threads you should migrate it to the
1447 new ithreads model.
1448 • The Gconvert macro ($Config{d_Gconvert}) used by perl for
1450 stringifying floating-point numbers is now more picky about using
1451 sprintf %.*g rules for the conversion. Some platforms that used to
1452 use gcvt may now resort to the slower sprintf.
1453 • The obsolete method of making a special (e.g., debugging) flavor of
1455 perl by saying
1456 make LIBPERL=libperld.a
1458 has been removed. Use -DDEBUGGING instead.
1460 New Or Improved Platforms
1462 For the list of platforms known to support Perl, see "Supported
1463 Platforms" in perlport.
1464 • AIX dynamic loading should be now better supported.
1466 • AIX should now work better with gcc, threads, and 64-bitness. Also
1468 the long doubles support in AIX should be better now. See perlaix.
1469 • AtheOS ( http://www.atheos.cx/ ) is a new platform.
1471 • BeOS has been reclaimed.
1473 • The DG/UX platform now supports 5.005-style threads. See perldgux.
1475 • The DYNIX/ptx platform (also known as dynixptx) is supported at or
1477 near osvers 4.5.2.
1478 • EBCDIC platforms (z/OS (also known as OS/390), POSIX-BC, and
1480 VM/ESA) have been regained. Many test suite tests still fail and
1481 the co-existence of Unicode and EBCDIC isn't quite settled, but the
1482 situation is much better than with Perl 5.6. See perlos390,
1483 perlbs2000 (for POSIX-BC), and perlvmesa for more information.
1484 (Note: support for VM/ESA was removed in Perl v5.18.0. The relevant
1485 information was in README.vmesa)
1486 • Building perl with -Duseithreads or -Duse5005threads now works
1488 under HP-UX 10.20 (previously it only worked under 10.30 or later).
1489 You will need a thread library package installed. See README.hpux.
1490 [561]
1491 • Mac OS Classic is now supported in the mainstream source package
1493 (MacPerl has of course been available since perl 5.004 but now the
1494 source code bases of standard Perl and MacPerl have been
1495 synchronised) [561]
1496 • Mac OS X (or Darwin) should now be able to build Perl even on HFS+
1498 filesystems. (The case-insensitivity used to confuse the Perl
1499 build process.)
1500 • NCR MP-RAS is now supported. [561]
1502 • All the NetBSD specific patches (except for the installation
1504 specific ones) have been merged back to the main distribution.
1505 • NetWare from Novell is now supported. See perlnetware.
1507 • NonStop-UX is now supported. [561]
1509 • NEC SUPER-UX is now supported.
1511 • All the OpenBSD specific patches (except for the installation
1513 specific ones) have been merged back to the main distribution.
1514 • Perl has been tested with the GNU pth userlevel thread package (
1516 http://www.gnu.org/software/pth/pth.html ). All thread tests of
1517 Perl now work, but not without adding some yield()s to the tests,
1518 so while pth (and other userlevel thread implementations) can be
1519 considered to be "working" with Perl ithreads, keep in mind the
1520 possible non-preemptability of the underlying thread
1521 implementation.
1522 • Stratus VOS is now supported using Perl's native build method
1524 (Configure). This is the recommended method to build Perl on VOS.
1525 The older methods, which build miniperl, are still available. See
1526 perlvos. [561+]
1527 • The Amdahl UTS Unix mainframe platform is now supported. [561]
1529 • WinCE is now supported. See perlce.
1531 • z/OS (formerly known as OS/390, formerly known as MVS OE) now has
1533 support for dynamic loading. This is not selected by default,
1534 however, you must specify -Dusedl in the arguments of Configure.
1535 [561]
1536
1538 Numerous memory leaks and uninitialized memory accesses have been
1539 hunted down. Most importantly, anonymous subs used to leak quite a
1540 bit. [561]
1541 • The autouse pragma didn't work for Multi::Part::Function::Names.
1543 • caller() could cause core dumps in certain situations. Carp was
1545 sometimes affected by this problem. In particular, caller() now
1546 returns a subroutine name of "(unknown)" for subroutines that have
1547 been removed from the symbol table.
1548 • chop(@list) in list context returned the characters chopped in
1550 reverse order. This has been reversed to be in the right order.
1551 [561]
1552 • Configure no longer includes the DBM libraries (dbm, gdbm, db,
1554 ndbm) when building the Perl binary. The only exception to this is
1555 SunOS 4.x, which needs them. [561]
1556 • The behaviour of non-decimal but numeric string constants such as
1558 "0x23" was platform-dependent: in some platforms that was seen as
1559 35, in some as 0, in some as a floating point number (don't ask).
1560 This was caused by Perl's using the operating system libraries in a
1561 situation where the result of the string to number conversion is
1562 undefined: now Perl consistently handles such strings as zero in
1563 numeric contexts.
1564 • Several debugger fixes: exit code now reflects the script exit
1566 code, condition "0" now treated correctly, the "d" command now
1567 checks line number, $. no longer gets corrupted, and all debugger
1568 output now goes correctly to the socket if RemotePort is set. [561]
1569 • The debugger (perl5db.pl) has been modified to present a more
1571 consistent commands interface, via (CommandSet=580). perl5db.t was
1572 also added to test the changes, and as a placeholder for further
1573 tests.
1574 See perldebug.
1576 • The debugger has a new "dumpDepth" option to control the maximum
1578 depth to which nested structures are dumped. The "x" command has
1579 been extended so that "x N EXPR" dumps out the value of EXPR to a
1580 depth of at most N levels.
1581 • The debugger can now show lexical variables if you have the CPAN
1583 module PadWalker installed.
1584 • The order of DESTROYs has been made more predictable.
1586 • Perl 5.6.0 could emit spurious warnings about redefinition of
1588 dl_error() when statically building extensions into perl. This has
1589 been corrected. [561]
1590 • dprofpp -R didn't work.
1592 • *foo{FORMAT} now works.
1594 • Infinity is now recognized as a number.
1596 • UNIVERSAL::isa no longer caches methods incorrectly. (This broke
1598 the Tk extension with 5.6.0.) [561]
1599 • Lexicals I: lexicals outside an eval "" weren't resolved correctly
1601 inside a subroutine definition inside the eval "" if they were not
1602 already referenced in the top level of the eval""ed code.
1603 • Lexicals II: lexicals leaked at file scope into subroutines that
1605 were declared before the lexicals.
1606 • Lexical warnings now propagating correctly between scopes and into
1608 "eval "..."".
1609 • "use warnings qw(FATAL all)" did not work as intended. This has
1611 been corrected. [561]
1612 • warnings::enabled() now reports the state of $^W correctly if the
1614 caller isn't using lexical warnings. [561]
1615 • Line renumbering with eval and "#line" now works. [561]
1617 • Fixed numerous memory leaks, especially in eval "".
1619 • Localised tied variables no longer leak memory
1621 use Tie::Hash;
1623 tie my %tied_hash => 'Tie::StdHash';
1624 ...
