[Summary view] [Print] [Text view]

   1  =head1 NAME
   2  
   3  perl5005delta - what's new for perl5.005
   4  
   5  =head1 DESCRIPTION
   6  
   7  This document describes differences between the 5.004 release and this one.
   8  
   9  =head1 About the new versioning system
  10  
  11  Perl is now developed on two tracks: a maintenance track that makes
  12  small, safe updates to released production versions with emphasis on
  13  compatibility; and a development track that pursues more aggressive
  14  evolution.  Maintenance releases (which should be considered production
  15  quality) have subversion numbers that run from C<1> to C<49>, and
  16  development releases (which should be considered "alpha" quality) run
  17  from C<50> to C<99>.
  18  
  19  Perl 5.005 is the combined product of the new dual-track development
  20  scheme.
  21  
  22  =head1 Incompatible Changes
  23  
  24  =head2 WARNING:  This version is not binary compatible with Perl 5.004.
  25  
  26  Starting with Perl 5.004_50 there were many deep and far-reaching changes
  27  to the language internals.  If you have dynamically loaded extensions
  28  that you built under perl 5.003 or 5.004, you can continue to use them
  29  with 5.004, but you will need to rebuild and reinstall those extensions
  30  to use them 5.005.  See F<INSTALL> for detailed instructions on how to
  31  upgrade.
  32  
  33  =head2 Default installation structure has changed
  34  
  35  The new Configure defaults are designed to allow a smooth upgrade from
  36  5.004 to 5.005, but you should read F<INSTALL> for a detailed
  37  discussion of the changes in order to adapt them to your system.
  38  
  39  =head2 Perl Source Compatibility
  40  
  41  When none of the experimental features are enabled, there should be
  42  very few user-visible Perl source compatibility issues.
  43  
  44  If threads are enabled, then some caveats apply. C<@_> and C<$_> become
  45  lexical variables.  The effect of this should be largely transparent to
  46  the user, but there are some boundary conditions under which user will
  47  need to be aware of the issues.  For example, C<local(@_)> results in
  48  a "Can't localize lexical variable @_ ..." message.  This may be enabled
  49  in a future version.
  50  
  51  Some new keywords have been introduced.  These are generally expected to
  52  have very little impact on compatibility.  See L<New C<INIT> keyword>,
  53  L<New C<lock> keyword>, and L<New C<qrE<sol>E<sol>> operator>.
  54  
  55  Certain barewords are now reserved.  Use of these will provoke a warning
  56  if you have asked for them with the C<-w> switch.
  57  See L<C<our> is now a reserved word>.
  58  
  59  =head2 C Source Compatibility
  60  
  61  There have been a large number of changes in the internals to support
  62  the new features in this release.
  63  
  64  =over 4
  65  
  66  =item *
  67  
  68  Core sources now require ANSI C compiler
  69  
  70  An ANSI C compiler is now B<required> to build perl.  See F<INSTALL>.
  71  
  72  =item *
  73  
  74  All Perl global variables must now be referenced with an explicit prefix
  75  
  76  All Perl global variables that are visible for use by extensions now
  77  have a C<PL_> prefix.  New extensions should C<not> refer to perl globals
  78  by their unqualified names.  To preserve sanity, we provide limited
  79  backward compatibility for globals that are being widely used like
  80  C<sv_undef> and C<na> (which should now be written as C<PL_sv_undef>,
  81  C<PL_na> etc.)
  82  
  83  If you find that your XS extension does not compile anymore because a
  84  perl global is not visible, try adding a C<PL_> prefix to the global
  85  and rebuild.
  86  
  87  It is strongly recommended that all functions in the Perl API that don't
  88  begin with C<perl> be referenced with a C<Perl_> prefix.  The bare function
  89  names without the C<Perl_> prefix are supported with macros, but this
  90  support may cease in a future release.
  91  
  92  See L<perlapi>.
  93  
  94  =item *
  95  
  96  Enabling threads has source compatibility issues
  97  
  98  Perl built with threading enabled requires extensions to use the new
  99  C<dTHR> macro to initialize the handle to access per-thread data.
 100  If you see a compiler error that talks about the variable C<thr> not
 101  being declared (when building a module that has XS code),  you need
 102  to add C<dTHR;> at the beginning of the block that elicited the error.
 103  
 104  The API function C<perl_get_sv("@",FALSE)> should be used instead of
 105  directly accessing perl globals as C<GvSV(errgv)>.  The API call is
 106  backward compatible with existing perls and provides source compatibility
 107  with threading is enabled.
