3 perldelta - what's new for perl v5.6 (as of v5.005_62)
7 This is an unsupported alpha release, meant for intrepid Perl developers
8 only. The included sources may not even build correctly on some platforms.
9 Subscribing to perl5-porters is the best way to monitor and contribute
10 to the progress of development releases (see www.perl.org for info).
12 This document describes differences between the 5.005 release and this one.
14 =head1 Incompatible Changes
16 =head2 Perl Source Incompatibilities
18 Beware that any new warnings that have been added are B<not> considered
21 Since all new warnings must be explicitly requested via the C<-w>
22 switch or the C<warnings> pragma, it is ultimately the programmer's
23 responsibility to ensure that warnings are enabled judiciously.
27 =item Treatment of list slices of undef has changed
29 When taking a slice of a literal list (as opposed to a slice of
30 an array or hash), Perl used to return an empty list if the
31 result happened to be composed of all undef values.
33 The new behavior is to produce an empty list if (and only if)
34 the original list was empty. Consider the following example:
36 @a = (1,undef,undef,2)[2,1,2];
38 The old behavior would have resulted in @a having no elements.
39 The new behavior ensures it has three undefined elements.
41 Note in particular that the behavior of slices of the following
42 cases remains unchanged:
46 @a = (anything_returning_empty_list())[2,1,2];
52 =item Possibly changed pseudo-random number generator
54 In 5.005_0x and earlier, perl's rand() function used the C library
55 rand(3) function. As of 5.005_52, Configure tests for drand48(),
56 random(), and rand() (in that order) and picks the first one it finds.
57 Perl programs that depend on reproducing a specific set of pseudo-random
58 numbers will now likely produce different output.
60 =item Hashing function for hash keys has changed
62 Perl hashes are not order preserving. The apparently random order
63 encountered when iterating on the contents of a hash is determined
64 by the hashing algorithm used. To improve the distribution of lower
65 bits in the hashed value, the algorithm has changed slightly as of
66 5.005_52. When iterating over hashes, this may yield a random order
67 that is B<different> from that of previous versions.
69 =item C<undef> fails on read only values
71 Using the C<undef> operator on a readonly value (such as $1) has
72 the same effect as assigning C<undef> to the readonly value--it
75 =item Close-on-exec bit may be set on pipe() handles
77 On systems that support a close-on-exec flag on filehandles, the
78 flag will be set for any handles created by pipe(), if that is
79 warranted by the value of $^F that may be in effect. Earlier
80 versions neglected to set the flag for handles created with
81 pipe(). See L<perlfunc/pipe> and L<perlvar/$^F>.
83 =item Writing C<"$$1"> to mean C<"${$}1"> is unsupported
85 Perl 5.004 deprecated the interpretation of C<$$1> and
86 similar within interpolated strings to mean C<$$ . "1">,
89 In Perl 5.6 and later, C<"$$1"> always means C<"${$1}">.
91 =item values(%h) and C<\(%h)> operate on aliases to values, not copies
93 each(), values() and hashes in a list context return the actual
94 values in the hash, instead of copies (as they used to in earlier
95 versions). Typical idioms for using these constructs copy the
96 returned values, but this is can make a significant difference when
97 creating references to the returned values.
99 Keys in the hash are still returned as copies when iterating on
102 =item vec(EXPR,OFFSET,BITS) enforces powers-of-two BITS
104 vec() generates a run-time error if the BITS argument is not
105 a valid power-of-two integer.
107 =item Text of some diagnostic output has changed
109 Most references to internal Perl operations in diagnostics
110 have been changed to be more descriptive. This may be an
111 issue for programs that may incorrectly rely on the exact
112 text of diagnostics for proper functioning.
114 =item C<%@> has been removed
116 The undocumented special variable C<%@> that used to accumulate
117 "background" errors (such as those that happen in DESTROY())
118 has been removed, because it could potentially result in memory
123 =head2 C Source Incompatibilities
127 =item C<PERL_POLLUTE>
129 Release 5.005 grandfathered old global symbol names by providing preprocessor
130 macros for extension source compatibility. As of release 5.6, these
131 preprocessor definitions are not available by default. You need to explicitly
132 compile perl with C<-DPERL_POLLUTE> to get these definitions. For
133 extensions still using the old symbols, this option can be
134 specified via MakeMaker:
136 perl Makefile.PL POLLUTE=1
138 =item C<PERL_IMPLICIT_CONTEXT>
140 This new build option provides a set of macros for all API functions
141 such that an implicit interpreter/thread context argument is passed to
142 every API function. As a result of this, something like C<sv_setsv(foo,bar)>
143 amounts to a macro invocation that actually translates to something like
144 C<Perl_sv_setsv(my_perl,foo,bar)>. While this is generally expected
145 to not have any significant source compatibility issues, the difference
146 between a macro and a real function call will need to be considered.
148 This means that there B<is> a source compatibility issue as a result of
149 this if your extensions attempt to use pointers to any of the Perl API
152 Note that the above issue is not relevant to the default build of
153 Perl, whose interfaces continue to match those of prior versions
154 (but subject to the other options described here).
156 PERL_IMPLICIT_CONTEXT is automatically enabled whenever Perl is built
157 with one of -Dusethreads, -Dusemultiplicity, or both.
159 See L<perlguts/"The Perl API"> for detailed information on the
160 ramifications of building Perl using this option.