1626 # Used to leak memory every time local() was called;
1628 # in a loop, this added up.
1629 local($tied_hash{Foo}) = 1;
1630 • Localised hash elements (and %ENV) are correctly unlocalised to not
1632 exist, if they didn't before they were localised.
1633 use Tie::Hash;
1635 tie my %tied_hash => 'Tie::StdHash';
1636 ...
1638 # Nothing has set the FOO element so far
1640 { local $tied_hash{FOO} = 'Bar' }
1642 # This used to print, but not now.
1644 print "exists!\n" if exists $tied_hash{FOO};
1645 As a side effect of this fix, tied hash interfaces must define the
1647 EXISTS and DELETE methods.
1648 • mkdir() now ignores trailing slashes in the directory name, as
1650 mandated by POSIX.
1651 • Some versions of glibc have a broken modfl(). This affects builds
1653 with "-Duselongdouble". This version of Perl detects this
1654 brokenness and has a workaround for it. The glibc release 2.2.2 is
1655 known to have fixed the modfl() bug.
1656 • Modulus of unsigned numbers now works (4063328477 % 65535 used to
1658 return 27406, instead of 27047). [561]
1659 • Some "not a number" warnings introduced in 5.6.0 eliminated to be
1661 more compatible with 5.005. Infinity is now recognised as a
1662 number. [561]
1663 • Numeric conversions did not recognize changes in the string value
1665 properly in certain circumstances. [561]
1666 • Attributes (such as :shared) didn't work with our().
1668 • our() variables will not cause bogus "Variable will not stay
1670 shared" warnings. [561]
1671 • "our" variables of the same name declared in two sibling blocks
1673 resulted in bogus warnings about "redeclaration" of the variables.
1674 The problem has been corrected. [561]
1675 • pack "Z" now correctly terminates the string with "\0".
1677 • Fix password routines which in some shadow password platforms (e.g.
1679 HP-UX) caused getpwent() to return every other entry.
1680 • The PERL5OPT environment variable (for passing command line
1682 arguments to Perl) didn't work for more than a single group of
1683 options. [561]
1684 • PERL5OPT with embedded spaces didn't work.
1686 • printf() no longer resets the numeric locale to "C".
1688 • "qw(a\\b)" now parses correctly as 'a\\b': that is, as three
1690 characters, not four. [561]
1691 • pos() did not return the correct value within s///ge in earlier
1693 versions. This is now handled correctly. [561]
1694 • Printing quads (64-bit integers) with printf/sprintf now works
1696 without the q L ll prefixes (assuming you are on a quad-capable
1697 platform).
1698 • Regular expressions on references and overloaded scalars now work.
1700 [561+]
1701 • Right-hand side magic (GMAGIC) could in many cases such as string
1703 concatenation be invoked too many times.
1704 • scalar() now forces scalar context even when used in void context.
1706 • SOCKS support is now much more robust.
1708 • sort() arguments are now compiled in the right wantarray context
1710 (they were accidentally using the context of the sort() itself).
1711 The comparison block is now run in scalar context, and the
1712 arguments to be sorted are always provided list context. [561]
1713 • Changed the POSIX character class "[[:space:]]" to include the
1715 (very rarely used) vertical tab character. Added a new POSIX-ish
1716 character class "[[:blank:]]" which stands for horizontal
1717 whitespace (currently, the space and the tab).
1718 • The tainting behaviour of sprintf() has been rationalized. It does
1720 not taint the result of floating point formats anymore, making the
1721 behaviour consistent with that of string interpolation. [561]
1722 • Some cases of inconsistent taint propagation (such as within hash
1724 values) have been fixed.
1725 • The RE engine found in Perl 5.6.0 accidentally pessimised certain
1727 kinds of simple pattern matches. These are now handled better.
1728 [561]
1729 • Regular expression debug output (whether through "use re 'debug'"
1731 or via "-Dr") now looks better. [561]
1732 • Multi-line matches like ""a\nxb\n" =~ /(?!\A)x/m" were flawed. The
1734 bug has been fixed. [561]
1735 • Use of $& could trigger a core dump under some situations. This is
1737 now avoided. [561]
1738 • The regular expression captured submatches ($1, $2, ...) are now
1740 more consistently unset if the match fails, instead of leaving
1741 false data lying around in them. [561]
1742 • readline() on files opened in "slurp" mode could return an extra ""
1744 (blank line) at the end in certain situations. This has been
1745 corrected. [561]
1746 • Autovivification of symbolic references of special variables
1748 described in perlvar (as in "${$num}") was accidentally disabled.
1749 This works again now. [561]
1750 • Sys::Syslog ignored the "LOG_AUTH" constant.
1752 • $AUTOLOAD, sort(), lock(), and spawning subprocesses in multiple
1754 threads simultaneously are now thread-safe.
1755 • Tie::Array's SPLICE method was broken.
1757 • Allow a read-only string on the left-hand side of a non-modifying
1759 tr///.
1760 • If "STDERR" is tied, warnings caused by "warn" and "die" now
1762 correctly pass to it.
1763 • Several Unicode fixes.
1765 • BOMs (byte order marks) at the beginning of Perl files
1767 (scripts, modules) should now be transparently skipped.
1768 UTF-16 and UCS-2 encoded Perl files should now be read
1769 correctly.
1770 • The character tables have been updated to Unicode 3.2.0.
1772 • Comparing with utf8 data does not magically upgrade
1774 non-utf8 data into utf8. (This was a problem for example
1775 if you were mixing data from I/O and Unicode data: your
1776 output might have got magically encoded as UTF-8.)
1777 • Generating illegal Unicode code points such as U+FFFE, or
1779 the UTF-16 surrogates, now also generates an optional
1780 warning.
1781 • "IsAlnum", "IsAlpha", and "IsWord" now match titlecase.
1783 • Concatenation with the "." operator or via variable
1785 interpolation, "eq", "substr", "reverse", "quotemeta", the
1786 "x" operator, substitution with "s///", single-quoted
1787 UTF-8, should now work.
1788 • The "tr///" operator now works. Note that the "tr///CU"
1790 functionality has been removed (but see pack('U0', ...)).
1791 • "eval "v200"" now works.
1793 • Perl 5.6.0 parsed m/\x{ab}/ incorrectly, leading to
1795 spurious warnings. This has been corrected. [561]
1796 • Zero entries were missing from the Unicode classes such as
1798 "IsDigit".
1799 • Large unsigned numbers (those above 2**31) could sometimes lose
1801 their unsignedness, causing bogus results in arithmetic operations.
1802 [561]
1803 • The Perl parser has been stress tested using both random input and
1805 Markov chain input and the few found crashes and lockups have been
1806 fixed.
1807 Platform Specific Changes and Fixes
1809 • BSDI 4.*
1810 Perl now works on post-4.0 BSD/OSes.
1812 • All BSDs
1814 Setting $0 now works (as much as possible; see perlvar for
1816 details).