 108  
 109  See L<"C Source Compatibility"> for more information.
 110  
 111  =back
 112  
 113  =head2 Binary Compatibility
 114  
 115  This version is NOT binary compatible with older versions.  All extensions
 116  will need to be recompiled.  Further binaries built with threads enabled
 117  are incompatible with binaries built without.  This should largely be
 118  transparent to the user, as all binary incompatible configurations have
 119  their own unique architecture name, and extension binaries get installed at
 120  unique locations.  This allows coexistence of several configurations in
 121  the same directory hierarchy.  See F<INSTALL>.
 122  
 123  =head2 Security fixes may affect compatibility
 124  
 125  A few taint leaks and taint omissions have been corrected.  This may lead
 126  to "failure" of scripts that used to work with older versions.  Compiling
 127  with -DINCOMPLETE_TAINTS provides a perl with minimal amounts of changes
 128  to the tainting behavior.  But note that the resulting perl will have
 129  known insecurities.
 130  
 131  Oneliners with the C<-e> switch do not create temporary files anymore.
 132  
 133  =head2 Relaxed new mandatory warnings introduced in 5.004
 134  
 135  Many new warnings that were introduced in 5.004 have been made
 136  optional.  Some of these warnings are still present, but perl's new
 137  features make them less often a problem.  See L<New Diagnostics>.
 138  
 139  =head2 Licensing
 140  
 141  Perl has a new Social Contract for contributors.  See F<Porting/Contract>.
 142  
 143  The license included in much of the Perl documentation has changed.
 144  Most of the Perl documentation was previously under the implicit GNU
 145  General Public License or the Artistic License (at the user's choice).
 146  Now much of the documentation unambiguously states the terms under which
 147  it may be distributed.  Those terms are in general much less restrictive
 148  than the GNU GPL.  See L<perl> and the individual perl manpages listed
 149  therein.
 150  
 151  =head1 Core Changes
 152  
 153  
 154  =head2 Threads
 155  
 156  WARNING: Threading is considered an B<experimental> feature.  Details of the
 157  implementation may change without notice.  There are known limitations
 158  and some bugs.  These are expected to be fixed in future versions.
 159  
 160  See F<README.threads>.
 161  
 162  =head2 Compiler
 163  
 164  WARNING: The Compiler and related tools are considered B<experimental>.
 165  Features may change without notice, and there are known limitations
 166  and bugs.  Since the compiler is fully external to perl, the default
 167  configuration will build and install it.
 168  
 169  The Compiler produces three different types of transformations of a
 170  perl program.  The C backend generates C code that captures perl's state
 171  just before execution begins.  It eliminates the compile-time overheads
 172  of the regular perl interpreter, but the run-time performance remains
 173  comparatively the same.  The CC backend generates optimized C code
 174  equivalent to the code path at run-time.  The CC backend has greater
 175  potential for big optimizations, but only a few optimizations are
 176  implemented currently.  The Bytecode backend generates a platform
 177  independent bytecode representation of the interpreter's state
 178  just before execution.  Thus, the Bytecode back end also eliminates
 179  much of the compilation overhead of the interpreter.
 180  
 181  The compiler comes with several valuable utilities.
 182  
 183  C<B::Lint> is an experimental module to detect and warn about suspicious
 184  code, especially the cases that the C<-w> switch does not detect.
 185  
 186  C<B::Deparse> can be used to demystify perl code, and understand
 187  how perl optimizes certain constructs.
 188  
 189  C<B::Xref> generates cross reference reports of all definition and use
 190  of variables, subroutines and formats in a program.
 191  
 192  C<B::Showlex> show the lexical variables used by a subroutine or file
 193  at a glance.
 194  
 195  C<perlcc> is a simple frontend for compiling perl.
 196  
 197  See C<ext/B/README>, L<B>, and the respective compiler modules.
 198  
 199  =head2 Regular Expressions
 200  
 201  Perl's regular expression engine has been seriously overhauled, and
 202  many new constructs are supported.  Several bugs have been fixed.