162 =item C<PERL_POLLUTE_MALLOC>
164 Enabling Perl's malloc in release 5.005 and earlier caused
165 the namespace of system versions of the malloc family of functions to
166 be usurped by the Perl versions, since by default they used the
169 Besides causing problems on platforms that do not allow these functions to
170 be cleanly replaced, this also meant that the system versions could not
171 be called in programs that used Perl's malloc. Previous versions of Perl
172 have allowed this behaviour to be suppressed with the HIDEMYMALLOC and
173 EMBEDMYMALLOC preprocessor definitions.
175 As of release 5.6, Perl's malloc family of functions have default names
176 distinct from the system versions. You need to explicitly compile perl with
177 C<-DPERL_POLLUTE_MALLOC> to get the older behaviour. HIDEMYMALLOC
178 and EMBEDMYMALLOC have no effect, since the behaviour they enabled is now
181 Note that these functions do B<not> constitute Perl's memory allocation API.
182 See L<perlguts/"Memory Allocation"> for further information about that.
184 =item C<PL_na> and C<dTHR> Issues
186 The C<PL_na> global is now thread local, so a C<dTHR> declaration is needed
187 in the scope in which the global appears. XSUBs should handle this automatically,
188 but if you have used C<PL_na> in support functions, you either need to
189 change the C<PL_na> to a local variable (which is recommended), or put in
194 =head2 Compatible C Source API Changes
198 =item C<PATCHLEVEL> is now C<PERL_VERSION>
200 The cpp macros C<PERL_REVISION>, C<PERL_VERSION>, and C<PERL_SUBVERSION>
201 are now available by default from perl.h, and reflect the base revision,
202 patchlevel, and subversion respectively. C<PERL_REVISION> had no
203 prior equivalent, while C<PERL_VERSION> and C<PERL_SUBVERSION> were
204 previously available as C<PATCHLEVEL> and C<SUBVERSION>.
206 The new names cause less pollution of the B<cpp> namespace and reflect what
207 the numbers have come to stand for in common practice. For compatibility,
208 the old names are still supported when F<patchlevel.h> is explicitly
209 included (as required before), so there is no source incompatibility
212 =item Support for C++ exceptions
214 change#3386, also needs perlguts documentation
215 [TODO - Chip Salzenberg <chip@perlsupport.com>]
219 =head2 Binary Incompatibilities
221 The default build of this release is binary compatible with the 5.005
222 release or its maintenance versions.
224 The usethreads or usemultiplicity builds are B<not> binary compatible
225 with the corresponding builds in 5.005.
227 =head1 Installation and Configuration Improvements
229 =head2 New Configure flags
231 The following new flags may be enabled on the Configure command line
232 by running Configure with C<-Dflag>.
240 =head2 -Dusethreads and -Duse64bits now more daring
242 The Configure options enabling the use of threads and the use of
243 64-bitness are now more daring in the sense that they no more have
244 an explicit list of operating systems of known threads/64-bit
245 capabilities. In other words: if your operating system has the
246 necessary APIs, you should be able just to go ahead and use them.
247 See also L<"64-bit support">.
251 Some platforms have "long doubles", floating point numbers of even
252 larger range than ordinary "doubles". To enable using ng doubles for
253 Perl's scalars, use -Duselongdouble.
257 You can enable both -Duse64bits and -Dlongdouble by -Dusemorebits.
258 See also L<"64-bit support">.
260 =head2 -Duselargefiles
262 Some platforms support large files, files larger than two gigabytes.
263 See L<"Large file support"> for more information.
265 =head2 installusrbinperl
267 You can use "Configure -Uinstallusrbinperl" which causes installperl
268 to skip installing perl also as /usr/bin/perl. This is useful if you
269 prefer not to modify /usr/bin for some reason or another but harmful
270 because many scripts assume to find Perl in /usr/bin/perl.
274 You can use "Configure -Dusesocks" which causes Perl to probe
275 for the SOCKS proxy protocol library, http://www.socks.nec.com/
279 You can "post-edit" the Configure variables using the Configure C<-A>
280 flag. The editing happens immediately after the platform specific
281 hints files have been processed but before the actual configuration
282 process starts. Run C<Configure -h> to find out the full C<-A> syntax.
284 =head2 New Installation Scheme
287 [TODO - Andy Dougherty <doughera@lafcol.lafayette.edu>]
291 =head2 Unicode and UTF-8 support
293 Perl can optionally use UTF-8 as its internal representation for character
294 strings. The C<utf8> pragma enables this support in the current lexical
295 scope. See L<utf8> for more information.
297 =head2 Lexically scoped warning categories
299 You can now control the granularity of warnings emitted by perl at a finer
300 level using the C<use warnings> pragma. See L<warnings> and L<perllexwarn>
303 =head2 Lvalue subroutines
305 WARNING: This is an experimental feature.
308 [TODO - Ilya Zakharevich <ilya@math.ohio-state.edu>,
309 Tuomas Lukka <lukka@fas.harvard.edu>)]
311 =head2 "our" declarations
313 An "our" declaration introduces a value that can be best understood
314 as a lexically scoped symbolic alias to a global variable in the
315 current package. This is mostly useful as an alternative to the
316 C<vars> pragma, but also provides the opportunity to introduce
317 typing and other attributes for such variables. See L<perlfunc/our>.