1817 • Cygwin
1819 Numerous updates; currently synchronised with Cygwin 1.3.10.
1821 • Previously DYNIX/ptx had problems in its Configure probe for non-
1823 blocking I/O.
1824 • EPOC
1826 EPOC now better supported. See README.epoc. [561]
1828 • FreeBSD 3.*
1830 Perl now works on post-3.0 FreeBSDs.
1832 • HP-UX
1834 README.hpux updated; "Configure -Duse64bitall" now works; now uses
1836 HP-UX malloc instead of Perl malloc.
1837 • IRIX
1839 Numerous compilation flag and hint enhancements; accidental mixing
1841 of 32-bit and 64-bit libraries (a doomed attempt) made much harder.
1842 • Linux
1844 • Long doubles should now work (see INSTALL). [561]
1846 • Linux previously had problems related to sockaddrlen when
1848 using accept(), recvfrom() (in Perl: recv()),
1849 getpeername(), and getsockname().
1850 • Mac OS Classic
1852 Compilation of the standard Perl distribution in Mac OS Classic
1854 should now work if you have the Metrowerks development environment
1855 and the missing Mac-specific toolkit bits. Contact the macperl
1856 mailing list for details.
1857 • MPE/iX
1859 MPE/iX update after Perl 5.6.0. See README.mpeix. [561]
1861 • NetBSD/threads: try installing the GNU pth (should be in the
1863 packages collection, or http://www.gnu.org/software/pth/), and
1864 Configure with -Duseithreads.
1865 • NetBSD/sparc
1867 Perl now works on NetBSD/sparc.
1869 • OS/2
1871 Now works with usethreads (see INSTALL). [561]
1873 • Solaris
1875 64-bitness using the Sun Workshop compiler now works.
1877 • Stratus VOS
1879 The native build method requires at least VOS Release 14.5.0 and
1881 GNU C++/GNU Tools 2.0.1 or later. The Perl pack function now maps
1882 overflowed values to +infinity and underflowed values to -infinity.
1883 • Tru64 (aka Digital UNIX, aka DEC OSF/1)
1885 The operating system version letter now recorded in
1887 $Config{osvers}. Allow compiling with gcc (previously explicitly
1888 forbidden). Compiling with gcc still not recommended because buggy
1889 code results, even with gcc 2.95.2.
1890 • Unicos
1892 Fixed various alignment problems that lead into core dumps either
1894 during build or later; no longer dies on math errors at runtime;
1895 now using full quad integers (64 bits), previously was using only
1896 46 bit integers for speed.
1897 • VMS
1899 See "Socket Extension Dynamic in VMS" and "IEEE-format Floating
1901 Point Default on OpenVMS Alpha" for important changes not otherwise
1902 listed here.
1903 chdir() now works better despite a CRT bug; now works with
1905 MULTIPLICITY (see INSTALL); now works with Perl's malloc.
1906 The tainting of %ENV elements via "keys" or "values" was previously
1908 unimplemented. It now works as documented.
1909 The "waitpid" emulation has been improved. The worst bug (now
1911 fixed) was that a pid of -1 would cause a wildcard search of all
1912 processes on the system.
1913 POSIX-style signals are now emulated much better on VMS versions
1915 prior to 7.0.
1916 The "system" function and backticks operator have improved
1918 functionality and better error handling. [561]
1919 File access tests now use current process privileges rather than
1921 the user's default privileges, which could sometimes result in a
1922 mismatch between reported access and actual access. This
1923 improvement is only available on VMS v6.0 and later.
1924 There is a new "kill" implementation based on "sys$sigprc" that
1926 allows older VMS systems (pre-7.0) to use "kill" to send signals
1927 rather than simply force exit. This implementation also allows
1928 later systems to call "kill" from within a signal handler.
1929 Iterative logical name translations are now limited to 10
1931 iterations in imitation of SHOW LOGICAL and other OpenVMS
1932 facilities.
1933 • Windows
1935 • Signal handling now works better than it used to. It is
1937 now implemented using a Windows message loop, and is
1938 therefore less prone to random crashes.
1939 • fork() emulation is now more robust, but still continues to
1941 have a few esoteric bugs and caveats. See perlfork for
1942 details. [561+]
1943 • A failed (pseudo)fork now returns undef and sets errno to
1945 EAGAIN. [561]
1946 • The following modules now work on Windows:
1948 ExtUtils::Embed [561]
1950 IO::Pipe
1951 IO::Poll
1952 Net::Ping
1953 • IO::File::new_tmpfile() is no longer limited to 32767
1955 invocations per-process.
1956 • Better chdir() return value for a non-existent directory.
1958 • Compiling perl using the 64-bit Platform SDK tools is now
1960 supported.
1961 • The Win32::SetChildShowWindow() builtin can be used to
1963 control the visibility of windows created by child
1964 processes. See Win32 for details.
1965 • Non-blocking waits for child processes (or pseudo-
1967 processes) are supported via "waitpid($pid,
1968 &POSIX::WNOHANG)".
1969 • The behavior of system() with multiple arguments has been
1971 rationalized. Each unquoted argument will be automatically
1972 quoted to protect whitespace, and any existing whitespace
1973 in the arguments will be preserved. This improves the
1974 portability of system(@args) by avoiding the need for
1975 Windows "cmd" shell specific quoting in perl programs.
1976 Note that this means that some scripts that may have relied
1978 on earlier buggy behavior may no longer work correctly.
1979 For example, "system("nmake /nologo", @args)" will now
1980 attempt to run the file "nmake /nologo" and will fail when
1981 such a file isn't found. On the other hand, perl will now
1982 execute code such as "system("c:/Program
1983 Files/MyApp/foo.exe", @args)" correctly.
1984 • The perl header files no longer suppress common warnings
1986 from the Microsoft Visual C++ compiler. This means that
1987 additional warnings may now show up when compiling XS code.
1988 • Borland C++ v5.5 is now a supported compiler that can build
1990 Perl. However, the generated binaries continue to be
1991 incompatible with those generated by the other supported
1992 compilers (GCC and Visual C++). [561]
1993 • Duping socket handles with open(F, ">&MYSOCK") now works
1995 under Windows 9x. [561]
1996 • Current directory entries in %ENV are now correctly
1998 propagated to child processes. [561]
1999 • New %ENV entries now propagate to subprocesses. [561]
2001 • Win32::GetCwd() correctly returns C:\ instead of C: when at
2003 the drive root. Other bugs in chdir() and Cwd::cwd() have
2004 also been fixed. [561]
2005 • The makefiles now default to the features enabled in
2007 ActiveState ActivePerl (a popular Win32 binary
2008 distribution). [561]
2009 • HTML files will now be installed in c:\perl\html instead of
2011 c:\perl\lib\pod\html
2012 • REG_EXPAND_SZ keys are now allowed in registry settings
2014 used by perl. [561]
2015 • Can now send() from all threads, not just the first one.