 203  
 204  Here is an itemized summary:
 205  
 206  =over 4
 207  
 208  =item Many new and improved optimizations
 209  
 210  Changes in the RE engine:
 211  
 212      Unneeded nodes removed;
 213      Substrings merged together;
 214      New types of nodes to process (SUBEXPR)* and similar expressions
 215          quickly, used if the SUBEXPR has no side effects and matches
 216          strings of the same length;
 217      Better optimizations by lookup for constant substrings;
 218      Better search for constants substrings anchored by $ ;
 219  
 220  Changes in Perl code using RE engine:
 221  
 222      More optimizations to s/longer/short/;
 223      study() was not working;
 224      /blah/ may be optimized to an analogue of index() if $& $` $' not seen;
 225      Unneeded copying of matched-against string removed;
 226      Only matched part of the string is copying if $` $' were not seen;
 227  
 228  =item Many bug fixes
 229  
 230  Note that only the major bug fixes are listed here.  See F<Changes> for others.
 231  
 232      Backtracking might not restore start of $3.
 233      No feedback if max count for * or + on "complex" subexpression
 234          was reached, similarly (but at compile time) for {3,34567}
 235      Primitive restrictions on max count introduced to decrease a 
 236          possibility of a segfault;
 237      (ZERO-LENGTH)* could segfault;
 238      (ZERO-LENGTH)* was prohibited;
 239      Long REs were not allowed;
 240      /RE/g could skip matches at the same position after a 
 241        zero-length match;
 242  
 243  =item New regular expression constructs
 244  
 245  The following new syntax elements are supported:
 246  
 247      (?<=RE)
 248      (?<!RE)
 249      (?{ CODE })
 250      (?i-x)
 251      (?i:RE)
 252      (?(COND)YES_RE|NO_RE)
 253      (?>RE)
 254      \z
 255  
 256  =item New operator for precompiled regular expressions
 257  
 258  See L<New C<qrE<sol>E<sol>> operator>.
 259  
 260  =item Other improvements
 261  
 262      Better debugging output (possibly with colors),
 263              even from non-debugging Perl;
 264      RE engine code now looks like C, not like assembler;
 265      Behaviour of RE modifiable by `use re' directive;
 266      Improved documentation;
 267      Test suite significantly extended;
 268      Syntax [:^upper:] etc., reserved inside character classes;
 269  
 270  =item Incompatible changes
 271  
 272      (?i) localized inside enclosing group;
 273      $( is not interpolated into RE any more;
 274      /RE/g may match at the same position (with non-zero length)
 275          after a zero-length match (bug fix).
 276  
 277  =back
 278  
 279  See L<perlre> and L<perlop>.
 280  
 281  =head2   Improved malloc()
 282  
 283  See banner at the beginning of C<malloc.c> for details.
 284  
 285  =head2 Quicksort is internally implemented
 286  
 287  Perl now contains its own highly optimized qsort() routine.  The new qsort()
 288  is resistant to inconsistent comparison functions, so Perl's C<sort()> will
 289  not provoke coredumps any more when given poorly written sort subroutines.
 290  (Some C library C<qsort()>s that were being used before used to have this
 291  problem.)  In our testing, the new C<qsort()> required the minimal number
 292  of pair-wise compares on average, among all known C<qsort()> implementations.
 293  
 294  See C<perlfunc/sort>.
 295  
 296  =head2 Reliable signals
 297  
 298  Perl's signal handling is susceptible to random crashes, because signals
 299  arrive asynchronously, and the Perl runtime is not reentrant at arbitrary
 300  times.
 301  
 302  However, one experimental implementation of reliable signals is available
 303  when threads are enabled.  See C<Thread::Signal>.  Also see F<INSTALL> for
 304  how to build a Perl capable of threads.
 305  
 306  =head2 Reliable stack pointers
 307  
 308  The internals now reallocate the perl stack only at predictable times.
 309  In particular, magic calls never trigger reallocations of the stack,
 310  because all reentrancy of the runtime is handled using a "stack of stacks".
 311  This should improve reliability of cached stack pointers in the internals
 312  and in XSUBs.
 313  
 314  =head2 More generous treatment of carriage returns
 315  
 316  Perl used to complain if it encountered literal carriage returns in
 317  scripts.  Now they are mostly treated like whitespace within program text.
 318  Inside string literals and here documents, literal carriage returns are
 319  ignored if they occur paired with linefeeds, or get interpreted as whitespace
 320  if they stand alone.  This behavior means that literal carriage returns
 321  in files should be avoided.  You can get the older, more compatible (but
 322  less generous) behavior by defining the preprocessor symbol
 323  C<PERL_STRICT_CR> when building perl.  Of course, all this has nothing
 324  whatever to do with how escapes like C<\r> are handled within strings.
 325  
 326  Note that this doesn't somehow magically allow you to keep all text files
 327  in DOS format.  The generous treatment only applies to files that perl
 328  itself parses.  If your C compiler doesn't allow carriage returns in
 329  files, you may still be unable to build modules that need a C compiler.
 330  
 331  =head2 Memory leaks
 332  
 333  C<substr>, C<pos> and C<vec> don't leak memory anymore when used in lvalue
 334  context.  Many small leaks that impacted applications that embed multiple
 335  interpreters have been fixed.