319 =head2 Weak references
321 WARNING: This is an experimental feature.
323 change#3385, also need perlguts documentation
325 [TODO - Tuomas Lukka <lukka@fas.harvard.edu>]
327 =head2 Binary numbers supported
329 Binary numbers are now supported as literals, in s?printf formats, and
333 printf "The answer is: %b\n", oct("0b101010");
335 =head2 Some arrows may be omitted in calls through references
337 Perl now allows the arrow to be omitted in many constructs
338 involving subroutine calls through references. For example,
339 C<$foo[10]->('foo')> may now be written C<$foo[10]('foo')>.
340 This is rather similar to how the arrow may be omitted from
341 C<$foo[10]->{'foo'}>. Note however, that the arrow is still
342 required for C<foo(10)->('bar')>.
344 =head2 syswrite() ease-of-use
346 The length argument of C<syswrite()> has become optional.
348 =head2 Filehandles can be autovivified
350 The construct C<open(my $fh, ...)> can be used to create filehandles
351 more easily. The filehandle will be automatically closed at the end
352 of the scope of $fh, provided there are no other references to it. This
353 largely eliminates the need for typeglobs when opening filehandles
354 that must be passed around, as in the following example:
358 or die "Can't open '@_': $!";
363 my $f = myopen("</etc/motd");
365 # $f implicitly closed here
368 [TODO - this idiom needs more pod penetration]
370 =head2 64-bit support
372 All platforms that have 64-bit integers either (a) natively as longs
373 or ints (b) via special compiler flags (c) using long long are able to
374 use "quads" (64-integers) as follows:
380 constants (decimal, hexadecimal, octal, binary) in the code
384 arguments to oct() and hex()
388 arguments to print(), printf() and sprintf() (flag prefixes ll, L, q)
396 pack() and unpack() "q" and "Q" formats
400 in basic arithmetics: + - * / %
404 vec() (but see the below note about bit arithmetics)
408 Note that unless you have the case (a) you will have to configure
409 and compile Perl using the -Duse64bits Configure flag.
411 Unfortunately bit arithmetics (&, |, ^, ~, <<, >>) for numbers are not
412 64-bit clean, they are explictly forced to be 32-bit. Bit arithmetics
413 for bit vectors (created by vec()) are not limited in their width.
415 Last but not least: note that due to Perl's habit of always using
416 floating point numbers the quads are still not true integers.
417 When quads overflow their limits (0...18_446_744_073_709_551_615 unsigned,
418 -9_223_372_036_854_775_808...9_223_372_036_854_775_807 signed), they
419 are silently promoted to floating point numbers, after which they will
420 start losing precision (their lower digits).
422 =head2 Large file support
424 If you have filesystems that support "large files" (files larger than
425 2 gigabytes), you may now also be able to create and access them from
426 Perl. You have to use Configure -Duselargefiles. Turning on the
427 large file support turns on also the 64-bit support, for obvious reasons.
429 Note that in addition to requiring a proper file system to do large
430 files you may also need to adjust your per-process (or your
431 per-system, or per-process-group, or per-user-group) maximum filesize
432 limits before running Perl scripts that try to handle large files,
433 especially if you intend to write such files.
435 Finally, in addition to your process/process group maximum filesize
436 limits, you may have quota limits on your filesystems that stop you
437 (your user id or your user group id) from using large files.
439 Adjusting your process/user/group/file system/operating system limits
440 is outside the scope of Perl core language. For process limits, you
441 may try increasing the limits using your shell's limits/limit/ulimit
442 command before running Perl. The BSD::Resource extension (not
443 included with the standard Perl distribution) may also be of use, it
444 offers the getrlimit/setrlimit interface that can be used to adjust
445 process resource usage limits, including the maximum filesize limit.
449 In some systems you may be able to use long doubles to enhance the
450 range of precision of your double precision floating point numbers
451 (that is, Perl's numbers). Use Configure -Duselongdouble to enable
452 this support (if it is available).
456 You can Configure -Dusemorebits to turn on both the 64-bit support
457 and the long double support.
459 =head2 Better syntax checks on parenthesized unary operators
463 print defined(&foo,&bar,&baz);
464 print uc("foo","bar","baz");
467 used to be accidentally allowed in earlier versions, and produced
468 unpredictable behaviour. Some produced ancillary warnings
469 when used in this way; others silently did the wrong thing.
471 The parenthesized forms of most unary operators that expect a single
472 argument now ensure that they are not called with more than one
473 argument, making the cases shown above syntax errors. The usual
476 print defined &foo, &bar, &baz;
477 print uc "foo", "bar", "baz";
480 remains unchanged. See L<perlop>.
482 =head2 POSIX character class syntax [: :] supported
484 For example to match alphabetic characters use /[[:alpha:]]/.
485 See L<perlre> for details.
487 =head2 Improved C<qw//> operator
489 The C<qw//> operator is now evaluated at compile time into a true list
490 instead of being replaced with a run time call to C<split()>. This
491 removes the confusing misbehaviour of C<qw//> in scalar context, which
492 had inherited that behaviour from split().
496 $foo = ($bar) = qw(a b c); print "$foo|$bar\n";
498 now correctly prints "3|a", instead of "2|a".
500 =head2 pack() format 'Z' supported
502 The new format type 'Z' is useful for packing and unpacking null-terminated
503 strings. See L<perlfunc/"pack">.
505 =head2 pack() format modifier '!' supported
507 The new format type modifier '!' is useful for packing and unpacking
508 native shorts, ints, and longs. See L<perlfunc/"pack">.
510 =head2 pack() and unpack() support counted strings
512 The template character '/' can be used to specify a counted string
513 type to be packed or unpacked. See L<perlfunc/"pack">.