2017 [561]
2018 • ExtUtils::MakeMaker now uses $ENV{LIB} to search for
2020 libraries. [561]
2021 • Less stack reserved per thread so that more threads can run
2023 concurrently. (Still 16M per thread.) [561]
2024 • "File::Spec->tmpdir()" now prefers C:/temp over /tmp (works
2026 better when perl is running as service).
2027 • Better UNC path handling under ithreads. [561]
2029 • wait(), waitpid(), and backticks now return the correct
2031 exit status under Windows 9x. [561]
2032 • A socket handle leak in accept() has been fixed. [561]
2034
2036 Please see perldiag for more details.
2037 • Ambiguous range in the transliteration operator (like a-z-9) now
2039 gives a warning.
2040 • chdir("") and chdir(undef) now give a deprecation warning because
2042 they cause a possible unintentional chdir to the home directory.
2043 Say chdir() if you really mean that.
2044 • Two new debugging options have been added: if you have compiled
2046 your Perl with debugging, you can use the -DT [561] and -DR options
2047 to trace tokenising and to add reference counts to displaying
2048 variables, respectively.
2049 • The lexical warnings category "deprecated" is no longer a sub-
2051 category of the "syntax" category. It is now a top-level category
2052 in its own right.
2053 • Unadorned dump() will now give a warning suggesting to use explicit
2055 CORE::dump() if that's what really is meant.
2056 • The "Unrecognized escape" warning has been extended to include
2058 "\8", "\9", and "\_". There is no need to escape any of the "\w"
2059 characters.
2060 • All regular expression compilation error messages are now hopefully
2062 easier to understand both because the error message now comes
2063 before the failed regex and because the point of failure is now
2064 clearly marked by a "<-- HERE" marker.
2065 • Various I/O (and socket) functions like binmode(), close(), and so
2067 forth now more consistently warn if they are used illogically
2068 either on a yet unopened or on an already closed filehandle (or
2069 socket).
2070 • Using lstat() on a filehandle now gives a warning. (It's a non-
2072 sensical thing to do.)
2073 • The "-M" and "-m" options now warn if you didn't supply the module
2075 name.
2076 • If you in "use" specify a required minimum version, modules
2078 matching the name and but not defining a $VERSION will cause a
2079 fatal failure.
2080 • Using negative offset for vec() in lvalue context is now a warnable
2082 offense.
2083 • Odd number of arguments to overload::constant now elicits a
2085 warning.
2086 • Odd number of elements in anonymous hash now elicits a warning.
2088 • The various "opened only for", "on closed", "never opened" warnings
2090 drop the "main::" prefix for filehandles in the "main" package, for
2091 example "STDIN" instead of "main::STDIN".
2092 • Subroutine prototypes are now checked more carefully, you may get
2094 warnings for example if you have used non-prototype characters.
2095 • If an attempt to use a (non-blessed) reference as an array index is
2097 made, a warning is given.
2098 • "push @a;" and "unshift @a;" (with no values to push or unshift)
2100 now give a warning. This may be a problem for generated and
2101 eval'ed code.
2102 • If you try to "pack" in perlfunc a number less than 0 or larger
2104 than 255 using the "C" format you will get an optional warning.
2105 Similarly for the "c" format and a number less than -128 or more
2106 than 127.
2107 • pack "P" format now demands an explicit size.
2109 • unpack "w" now warns of unterminated compressed integers.
2111 • Warnings relating to the use of PerlIO have been added.
2113 • Certain regex modifiers such as "(?o)" make sense only if applied
2115 to the entire regex. You will get an optional warning if you try
2116 to do otherwise.
2117 • Variable length lookbehind has not yet been implemented, trying to
2119 use it will tell that.
2120 • Using arrays or hashes as references (e.g. "%foo->{bar}" has been
2122 deprecated for a while. Now you will get an optional warning.
2123 • Warnings relating to the use of the new restricted hashes feature
2125 have been added.
2126 • Self-ties of arrays and hashes are not supported and fatal errors
2128 will happen even at an attempt to do so.
2129 • Using "sort" in scalar context now issues an optional warning.
2131 This didn't do anything useful, as the sort was not performed.
2132 • Using the /g modifier in split() is meaningless and will cause a
2134 warning.
2135 • Using splice() past the end of an array now causes a warning.
2137 • Malformed Unicode encodings (UTF-8 and UTF-16) cause a lot of
2139 warnings, as does trying to use UTF-16 surrogates (which are
2140 unimplemented).
2141 • Trying to use Unicode characters on an I/O stream without marking
2143 the stream's encoding (using open() or binmode()) will cause "Wide
2144 character" warnings.
2145 • Use of v-strings in use/require causes a (backward) portability
2147 warning.
2148 • Warnings relating to the use interpreter threads and their shared
2150 data have been added.
2151
2153 • PerlIO is now the default.
2154 • perlapi.pod (a companion to perlguts) now attempts to document the
2156 internal API.
2157 • You can now build a really minimal perl called microperl. Building
2159 microperl does not require even running Configure; "make -f
2160 Makefile.micro" should be enough. Beware: microperl makes many
2161 assumptions, some of which may be too bold; the resulting
2162 executable may crash or otherwise misbehave in wondrous ways. For
2163 careful hackers only.
2164 • Added rsignal(), whichsig(), do_join(), op_clear, op_null,
2166 ptr_table_clear(), ptr_table_free(), sv_setref_uv(), and several
2167 UTF-8 interfaces to the publicised API. For the full list of the
2168 available APIs see perlapi.
2169 • Made possible to propagate customised exceptions via croak()ing.
2171 • Now xsubs can have attributes just like subs. (Well, at least the
2173 built-in attributes.)
2174 • dTHR and djSP have been obsoleted; the former removed (because it's
2176 a no-op) and the latter replaced with dSP.
2177 • PERL_OBJECT has been completely removed.
2179 • The MAGIC constants (e.g. 'P') have been macrofied (e.g.
2181 "PERL_MAGIC_TIED") for better source code readability and
2182 maintainability.
2183 • The regex compiler now maintains a structure that identifies nodes
2185 in the compiled bytecode with the corresponding syntactic features
2186 of the original regex expression. The information is attached to
2187 the new "offsets" member of the "struct regexp". See perldebguts
2188 for more complete information.
2189 • The C code has been made much more "gcc -Wall" clean. Some warning
2191 messages still remain in some platforms, so if you are compiling
2192 with gcc you may see some warnings about dubious practices. The
2193 warnings are being worked on.
2194 • perly.c, sv.c, and sv.h have now been extensively commented.
2196 • Documentation on how to use the Perl source repository has been
2198 added to Porting/repository.pod.
2199 • There are now several profiling make targets.
2201
2203 (This change was already made in 5.7.0 but bears repeating here.)
2204 (5.7.0 came out before 5.6.1: the development branch 5.7 released
2205 earlier than the maintenance branch 5.6)
2206 A potential security vulnerability in the optional suidperl component
2208 of Perl was identified in August 2000. suidperl is neither built nor
2209 installed by default. As of November 2001 the only known vulnerable
2210 platform is Linux, most likely all Linux distributions. CERT and
2211 various vendors and distributors have been alerted about the
2212 vulnerability. See
2213 http://www.cpan.org/src/5.0/sperl-2000-08-05/sperl-2000-08-05.txt for
2214 more information.