 336  
 337  =head2 Better support for multiple interpreters
 338  
 339  The build-time option C<-DMULTIPLICITY> has had many of the details
 340  reworked.  Some previously global variables that should have been
 341  per-interpreter now are.  With care, this allows interpreters to call
 342  each other.  See the C<PerlInterp> extension on CPAN.
 343  
 344  =head2 Behavior of local() on array and hash elements is now well-defined
 345  
 346  See L<perlsub/"Temporary Values via local()">.
 347  
 348  =head2 C<%!> is transparently tied to the L<Errno> module
 349  
 350  See L<perlvar>, and L<Errno>.
 351  
 352  =head2 Pseudo-hashes are supported
 353  
 354  See L<perlref>.
 355  
 356  =head2 C<EXPR foreach EXPR> is supported
 357  
 358  See L<perlsyn>.
 359  
 360  =head2 Keywords can be globally overridden
 361  
 362  See L<perlsub>.
 363  
 364  =head2 C<$^E> is meaningful on Win32
 365  
 366  See L<perlvar>.
 367  
 368  =head2 C<foreach (1..1000000)> optimized
 369  
 370  C<foreach (1..1000000)> is now optimized into a counting loop.  It does
 371  not try to allocate a 1000000-size list anymore.
 372  
 373  =head2 C<Foo::> can be used as implicitly quoted package name
 374  
 375  Barewords caused unintuitive behavior when a subroutine with the same
 376  name as a package happened to be defined.  Thus, C<new Foo @args>,
 377  use the result of the call to C<Foo()> instead of C<Foo> being treated
 378  as a literal.  The recommended way to write barewords in the indirect
 379  object slot is C<new Foo:: @args>.  Note that the method C<new()> is
 380  called with a first argument of C<Foo>, not C<Foo::> when you do that.
 381  
 382  =head2 C<exists $Foo::{Bar::}> tests existence of a package
 383  
 384  It was impossible to test for the existence of a package without
 385  actually creating it before.  Now C<exists $Foo::{Bar::}> can be
 386  used to test if the C<Foo::Bar> namespace has been created.
 387  
 388  =head2 Better locale support
 389  
 390  See L<perllocale>.
 391  
 392  =head2 Experimental support for 64-bit platforms
 393  
 394  Perl5 has always had 64-bit support on systems with 64-bit longs.
 395  Starting with 5.005, the beginnings of experimental support for systems
 396  with 32-bit long and 64-bit 'long long' integers has been added.
 397  If you add -DUSE_LONG_LONG to your ccflags in config.sh (or manually
 398  define it in perl.h) then perl will be built with 'long long' support.
 399  There will be many compiler warnings, and the resultant perl may not
 400  work on all systems.  There are many other issues related to
 401  third-party extensions and libraries.  This option exists to allow
 402  people to work on those issues.
 403  
 404  =head2 prototype() returns useful results on builtins
 405  
 406  See L<perlfunc/prototype>.
 407  
 408  =head2 Extended support for exception handling
 409  
 410  C<die()> now accepts a reference value, and C<$@> gets set to that
 411  value in exception traps.  This makes it possible to propagate
 412  exception objects.  This is an undocumented B<experimental> feature.
 413  
 414  =head2 Re-blessing in DESTROY() supported for chaining DESTROY() methods
 415  
 416  See L<perlobj/Destructors>.
 417  
 418  =head2 All C<printf> format conversions are handled internally
 419  
 420  See L<perlfunc/printf>.
 421  
 422  =head2 New C<INIT> keyword
 423  
 424  C<INIT> subs are like C<BEGIN> and C<END>, but they get run just before
 425  the perl runtime begins execution.  e.g., the Perl Compiler makes use of
 426  C<INIT> blocks to initialize and resolve pointers to XSUBs.
 427  
 428  =head2 New C<lock> keyword
 429  
 430  The C<lock> keyword is the fundamental synchronization primitive
 431  in threaded perl.  When threads are not enabled, it is currently a noop.
 432  
 433  To minimize impact on source compatibility this keyword is "weak", i.e., any
 434  user-defined subroutine of the same name overrides it, unless a C<use Thread>
 435  has been seen.
 436  
 437  =head2 New C<qr//> operator
 438  
 439  The C<qr//> operator, which is syntactically similar to the other quote-like
 440  operators, is used to create precompiled regular expressions.  This compiled
 441  form can now be explicitly passed around in variables, and interpolated in
 442  other regular expressions.  See L<perlop>.