515 =head2 Comments in pack() templates
517 The '#' character in a template introduces a comment up to
518 end of the line. This facilitates documentation of pack()
521 =head2 $^X variables may now have names longer than one character
523 Formerly, $^X was synonymous with ${"\cX"}, but $^XY was a syntax
524 error. Now variable names that begin with a control character may be
525 arbitrarily long. However, for compatibility reasons, these variables
526 I<must> be written with explicit braces, as C<${^XY}> for example.
527 C<${^XYZ}> is synonymous with ${"\cXYZ"}. Variable names with more
528 than one control character, such as C<${^XY^Z}>, are illegal.
530 The old syntax has not changed. As before, `^X' may be either a
531 literal control-X character or the two-character sequence `caret' plus
532 `X'. When braces are omitted, the variable name stops after the
533 control character. Thus C<"$^XYZ"> continues to be synonymous with
534 C<$^X . "YZ"> as before.
536 As before, lexical variables may not have names beginning with control
537 characters. As before, variables whose names begin with a control
538 character are always forced to be in package `main'. All such variables
539 are reserved for future extensions, except those that begin with
540 C<^_>, which may be used by user programs and are guaranteed not to
541 acquire special meaning in any future version of Perl.
543 =head2 C<use attrs> implicit in subroutine attributes
545 Formerly, if you wanted to mark a subroutine as being a method call or
546 as requiring an automatic lock() when it is entered, you had to declare
547 that with a C<use attrs> pragma in the body of the subroutine.
548 That can now be accomplished with a declaration syntax, like this:
550 sub mymethod : locked, method ;
552 sub mymethod : locked, method {
556 F<AutoSplit.pm> and F<SelfLoader.pm> have been updated to keep the attributes
557 with the stubs they provide. See L<attributes>.
559 =head2 Regular expression improvements
561 change#2827,2373,2372,2365,1813,1800,4112,4158,4215,4301
562 [TODO - Ilya Zakharevich <ilya@math.ohio-state.edu>]
564 =head2 Overloading improvements
567 [TODO - Ilya Zakharevich <ilya@math.ohio-state.edu>]
569 =head2 open() with more than two arguments
571 [TODO - Ilya Zakharevich <ilya@math.ohio-state.edu>]
573 =head2 Support for interpolating named characters
576 [TODO - Ilya Zakharevich <ilya@math.ohio-state.edu>]
578 =head2 Experimental support for user-hooks in @INC
580 [TODO - Ken Fox <kfox@ford.com>]
582 =head2 C<require> and C<do> may be overridden
584 C<require> and C<do 'file'> operations may be overridden locally
585 by importing subroutines of the same name into the current package
586 (or globally by importing them into the CORE::GLOBAL:: namespace).
587 Overriding C<require> will also affect C<use>, provided the override
588 is visible at compile-time.
589 See L<perlsub/"Overriding Built-in Functions">.
591 =head2 New variable $^C reflects C<-c> switch
593 C<$^C> has a boolean value that reflects whether perl is being run
594 in compile-only mode (i.e. via the C<-c> switch). Since
595 BEGIN blocks are executed under such conditions, this variable
596 enables perl code to determine whether actions that make sense
597 only during normal running are warranted. See L<perlvar>.
599 =head2 Optional Y2K warnings
601 If Perl is built with the cpp macro C<PERL_Y2KWARN> defined,
602 it emits optional warnings when concatenating the number 19
605 This behavior must be specifically enabled when running Configure.
606 See L<INSTALL> and L<README.Y2K>.
608 =head1 Significant bug fixes
610 =head2 E<lt>HANDLEE<gt> on empty files
612 With C<$/> set to C<undef>, slurping an empty file returns a string of
613 zero length (instead of C<undef>, as it used to) the first time the
614 HANDLE is read. Further reads yield C<undef>.
616 This means that the following will append "foo" to an empty file (it used
619 perl -0777 -pi -e 's/^/foo/' empty_file
623 perl -pi -e 's/^/foo/' empty_file
625 is unchanged (it continues to leave the file empty).
627 =head2 C<eval '...'> improvements
629 Line numbers (as reflected by caller() and most diagnostics) within
630 C<eval '...'> were often incorrect when here documents were involved.
631 This has been corrected.
633 Lexical lookups for variables appearing in C<eval '...'> within
634 functions that were themselves called within an C<eval '...'> were
635 searching the wrong place for lexicals. The lexical search now
636 correctly ends at the subroutine's block boundary.
638 Parsing of here documents used to be flawed when they appeared as
639 the replacement expression in C<eval 's/.../.../e'>. This has
642 =head2 All compilation errors are true errors
644 Some "errors" encountered at compile time were by neccessity
645 generated as warnings followed by eventual termination of the
646 program. This enabled more such errors to be reported in a
647 single run, rather than causing a hard stop at the first error
648 that was encountered.
650 The mechanism for reporting such errors has been reimplemented
651 to queue compile-time errors and report them at the end of the
652 compilation as true errors rather than as warnings. This fixes
653 cases where error messages leaked through in the form of warnings
654 when code was compiled at run time using C<eval STRING>, and
655 also allows such errors to be reliably trapped using __DIE__ hooks.
657 =head2 Automatic flushing of output buffers
659 fork(), exec(), system(), qx//, and pipe open()s now flush buffers
660 of all files opened for output when the operation
661 was attempted. This mostly eliminates confusing
662 buffering mishaps suffered by users unaware of how Perl internally
665 =head2 Better diagnostics on meaningless filehandle operations
667 Constructs such as C<open(E<lt>FHE<gt>)> and C<close(E<lt>FHE<gt>)>
668 are compile time errors. Attempting to read from filehandles that
669 were opened only for writing will now produce warnings (just as
670 writing to read-only filehandles does).