2215 The problem was caused by Perl trying to report a suspected security
2217 exploit attempt using an external program, /bin/mail. On Linux
2218 platforms the /bin/mail program had an undocumented feature which when
2219 combined with suidperl gave access to a root shell, resulting in a
2220 serious compromise instead of reporting the exploit attempt. If you
2221 don't have /bin/mail, or if you have 'safe setuid scripts', or if
2222 suidperl is not installed, you are safe.
2223 The exploit attempt reporting feature has been completely removed from
2225 Perl 5.8.0 (and the maintenance release 5.6.1, and it was removed also
2226 from all the Perl 5.7 releases), so that particular vulnerability isn't
2227 there anymore. However, further security vulnerabilities are,
2228 unfortunately, always possible. The suidperl functionality is most
2229 probably going to be removed in Perl 5.10. In any case, suidperl
2230 should only be used by security experts who know exactly what they are
2231 doing and why they are using suidperl instead of some other solution
2232 such as sudo ( see http://www.courtesan.com/sudo/ ).
2233
2235 Several new tests have been added, especially for the lib and ext
2236 subsections. There are now about 69 000 individual tests (spread over
2237 about 700 test scripts), in the regression suite (5.6.1 has about 11
2238 700 tests, in 258 test scripts) The exact numbers depend on the
2239 platform and Perl configuration used. Many of the new tests are of
2240 course introduced by the new modules, but still in general Perl is now
2241 more thoroughly tested.
2242 Because of the large number of tests, running the regression suite will
2244 take considerably longer time than it used to: expect the suite to take
2245 up to 4-5 times longer to run than in perl 5.6. On a really fast
2246 machine you can hope to finish the suite in about 6-8 minutes
2247 (wallclock time).
2248 The tests are now reported in a different order than in earlier Perls.
2250 (This happens because the test scripts from under t/lib have been moved
2251 to be closer to the library/extension they are testing.)
2252
2254 The Compiler Suite Is Still Very Experimental
2255 The compiler suite is slowly getting better but it continues to be
2256 highly experimental. Use in production environments is discouraged.
2257 Localising Tied Arrays and Hashes Is Broken
2259 local %tied_array;
2260 doesn't work as one would expect: the old value is restored
2262 incorrectly. This will be changed in a future release, but we don't
2263 know yet what the new semantics will exactly be. In any case, the
2264 change will break existing code that relies on the current (ill-
2265 defined) semantics, so just avoid doing this in general.
2266 Building Extensions Can Fail Because Of Largefiles
2268 Some extensions like mod_perl are known to have issues with
2269 `largefiles', a change brought by Perl 5.6.0 in which file offsets
2270 default to 64 bits wide, where supported. Modules may fail to compile
2271 at all, or they may compile and work incorrectly. Currently, there is
2272 no good solution for the problem, but Configure now provides
2273 appropriate non-largefile ccflags, ldflags, libswanted, and libs in the
2274 %Config hash (e.g., $Config{ccflags_nolargefiles}) so the extensions
2275 that are having problems can try configuring themselves without the
2276 largefileness. This is admittedly not a clean solution, and the
2277 solution may not even work at all. One potential failure is whether
2278 one can (or, if one can, whether it's a good idea to) link together at
2279 all binaries with different ideas about file offsets; all this is
2280 platform-dependent.
2281 Modifying $_ Inside for(..)
2283 for (1..5) { $_++ }
2284 works without complaint. It shouldn't. (You should be able to modify
2286 only lvalue elements inside the loops.) You can see the correct
2287 behaviour by replacing the 1..5 with 1, 2, 3, 4, 5.
2288 mod_perl 1.26 Doesn't Build With Threaded Perl
2290 Use mod_perl 1.27 or higher.
2291 lib/ftmp-security tests warn 'system possibly insecure'
2293 Don't panic. Read the 'make test' section of INSTALL instead.
2294 libwww-perl (LWP) fails base/date #51
2296 Use libwww-perl 5.65 or later.
2297 PDL failing some tests
2299 Use PDL 2.3.4 or later.
2300 Perl_get_sv
2302 You may get errors like 'Undefined symbol "Perl_get_sv"' or "can't
2303 resolve symbol 'Perl_get_sv'", or the symbol may be "Perl_sv_2pv".
2304 This probably means that you are trying to use an older shared Perl
2305 library (or extensions linked with such) with Perl 5.8.0 executable.
2306 Perl used to have such a subroutine, but that is no more the case.
2307 Check your shared library path, and any shared Perl libraries in those
2308 directories.
2309 Sometimes this problem may also indicate a partial Perl 5.8.0
2311 installation, see "Mac OS X dyld undefined symbols" for an example and
2312 how to deal with it.
2313 Self-tying Problems
2315 Self-tying of arrays and hashes is broken in rather deep and hard-to-
2316 fix ways. As a stop-gap measure to avoid people from getting
2317 frustrated at the mysterious results (core dumps, most often), it is
2318 forbidden for now (you will get a fatal error even from an attempt).
2319 A change to self-tying of globs has caused them to be recursively
2321 referenced (see: "Two-Phased Garbage Collection" in perlobj). You will
2322 now need an explicit untie to destroy a self-tied glob. This behaviour
2323 may be fixed at a later date.
2324 Self-tying of scalars and IO thingies works.
2326 ext/threads/t/libc
2328 If this test fails, it indicates that your libc (C library) is not
2329 threadsafe. This particular test stress tests the localtime() call to
2330 find out whether it is threadsafe. See perlthrtut for more
2331 information.
2332 Failure of Thread (5.005-style) tests
2334 Note that support for 5.005-style threading is deprecated, experimental
2335 and practically unsupported. In 5.10, it is expected to be removed.
2336 You should migrate your code to ithreads.
2337 The following tests are known to fail due to fundamental problems in
2339 the 5.005 threading implementation. These are not new failures--Perl
2340 5.005_0x has the same bugs, but didn't have these tests.
2341 ../ext/B/t/xref.t 255 65280 14 12 85.71% 3-14
2343 ../ext/List/Util/t/first.t 255 65280 7 4 57.14% 2 5-7
2344 ../lib/English.t 2 512 54 2 3.70% 2-3
2345 ../lib/FileCache.t 5 1 20.00% 5
2346 ../lib/Filter/Simple/t/data.t 6 3 50.00% 1-3
2347 ../lib/Filter/Simple/t/filter_only. 9 3 33.33% 1-2 5
2348 ../lib/Math/BigInt/t/bare_mbf.t 1627 4 0.25% 8 11 1626-1627
2349 ../lib/Math/BigInt/t/bigfltpm.t 1629 4 0.25% 10 13 1628-
2350 1629
2351 ../lib/Math/BigInt/t/sub_mbf.t 1633 4 0.24% 8 11 1632-1633
2352 ../lib/Math/BigInt/t/with_sub.t 1628 4 0.25% 9 12 1627-1628
2353 ../lib/Tie/File/t/31_autodefer.t 255 65280 65 32 49.23% 34-65
2354 ../lib/autouse.t 10 1 10.00% 4
2355 op/flip.t 15 1 6.67% 15
2356 These failures are unlikely to get fixed as 5.005-style threads are
2358 considered fundamentally broken. (Basically what happens is that
2359 competing threads can corrupt shared global state, one good example
2360 being regular expression engine's state.)
2361 Timing problems
2363 The following tests may fail intermittently because of timing problems,
2364 for example if the system is heavily loaded.
2365 t/op/alarm.t
2367 ext/Time/HiRes/HiRes.t
2368 lib/Benchmark.t
2369 lib/Memoize/t/expmod_t.t
2370 lib/Memoize/t/speed.t
2371 In case of failure please try running them manually, for example
2373 ./perl -Ilib ext/Time/HiRes/HiRes.t
2375 Tied/Magical Array/Hash Elements Do Not Autovivify
2377 For normal arrays "$foo = \$bar[1]" will assign "undef" to $bar[1]
2378 (assuming that it didn't exist before), but for tied/magical arrays and
2379 hashes such autovivification does not happen because there is currently
2380 no way to catch the reference creation. The same problem affects
2381 slicing over non-existent indices/keys of a tied/magical array/hash.
2382 Unicode in package/class and subroutine names does not work
2384 One can have Unicode in identifier names, but not in package/class or
2385 subroutine names. While some limited functionality towards this does
2386 exist as of Perl 5.8.0, that is more accidental than designed; use of
2387 Unicode for the said purposes is unsupported.
2388 One reason of this unfinishedness is its (currently) inherent
2390 unportability: since both package names and subroutine names may need
2391 to be mapped to file and directory names, the Unicode capability of the
2392 filesystem becomes important-- and there unfortunately aren't portable
2393 answers.
2394
2396 AIX
2397 • If using the AIX native make command, instead of just "make" issue
2398 "make all". In some setups the former has been known to spuriously
2399 also try to run "make install". Alternatively, you may want to use
2400 GNU make.
2401 • In AIX 4.2, Perl extensions that use C++ functions that use statics
2403 may have problems in that the statics are not getting initialized.
2404 In newer AIX releases, this has been solved by linking Perl with
2405 the libC_r library, but unfortunately in AIX 4.2 the said library
2406 has an obscure bug where the various functions related to time
2407 (such as time() and gettimeofday()) return broken values, and
2408 therefore in AIX 4.2 Perl is not linked against libC_r.
2409 • vac 5.0.0.0 May Produce Buggy Code For Perl
2411 The AIX C compiler vac version 5.0.0.0 may produce buggy code,
2413 resulting in a few random tests failing when run as part of "make
2414 test", but when the failing tests are run by hand, they succeed.
2415 We suggest upgrading to at least vac version 5.0.1.0, that has been
2416 known to compile Perl correctly. "lslpp -L|grep vac.C" will tell
2417 you the vac version. See README.aix.
2418 • If building threaded Perl, you may get compilation warning from
2420 pp_sys.c:
2421 "pp_sys.c", line 4651.39: 1506-280 (W) Function argument assignment between types "unsigned char*" and "const void*" is not allowed.
2423 This is harmless; it is caused by the getnetbyaddr() and
2425 getnetbyaddr_r() having slightly different types for their first
2426 argument.
2427 Alpha systems with old gccs fail several tests
2429 If you see op/pack, op/pat, op/regexp, or ext/Storable tests failing in
2430 a Linux/alpha or *BSD/Alpha, it's probably time to upgrade your gcc.
2431 gccs prior to 2.95.3 are definitely not good enough, and gcc 3.1 may be
2432 even better. (RedHat Linux/alpha with gcc 3.1 reported no problems, as
2433 did Linux 2.4.18 with gcc 2.95.4.) (In Tru64, it is preferable to use
2434 the bundled C compiler.)
2435 AmigaOS
2437 Perl 5.8.0 doesn't build in AmigaOS. It broke at some point during the
2438 ithreads work and we could not find Amiga experts to unbreak the
2439 problems. Perl 5.6.1 still works for AmigaOS (as does the 5.7.2
2440 development release).
2441 BeOS
2443 The following tests fail on 5.8.0 Perl in BeOS Personal 5.03:
2444 t/op/lfs............................FAILED at test 17
2446 t/op/magic..........................FAILED at test 24
2447 ext/Fcntl/t/syslfs..................FAILED at test 17
2448 ext/File/Glob/t/basic...............FAILED at test 3
2449 ext/POSIX/t/sigaction...............FAILED at test 13
2450 ext/POSIX/t/waitpid.................FAILED at test 1
2451 (Note: more information was available in README.beos until support for
2453 BeOS was removed in Perl v5.18.0)
2454 Cygwin "unable to remap"
2456 For example when building the Tk extension for Cygwin, you may get an
2457 error message saying "unable to remap". This is known problem with
2458 Cygwin, and a workaround is detailed in here:
2459 http://sources.redhat.com/ml/cygwin/2001-12/msg00894.html
2460 Cygwin ndbm tests fail on FAT
2462 One can build but not install (or test the build of) the NDBM_File on
2463 FAT filesystems. Installation (or build) on NTFS works fine. If one
2464 attempts the test on a FAT install (or build) the following failures
2465 are expected:
2466 ../ext/NDBM_File/ndbm.t 13 3328 71 59 83.10% 1-2 4 16-71
2468 ../ext/ODBM_File/odbm.t 255 65280 ?? ?? % ??
2469 ../lib/AnyDBM_File.t 2 512 12 2 16.67% 1 4
2470 ../lib/Memoize/t/errors.t 0 139 11 5 45.45% 7-11
2471 ../lib/Memoize/t/tie_ndbm.t 13 3328 4 4 100.00% 1-4
2472 run/fresh_perl.t 97 1 1.03% 91
2473 NDBM_File fails and ODBM_File just coredumps.
2475 If you intend to run only on FAT (or if using AnyDBM_File on FAT), run
2477 Configure with the -Ui_ndbm and -Ui_dbm options to prevent NDBM_File
2478 and ODBM_File being built.
2479 DJGPP Failures
2481 t/op/stat............................FAILED at test 29
2482 lib/File/Find/t/find.................FAILED at test 1
2483 lib/File/Find/t/taint................FAILED at test 1
2484 lib/h2xs.............................FAILED at test 15
2485 lib/Pod/t/eol........................FAILED at test 1
2486 lib/Test/Harness/t/strap-analyze.....FAILED at test 8
2487 lib/Test/Harness/t/test-harness......FAILED at test 23
2488 lib/Test/Simple/t/exit...............FAILED at test 1
2489 The above failures are known as of 5.8.0 with native builds with long
2491 filenames, but there are a few more if running under dosemu because of
2492 limitations (and maybe bugs) of dosemu:
2493 t/comp/cpp...........................FAILED at test 3
2495 t/op/inccode.........................(crash)
2496 and a few lib/ExtUtils tests, and several hundred Encode/t/Aliases.t
2498 failures that work fine with long filenames. So you really might
2499 prefer native builds and long filenames.