672 =head2 Where possible, buffered data discarded from duped input filehandle
674 C<open(NEW, "E<lt>&OLD")> now attempts to discard any data that
675 was previously read and buffered in C<OLD> before duping the handle.
676 On platforms where doing this is allowed, the next read operation
677 on C<NEW> will return the same data as the corresponding operation
678 on C<OLD>. Formerly, it would have returned the data from the start
679 of the following disk block instead.
681 =head2 system(), backticks and pipe open now reflect exec() failure
683 On Unix and similar platforms, system(), qx() and open(FOO, "cmd |")
684 etc., are implemented via fork() and exec(). When the underlying
685 exec() fails, earlier versions did not report the error properly,
686 since the exec() happened to be in a different process.
688 The child process now communicates with the parent about the
689 error in launching the external command, which allow these
690 constructs to return with their usual error value and set $!.
692 =head2 Implicitly closed filehandles are safer
694 Sometimes implicitly closed filehandles (as when they are localized,
695 and Perl automatically closes them on exiting the scope) could
696 inadvertently set $? or $!. This has been corrected.
698 =head2 C<(\$)> prototype and C<$foo{a}>
700 An scalar reference prototype now correctly allows a hash or
701 array element in that slot.
703 =head2 Pseudo-hashes work better
705 Dereferencing some types of reference values in a pseudo-hash,
706 such as C<$ph->{foo}[1]>, was accidentally disallowed. This has
709 When applied to a pseudo-hash element, exists() now reports whether
710 the specified value exists, not merely if the key is valid.
712 =head2 C<goto &sub> and AUTOLOAD
714 The C<goto &sub> construct works correctly when C<&sub> happens
717 =head2 C<-bareword> allowed under C<use integer>
719 The autoquoting of barewords preceded by C<-> did not work
720 in prior versions when the C<integer> pragma was enabled.
723 =head2 Boolean assignment operators are legal lvalues
725 Constructs such as C<($a ||= 2) += 1> are now allowed.
727 =head2 C<sort $coderef @foo> allowed
729 sort() did not accept a subroutine reference as the comparison
730 function in earlier versions. This is now permitted.
732 =head2 Failures in DESTROY()
734 When code in a destructor threw an exception, it went unnoticed
735 in earlier versions of Perl, unless someone happened to be
736 looking in $@ just after the point the destructor happened to
737 run. Such failures are now visible as warnings when warnings are
740 =head2 Locale bugs fixed
742 printf() and sprintf() previously did reset the numeric locale
743 back to the default "C" locale. This has been fixed.
745 Numbers formatted according to the local numeric locale
746 (such as using a decimal comma instead of a decimal dot) caused
747 "isn't numeric" warnings, even while the operations accessing
748 those numbers produced correct results. The warnings are gone.
752 The C<eval 'return sub {...}'> construct could sometimes leak
753 memory. This has been fixed.
755 Operations that aren't filehandle constructors used to leak memory
756 when used on invalid filehandles. This has been fixed.
758 Constructs that modified C<@_> could fail to deallocate values
759 in C<@_> and thus leak memory. This has been corrected.
761 =head2 Spurious subroutine stubs after failed subroutine calls
763 Perl could sometimes create empty subroutine stubs when a
764 subroutine was not found in the package. Such cases stopped
765 later method lookups from progressing into base packages.
766 This has been corrected.
768 =head2 Consistent numeric conversions
771 [TODO - Ilya Zakharevich <ilya@math.ohio-state.edu>]
773 =head2 Taint failures under C<-U>
775 When running in unsafe mode, taint violations could sometimes
776 cause silent failures. This has been fixed.
778 =head2 END blocks and the C<-c> switch
780 Prior versions used to run BEGIN B<and> END blocks when Perl was
781 run in compile-only mode. Since this is typically not the expected
782 behavior, END blocks are not executed anymore when the C<-c> switch
785 Note that something resembling the previous behavior can still be
786 obtained by putting C<BEGIN { $^C = 0; exit; } at the very end of
787 the top level source file.
789 =head2 Potential to leak DATA filehandles
791 Using the C<__DATA__> token creates an implicit filehandle to
792 the file that contains the token. It is the program's
793 responsibility to close it when it is done reading from it.
795 This caveat is now better explained in the documentation.
798 =head2 Diagnostics follow STDERR
800 Diagnostic output now goes to whichever file the C<STDERR> handle
801 is pointing at, instead of always going to the underlying C runtime
804 =head2 Other fixes for better diagnostics
806 Line numbers are suppressed no more (under most likely circumstances)
807 during the global destruction phase.
809 Diagnostics emitted from code running in threads other than the main
810 thread are now accompanied by the thread ID.
812 Embedded null characters in diagnostics now actually show up. They
813 used to truncate the message in prior versions.
815 $foo::a and $foo::b are now exempt from "possible typo" warnings only
816 if sort() is encountered in package foo.
818 Unrecognized alphabetic escapes encountered when parsing quoting
819 constructs now generate a warning, since they may take on new
820 semantics in later versions of Perl.
822 =head1 Performance enhancements
824 =head2 Simple sort() using { $a <=> $b } and the like are optimized
826 Many common sort() operations using a simple inlined block are now
827 optimized for faster performance.
829 =head2 Optimized assignments to lexical variables
831 Certain operations in the RHS of assignment statements have been
832 optimized to directly set the lexical variable on the LHS,
833 eliminating redundant copying overheads.
835 =head2 Method lookups optimized
837 [TODO - Chip Salzenberg <chip@perlsupport.com>]
839 =head2 Faster mechanism to invoke XSUBs
842 [TODO - Ilya Zakharevich <ilya@math.ohio-state.edu>]
844 =head2 Perl_malloc() improvements
847 [TODO - Ilya Zakharevich <ilya@math.ohio-state.edu>]
849 =head2 Faster subroutine calls
851 Minor changes in how subroutine calls are handled internally
852 provide marginal improvements in performance.
854 =head1 Platform specific changes
856 =head2 Additional supported platforms
862 VM/ESA is now supported.
866 Siemens BS2000 is now supported under the POSIX Shell.
870 The Mach CThreads (NEXTSTEP, OPENSTEP) are now supported by the Thread
875 GNU/Hurd is now supported.
879 Rhapsody is now supported.
883 EPOC is is now supported (on Psion 5).
889 [TODO - Laszlo Molnar <laszlo.molnar@eth.ericsson.se>]
893 [TODO - Ilya Zakharevich <ilya@math.ohio-state.edu>]
897 [TODO - Charles Bailey <bailey@newman.upenn.edu>]
901 Site library searches failed to look for ".../site/5.XXX/lib"
902 if ".../site/5.XXXYY/lib" wasn't found. This has been corrected.
904 When given a pathname that consists only of a drivename, such
905 as C<A:>, opendir() and stat() now use the current working
906 directory for the drive rather than the drive root.
908 The builtin XSUB functions in the Win32:: namespace are
909 documented. See L<Win32>.
911 $^X now contains the full path name of the running executable.
913 A Win32::GetLongPathName() function is provided to complement
914 Win32::GetFullPathName() and Win32::GetShortPathName(). See L<Win32>.
916 POSIX::uname() is supported.
918 system(1,...) now returns true process IDs rather than process
919 handles. kill() accepts any real process id, rather than strictly
920 return values from system(1,...).
922 The C<Shell> module is supported.
932 Compatibility tests for C<sub : attrs> vs the older C<use attrs>.
936 IO constants (SEEK_*, _IO*).
940 Directory-related IO methods (new, read, close, rewind, tied delete).
942 =item lib/io_multihomed
944 INET sockets with multi-homed hosts.
956 Regression tests for C<my ($x,@y,%z) : attrs> and <sub : attrs>.
964 Verify operations that access pad objects (lexicals and temporaries).
968 =head1 Modules and Pragmata
976 While used internally by Perl as a pragma, this module also
977 provides a way to fetch subroutine and variable attributes.
982 [TODO - Vishal Bhatia <vishal@gol.com>,
983 Nick Ing-Simmons <nick@ni-s.u-net.com>]
987 The ByteLoader is a dedicated extension to generate and run
988 Perl bytecode. See L<ByteLoader>.
992 The Perl Compiler suite has been extensively reworked for this
997 References can now be used. See L<constant>.
1002 [TODO - Ilya Zakharevich <ilya@math.ohio-state.edu>]
1006 A C<Maxdepth> setting can be specified to avoid venturing
1007 too deeply into data structures that may be very deep.
1008 See L<Data::Dumper>.
1010 Dumping C<qr//> objects works correctly.
1014 C<DB> is an experimental module that exposes a clean abstraction
1015 to Perl's debugging API.
1019 [TODO - Paul Marquess <paul.marquess@bt.com>]
1023 Devel::DProf, a Perl source code profiler has been added. See L<DProf>.
1027 Added Dumpvalue module provides screen dumps of Perl data.
1031 You can now run tests for I<n> seconds instead of guessing the right
1032 number of tests to run: e.g. timethese(-5, ...) will run each
1033 code for at least 5 CPU seconds. Zero as the "number of repetitions"
1034 means "for at least 3 CPU seconds". The output format has also
1035 changed. For example:
1037 use Benchmark;$x=3;timethese(-5,{a=>sub{$x*$x},b=>sub{$x**2}})
1039 will now output something like this:
1041 Benchmark: running a, b, each for at least 5 CPU seconds...
1042 a: 5 wallclock secs ( 5.77 usr + 0.00 sys = 5.77 CPU) @ 200551.91/s (n=1156516)
1043 b: 4 wallclock secs ( 5.00 usr + 0.02 sys = 5.02 CPU) @ 159605.18/s (n=800686)
1045 New features: "each for at least N CPU seconds...", "wallclock secs",
1046 and the "@ operations/CPU second (n=operations)".
1048 change#4265,4266,4292
1049 [TODO - Barrie Slaymaker <barries@slaysys.com>]
1053 The Devel::Peek module provides access to the internal representation
1054 of Perl variables and data. It is a data debugging tool for the XS programmer.
1056 =item ExtUtils::MakeMaker
1058 change#4135, also needs docs in module pod
1059 [TODO - Ilya Zakharevich <ilya@math.ohio-state.edu>]
1063 More Fcntl constants added: F_SETLK64, F_SETLKW64, O_LARGEFILE for
1064 large (more than 4G) file access (64-bit support is not yet
1065 working, though, so no need to get overly excited), Free/Net/OpenBSD
1066 locking behaviour flags F_FLOCK, F_POSIX, Linux F_SHLCK, and
1067 O_ACCMODE: the mask of O_RDONLY, O_WRONLY, and O_RDWR.
1071 A compare_text() function has been added, which allows custom
1072 comparison functions. See L<File::Compare>.