2500 FreeBSD built with ithreads coredumps reading large directories
2502 This is a known bug in FreeBSD 4.5's readdir_r(), it has been fixed in
2503 FreeBSD 4.6 (see perlfreebsd (README.freebsd)).
2504 FreeBSD Failing locale Test 117 For ISO 8859-15 Locales
2506 The ISO 8859-15 locales may fail the locale test 117 in FreeBSD. This
2507 is caused by the characters \xFF (y with diaeresis) and \xBE (Y with
2508 diaeresis) not behaving correctly when being matched case-
2509 insensitively. Apparently this problem has been fixed in the latest
2510 FreeBSD releases. ( http://www.freebsd.org/cgi/query-pr.cgi?pr=34308 )
2511 IRIX fails ext/List/Util/t/shuffle.t or Digest::MD5
2513 IRIX with MIPSpro 7.3.1.2m or 7.3.1.3m compiler may fail the List::Util
2514 test ext/List/Util/t/shuffle.t by dumping core. This seems to be a
2515 compiler error since if compiled with gcc no core dump ensues, and no
2516 failures have been seen on the said test on any other platform.
2517 Similarly, building the Digest::MD5 extension has been known to fail
2519 with "*** Termination code 139 (bu21)".
2520 The cure is to drop optimization level (Configure -Doptimize=-O2).
2522 HP-UX lib/posix Subtest 9 Fails When LP64-Configured
2524 If perl is configured with -Duse64bitall, the successful result of the
2525 subtest 10 of lib/posix may arrive before the successful result of the
2526 subtest 9, which confuses the test harness so much that it thinks the
2527 subtest 9 failed.
2528 Linux with glibc 2.2.5 fails t/op/int subtest #6 with -Duse64bitint
2530 This is a known bug in the glibc 2.2.5 with long long integers. (
2531 http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=65612 )
2532 Linux With Sfio Fails op/misc Test 48
2534 No known fix.
2535 Mac OS X
2537 Please remember to set your environment variable LC_ALL to "C" (setenv
2538 LC_ALL C) before running "make test" to avoid a lot of warnings about
2539 the broken locales of Mac OS X.
2540 The following tests are known to fail in Mac OS X 10.1.5 because of
2542 buggy (old) implementations of Berkeley DB included in Mac OS X:
2543 Failed Test Stat Wstat Total Fail Failed List of Failed
2545 -------------------------------------------------------------------------
2546 ../ext/DB_File/t/db-btree.t 0 11 ?? ?? % ??
2547 ../ext/DB_File/t/db-recno.t 149 3 2.01% 61 63 65
2548 If you are building on a UFS partition, you will also probably see
2550 t/op/stat.t subtest #9 fail. This is caused by Darwin's UFS not
2551 supporting inode change time.
2552 Also the ext/POSIX/t/posix.t subtest #10 fails but it is skipped for
2554 now because the failure is Apple's fault, not Perl's (blocked signals
2555 are lost).
2556 If you Configure with ithreads, ext/threads/t/libc.t will fail. Again,
2558 this is not Perl's fault-- the libc of Mac OS X is not threadsafe (in
2559 this particular test, the localtime() call is found to be
2560 threadunsafe.)
2561 Mac OS X dyld undefined symbols
2563 If after installing Perl 5.8.0 you are getting warnings about missing
2564 symbols, for example
2565 dyld: perl Undefined symbols
2567 _perl_sv_2pv
2568 _perl_get_sv
2569 you probably have an old pre-Perl-5.8.0 installation (or parts of one)
2571 in /Library/Perl (the undefined symbols used to exist in pre-5.8.0
2572 Perls). It seems that for some reason "make install" doesn't always
2573 completely overwrite the files in /Library/Perl. You can move the old
2574 Perl shared library out of the way like this:
2575 cd /Library/Perl/darwin/CORE
2577 mv libperl.dylib libperlold.dylib
2578 and then reissue "make install". Note that the above of course is
2580 extremely disruptive for anything using the /usr/local/bin/perl. If
2581 that doesn't help, you may have to try removing all the .bundle files
2582 from beneath /Library/Perl, and again "make install"-ing.
2583 OS/2 Test Failures
2585 The following tests are known to fail on OS/2 (for clarity only the
2586 failures are shown, not the full error messages):
2587 ../lib/ExtUtils/t/Mkbootstrap.t 1 256 18 1 5.56% 8
2589 ../lib/ExtUtils/t/Packlist.t 1 256 34 1 2.94% 17
2590 ../lib/ExtUtils/t/basic.t 1 256 17 1 5.88% 14
2591 lib/os2_process.t 2 512 227 2 0.88% 174 209
2592 lib/os2_process_kid.t 227 2 0.88% 174 209
2593 lib/rx_cmprt.t 255 65280 18 3 16.67% 16-18
2594 op/sprintf tests 91, 129, and 130
2596 The op/sprintf tests 91, 129, and 130 are known to fail on some
2597 platforms. Examples include any platform using sfio, and
2598 Compaq/Tandem's NonStop-UX.
2599 Test 91 is known to fail on QNX6 (nto), because "sprintf '%e',0"
2601 incorrectly produces 0.000000e+0 instead of 0.000000e+00.
2602 For tests 129 and 130, the failing platforms do not comply with the
2604 ANSI C Standard: lines 19ff on page 134 of ANSI X3.159 1989, to be
2605 exact. (They produce something other than "1" and "-1" when formatting
2606 0.6 and -0.6 using the printf format "%.0f"; most often, they produce
2607 "0" and "-0".)
2608 SCO
2610 The socketpair tests are known to be unhappy in SCO 3.2v5.0.4:
2611 ext/Socket/socketpair.t...............FAILED tests 15-45
2613 Solaris 2.5
2615 In case you are still using Solaris 2.5 (aka SunOS 5.5), you may
2616 experience failures (the test core dumping) in lib/locale.t. The
2617 suggested cure is to upgrade your Solaris.
2618 Solaris x86 Fails Tests With -Duse64bitint
2620 The following tests are known to fail in Solaris x86 with Perl
2621 configured to use 64 bit integers:
2622 ext/Data/Dumper/t/dumper.............FAILED at test 268
2624 ext/Devel/Peek/Peek..................FAILED at test 7
2625 SUPER-UX (NEC SX)
2627 The following tests are known to fail on SUPER-UX:
2628 op/64bitint...........................FAILED tests 29-30, 32-33, 35-36
2630 op/arith..............................FAILED tests 128-130
2631 op/pack...............................FAILED tests 25-5625
2632 op/pow................................