1076 File::Find now works correctly when the wanted() function is either
1077 autoloaded or is a symbolic reference.
1079 A bug that caused File::Find to lose track of the working directory
1080 when pruning top-level directories has been fixed.
1084 New methods have been added to the File::Spec module: devnull() returns
1085 the name of the null device (/dev/null on Unix) and tmpdir() the name of
1086 the temp directory (normally /tmp on Unix). There are now also methods
1087 to convert between absolute and relative filenames: abs2rel() and
1088 rel2abs(). For compatibility with operating systems that specify volume
1089 names in file paths, the splitpath(), splitdir(), and catdir() methods
1092 =item File::Spec::Functions
1094 The new File::Spec::Functions modules provides a function interface
1095 to the File::Spec module. Allows shorthand
1097 $fullname = catfile($dir1, $dir2, $file);
1101 $fullname = File::Spec->catfile($dir1, $dir2, $file);
1105 [TODO - Johan Vromans <jvromans@squirrel.nl>]
1109 write() and syswrite() will now accept a single-argument
1110 form of the call, for consistency with Perl's syswrite().
1112 You can now create a TCP-based IO::Socket::INET without forcing
1113 a connect attempt. This allows you to configure its options
1114 (like making it non-blocking) and then call connect() manually.
1116 A bug that prevented the IO::Socket::protocol() accessor
1117 from ever returning the correct value has been corrected.
1121 Java Perl Lingo is now distributed with Perl. See jpl/README
1122 for more information.
1126 The logical operations C<E<lt>E<lt>>, C<E<gt>E<gt>>, C<&>, C<|>,
1127 and C<~> are now supported on bigints.
1131 The accessor methods Re, Im, arg, abs, rho, and theta can now also
1132 act as mutators (accessor $z->Re(), mutator $z->Re(3)).
1136 A little bit of radial trigonometry (cylindrical and spherical),
1137 radial coordinate conversions, and the great circle distance were added.
1141 [TODO - Brad Appleton <bradapp@enteract.com>]
1143 =item Pod::Text and Pod::Man
1145 [TODO - Russ Allbery <rra@stanford.edu>]
1149 An EXISTS method has been added to this module (and sdbm_exists() has
1150 been added to the underlying sdbm library), so one can now call exists
1151 on an SDBM_File tied hash and get the correct result, rather than a
1154 A bug that may have caused data loss when more than one disk block
1155 happens to be read from the database in a single FETCH() has been
1160 The timelocal() and timegm() functions used to silently return bogus
1161 results when the date exceeded the machine's integer range. They
1162 now consistently croak() if the date falls in an unsupported range.
1166 The error return value in list context has been changed for all functions
1167 that return a list of values. Previously these functions returned a list
1168 with a single element C<undef> if an error occurred. Now these functions
1169 return the empty list in these situations. This applies to the following
1175 The remaining functions are unchanged and continue to return C<undef> on
1176 error even in list context.
1178 The Win32::SetLastError(ERROR) function has been added as a complement
1179 to the Win32::GetLastError() function.
1181 The new Win32::GetFullPathName(FILENAME) returns the full absolute
1182 pathname for FILENAME in scalar context. In list context it returns
1183 a two-element list containing the fully qualified directory name and
1188 A new feature called "DBM Filters" has been added to all the
1189 DBM modules--DB_File, GDBM_File, NDBM_File, ODBM_File, and SDBM_File.
1190 DBM Filters add four new methods to each DBM module:
1197 These can be used to filter key-value pairs before the pairs are
1198 written to the database or just after they are read from the database.
1199 See L<perldbmfilter> for further information.
1205 C<use attrs> is now obsolescent, and is only provided for
1206 backward-compatibility. It's been replaced by the C<sub : attributes>
1207 syntax. See L<perlsub/"Subroutine Attributes"> and L<attributes>.
1209 C<use utf8> to enable UTF-8 and Unicode support.
1211 C<use caller 'encoding'> allows modules to inherit pragmatic attributes
1212 from the caller's context. C<encoding> is currently the only supported
1215 Lexical warnings pragma, C<use warnings;>, to control optional warnings.
1218 C<use filetest> to control the behaviour of filetests (C<-r> C<-w>
1219 ...). Currently only one subpragma implemented, "use filetest
1220 'access';", that uses access(2) or equivalent to check permissions
1221 instead of using stat(2) as usual. This matters in filesystems
1222 where there are ACLs (access control lists): the stat(2) might lie,
1223 but access(2) knows better.
1225 =head1 Utility Changes
1229 [TODO - Kurt Starsinic <kstar@chapin.edu>]
1233 C<perlcc> now supports the C and Bytecode backends. By default,
1234 it generates output from the simple C backend rather than the
1235 optimized C backend.
1237 Support for non-Unix platforms has been improved.
1242 [TODO - Ilya Zakharevich <ilya@math.ohio-state.edu>]
1244 =head1 Documentation Changes
1248 =item perlopentut.pod
1250 A tutorial on using open() effectively.
1252 =item perlreftut.pod
1254 A tutorial that introduces the essentials of references.
1258 A tutorial on managing class data for object modules.
1260 =item perlcompile.pod
1262 An introduction to using the Perl Compiler suite.
1266 =head1 New Diagnostics
1268 =item "my sub" not yet implemented
1270 (F) Lexically scoped subroutines are not yet implemented. Don't try that
1273 =item %s package attribute may clash with future reserved word: %s
1275 (W) A lowercase attribute name was used that had a package-specific handler.
1276 That name might have a meaning to Perl itself some day, even though it
1277 doesn't yet. Perhaps you should use a mixed-case attribute name, instead.
1280 =item /%s/: Unrecognized escape \\%c passed through
1282 (W) You used a backslash-character combination which is not recognized
1283 by Perl. This combination appears in an interpolated variable or a
1284 C<'>-delimited regular expression.
1286 =item Filehandle %s opened only for output
1288 (W) You tried to read from a filehandle opened only for writing. If you
1289 intended it to be a read-write filehandle, you needed to open it with
1290 "+E<lt>" or "+E<gt>" or "+E<gt>E<gt>" instead of with "E<lt>" or nothing. If
1291 you intended only to read from the file, use "E<lt>". See
1294 =item Invalid %s attribute: %s
1296 The indicated attribute for a subroutine or variable was not recognized
1297 by Perl or by a user-supplied handler. See L<attributes>.
1299 =item Invalid %s attributes: %s
1301 The indicated attributes for a subroutine or variable were not recognized
1302 by Perl or by a user-supplied handler. See L<attributes>.
1304 =item Invalid separator character %s in attribute list
1306 (F) Something other than a comma or whitespace was seen between the
1307 elements of an attribute list. If the previous attribute
1308 had a parenthesised parameter list, perhaps that list was terminated
1309 too soon. See L<attributes>.
1311 =item Missing command in piped open
1313 (W) You used the C<open(FH, "| command")> or C<open(FH, "command |")>
1314 construction, but the command was missing or blank.
1316 =item Missing name in "my sub"
1318 (F) The reserved syntax for lexically scoped subroutines requires that they
1319 have a name with which they can be found.
1321 =item Unrecognized escape \\%c passed through
1323 (W) You used a backslash-character combination which is not recognized
1326 =item Unterminated attribute parameter in attribute list
1328 (F) The lexer saw an opening (left) parenthesis character while parsing an
1329 attribute list, but the matching closing (right) parenthesis
1330 character was not found. You may need to add (or remove) a backslash
1331 character to get your parentheses to balance. See L<attributes>.
1333 =item Unterminated attribute list
1335 (F) The lexer found something other than a simple identifier at the start
1336 of an attribute, and it wasn't a semicolon or the start of a
1337 block. Perhaps you terminated the parameter list of the previous attribute
1338 too soon. See L<attributes>.
1340 =item defined(@array) is deprecated
1342 (D) defined() is not usually useful on arrays because it checks for an
1343 undefined I<scalar> value. If you want to see if the array is empty,
1344 just use C<if (@array) { # not empty }> for example.
1346 =item defined(%hash) is deprecated
1348 (D) defined() is not usually useful on hashes because it checks for an
1349 undefined I<scalar> value. If you want to see if the hash is empty,
1350 just use C<if (%hash) { # not empty }> for example.
1352 =item Invalid separator character %s in subroutine attribute list
1354 (F) Something other than a comma or whitespace was seen between the
1355 elements of a subroutine attribute list. If the previous attribute
1356 had a parenthesised parameter list, perhaps that list was terminated
1359 =item Possible Y2K bug: %s
1361 (W) You are concatenating the number 19 with another number, which
1362 could be a potential Year 2000 problem.
1364 =item Unterminated attribute parameter in subroutine attribute list
1366 (F) The lexer saw an opening (left) parenthesis character while parsing a
1367 subroutine attribute list, but the matching closing (right) parenthesis
1368 character was not found. You may need to add (or remove) a backslash
1369 character to get your parentheses to balance.
1371 =item Unterminated subroutine attribute list
1373 (F) The lexer found something other than a simple identifier at the start
1374 of a subroutine attribute, and it wasn't a semicolon or the start of a
1375 block. Perhaps you terminated the parameter list of the previous attribute
1378 =item /%s/ should probably be written as "%s"
1380 (W) You have used a pattern where Perl expected to find a string,
1381 like in the first argument to C<join>. Perl will treat the true
1382 or false result of matching the pattern against $_ as the string,
1383 which is probably not what you had in mind.
1385 =item %s() called too early to check prototype
1387 (W) You've called a function that has a prototype before the parser saw a
1388 definition or declaration for it, and Perl could not check that the call
1389 conforms to the prototype. You need to either add an early prototype
1390 declaration for the subroutine in question, or move the subroutine
1391 definition ahead of the call to get proper prototype checking. Alternatively,
1392 if you are certain that you're calling the function correctly, you may put
1393 an ampersand before the name to avoid the warning. See L<perlsub>.
1395 =head1 Obsolete Diagnostics
1401 If you find what you think is a bug, you might check the headers of
1402 articles recently posted to the comp.lang.perl.misc newsgroup.
1403 There may also be information at http://www.perl.com/perl/, the Perl
1406 If you believe you have an unreported bug, please run the B<perlbug>
1407 program included with your release. Make sure to trim your bug down
1408 to a tiny but sufficient test case. Your bug report, along with the
1409 output of C<perl -V>, will be sent off to perlbug@perl.com to be
1410 analysed by the Perl porting team.
1414 The F<Changes> file for exhaustive details on what changed.
1416 The F<INSTALL> file for how to build Perl.
1418 The F<README> file for general stuff.
1420 The F<Artistic> and F<Copying> files for copyright information.
1424 Written by Gurusamy Sarathy <F<gsar@activestate.com>>, with many
1425 contributions from The Perl Porters.
1427 Send omissions or corrections to <F<perlbug@perl.com>>.