2633 op/taint..............................# msgsnd failed
2634 ../ext/IO/lib/IO/t/io_poll............FAILED tests 3-4
2635 ../ext/IPC/SysV/ipcsysv...............FAILED tests 2, 5-6
2636 ../ext/IPC/SysV/t/msg.................FAILED tests 2, 4-6
2637 ../ext/Socket/socketpair..............FAILED tests 12
2638 ../lib/IPC/SysV.......................FAILED tests 2, 5-6
2639 ../lib/warnings.......................FAILED tests 115-116, 118-119
2640 The op/pack failure ("Cannot compress negative numbers at op/pack.t
2642 line 126") is serious but as of yet unsolved. It points at some
2643 problems with the signedness handling of the C compiler, as do the
2644 64bitint, arith, and pow failures. Most of the rest point at problems
2645 with SysV IPC.
2646 Term::ReadKey not working on Win32
2648 Use Term::ReadKey 2.20 or later.
2649 UNICOS/mk
2651 • During Configure, the test
2652 Guessing which symbols your C compiler and preprocessor define...
2654 will probably fail with error messages like
2656 CC-20 cc: ERROR File = try.c, Line = 3
2658 The identifier "bad" is undefined.
2659 bad switch yylook 79bad switch yylook 79bad switch yylook 79bad switch yylook 79#ifdef A29K
2661 ^
2662 CC-65 cc: ERROR File = try.c, Line = 3
2664 A semicolon is expected at this point.
2665 This is caused by a bug in the awk utility of UNICOS/mk. You can
2667 ignore the error, but it does cause a slight problem: you cannot
2668 fully benefit from the h2ph utility (see h2ph) that can be used to
2669 convert C headers to Perl libraries, mainly used to be able to
2670 access from Perl the constants defined using C preprocessor, cpp.
2671 Because of the above error, parts of the converted headers will be
2672 invisible. Luckily, these days the need for h2ph is rare.
2673 • If building Perl with interpreter threads (ithreads), the
2675 getgrent(), getgrnam(), and getgrgid() functions cannot return the
2676 list of the group members due to a bug in the multithreaded support
2677 of UNICOS/mk. What this means is that in list context the
2678 functions will return only three values, not four.
2679 UTS
2681 There are a few known test failures. (Note: the relevant information
2682 was available in README.uts until support for UTS was removed in Perl
2683 v5.18.0)
2684 VOS (Stratus)
2686 When Perl is built using the native build process on VOS Release 14.5.0
2687 and GNU C++/GNU Tools 2.0.1, all attempted tests either pass or result
2688 in TODO (ignored) failures.
2689 VMS
2691 There should be no reported test failures with a default configuration,
2692 though there are a number of tests marked TODO that point to areas
2693 needing further debugging and/or porting work.
2694 Win32
2696 In multi-CPU boxes, there are some problems with the I/O buffering:
2697 some output may appear twice.
2698 XML::Parser not working
2700 Use XML::Parser 2.31 or later.
2701 z/OS (OS/390)
2703 z/OS has rather many test failures but the situation is actually much
2704 better than it was in 5.6.0; it's just that so many new modules and
2705 tests have been added.
2706 Failed Test Stat Wstat Total Fail Failed List of Failed
2708 ---------------------------------------------------------------------------
2709 ../ext/Data/Dumper/t/dumper.t 357 8 2.24% 311 314 325 327
2710 331 333 337 339
2711 ../ext/IO/lib/IO/t/io_unix.t 5 4 80.00% 2-5
2712 ../ext/Storable/t/downgrade.t 12 3072 169 12 7.10% 14-15 46-47 78-79
2713 110-111 150 161
2714 ../lib/ExtUtils/t/Constant.t 121 30976 48 48 100.00% 1-48
2715 ../lib/ExtUtils/t/Embed.t 9 9 100.00% 1-9
2716 op/pat.t 922 7 0.76% 665 776 785 832-
2717 834 845
2718 op/sprintf.t 224 3 1.34% 98 100 136
2719 op/tr.t 97 5 5.15% 63 71-74
2720 uni/fold.t 780 6 0.77% 61 169 196 661
2721 710-711
2722 The failures in dumper.t and downgrade.t are problems in the tests,
2724 those in io_unix and sprintf are problems in the USS (UDP sockets and
2725 printf formats). The pat, tr, and fold failures are genuine Perl
2726 problems caused by EBCDIC (and in the pat and fold cases, combining
2727 that with Unicode). The Constant and Embed are probably problems in
2728 the tests (since they test Perl's ability to build extensions, and that
2729 seems to be working reasonably well.)
2730 Unicode Support on EBCDIC Still Spotty
2732 Though mostly working, Unicode support still has problem spots on
2733 EBCDIC platforms. One such known spot are the "\p{}" and "\P{}"
2734 regular expression constructs for code points less than 256: the "pP"
2735 are testing for Unicode code points, not knowing about EBCDIC.
2736 Seen In Perl 5.7 But Gone Now
2738 "Time::Piece" (previously known as "Time::Object") was removed because
2739 it was felt that it didn't have enough value in it to be a core module.
2740 It is still a useful module, though, and is available from the CPAN.
2741 Perl 5.8 unfortunately does not build anymore on AmigaOS; this broke
2743 accidentally at some point. Since there are not that many Amiga
2744 developers available, we could not get this fixed and tested in time
2745 for 5.8.0. Perl 5.6.1 still works for AmigaOS (as does the 5.7.2
2746 development release).
2747 The "PerlIO::Scalar" and "PerlIO::Via" (capitalised) were renamed as
2749 "PerlIO::scalar" and "PerlIO::via" (all lowercase) just before 5.8.0.
2750 The main rationale was to have all core PerlIO layers to have all
2751 lowercase names. The "plugins" are named as usual, for example
2752 "PerlIO::via::QuotedPrint".
2753 The "threads::shared::queue" and "threads::shared::semaphore" were
2755 renamed as "Thread::Queue" and "Thread::Semaphore" just before 5.8.0.
2756 The main rationale was to have thread modules to obey normal naming,
2757 "Thread::" (the "threads" and "threads::shared" themselves are more
2758 pragma-like, they affect compile-time, so they stay lowercase).
2759
2761 If you find what you think is a bug, you might check the articles
2762 recently posted to the comp.lang.perl.misc newsgroup and the perl bug
2763 database at http://bugs.perl.org/ . There may also be information at
2764 http://www.perl.com/ , the Perl Home Page.
2765 If you believe you have an unreported bug, please run the perlbug
2767 program included with your release. Be sure to trim your bug down to a
2768 tiny but sufficient test case. Your bug report, along with the output
2769 of "perl -V", will be sent off to perlbug@perl.org to be analysed by
2770 the Perl porting team.
2771
2773 The Changes file for exhaustive details on what changed.
2774 The INSTALL file for how to build Perl.
2776 The README file for general stuff.
2778 The Artistic and Copying files for copyright information.
2780
2782 Written by Jarkko Hietaniemi <jhi@iki.fi>.
2783perl v5.36.3 2023-11-30 PERL58DELTA(1)