 443  
 444  =head2 C<our> is now a reserved word
 445  
 446  Calling a subroutine with the name C<our> will now provoke a warning when
 447  using the C<-w> switch.
 448  
 449  =head2 Tied arrays are now fully supported
 450  
 451  See L<Tie::Array>.
 452  
 453  =head2 Tied handles support is better
 454  
 455  Several missing hooks have been added.  There is also a new base class for
 456  TIEARRAY implementations.  See L<Tie::Array>.
 457  
 458  =head2 4th argument to substr
 459  
 460  substr() can now both return and replace in one operation.  The optional
 461  4th argument is the replacement string.  See L<perlfunc/substr>.
 462  
 463  =head2 Negative LENGTH argument to splice
 464  
 465  splice() with a negative LENGTH argument now work similar to what the
 466  LENGTH did for substr().  Previously a negative LENGTH was treated as
 467  0.  See L<perlfunc/splice>.
 468  
 469  =head2 Magic lvalues are now more magical
 470  
 471  When you say something like C<substr($x, 5) = "hi">, the scalar returned
 472  by substr() is special, in that any modifications to it affect $x.
 473  (This is called a 'magic lvalue' because an 'lvalue' is something on
 474  the left side of an assignment.)  Normally, this is exactly what you
 475  would expect to happen, but Perl uses the same magic if you use substr(),
 476  pos(), or vec() in a context where they might be modified, like taking
 477  a reference with C<\> or as an argument to a sub that modifies C<@_>.
 478  In previous versions, this 'magic' only went one way, but now changes
 479  to the scalar the magic refers to ($x in the above example) affect the
 480  magic lvalue too. For instance, this code now acts differently:
 481  
 482      $x = "hello";
 483      sub printit {
 484      $x = "g'bye";
 485      print $_[0], "\n";
 486      }
 487      printit(substr($x, 0, 5));
 488  
 489  In previous versions, this would print "hello", but it now prints "g'bye".
 490  
 491  =head2 <> now reads in records
 492  
 493  If C<$/> is a reference to an integer, or a scalar that holds an integer,
 494  <> will read in records instead of lines. For more info, see
 495  L<perlvar/$E<sol>>.
 496  
 497  =head1 Supported Platforms
 498  
 499  Configure has many incremental improvements.  Site-wide policy for building
 500  perl can now be made persistent, via Policy.sh.  Configure also records
 501  the command-line arguments used in F<config.sh>.
 502  
 503  =head2 New Platforms
 504  
 505  BeOS is now supported.  See F<README.beos>.
 506  
 507  DOS is now supported under the DJGPP tools.  See F<README.dos> (installed 
 508  as L<perldos> on some systems).
 509  
 510  MiNT is now supported.  See F<README.mint>.
 511  
 512  MPE/iX is now supported.  See F<README.mpeix>.
 513  
 514  MVS (aka OS390, aka Open Edition) is now supported.  See F<README.os390> 
 515  (installed as L<perlos390> on some systems).
 516  
 517  Stratus VOS is now supported.  See F<README.vos>.
 518  
 519  =head2 Changes in existing support
 520  
 521  Win32 support has been vastly enhanced.  Support for Perl Object, a C++
 522  encapsulation of Perl.  GCC and EGCS are now supported on Win32.
 523  See F<README.win32>, aka L<perlwin32>.
 524  
 525  VMS configuration system has been rewritten.  See F<README.vms> (installed 
 526  as L<README_vms> on some systems).
 527  
 528  The hints files for most Unix platforms have seen incremental improvements.
 529  
 530  =head1 Modules and Pragmata
 531  
 532  =head2 New Modules
 533  
 534  =over 4
 535  
 536  =item B
 537  
 538  Perl compiler and tools.  See L<B>.
 539  
 540  =item Data::Dumper
 541  
 542  A module to pretty print Perl data.  See L<Data::Dumper>.
 543  
 544  =item Dumpvalue
 545  
 546  A module to dump perl values to the screen. See L<Dumpvalue>.
 547  
 548  =item Errno
 549  
 550  A module to look up errors more conveniently.  See L<Errno>.
 551  
 552  =item File::Spec
 553  
 554  A portable API for file operations.
 555  
 556  =item ExtUtils::Installed
 557  
 558  Query and manage installed modules.
 559  
 560  =item ExtUtils::Packlist
 561  
 562  Manipulate .packlist files.
 563  
 564  =item Fatal
 565  
 566  Make functions/builtins succeed or die.
 567  
 568  =item IPC::SysV
 569  
 570  Constants and other support infrastructure for System V IPC operations
 571  in perl.
 572  
 573  =item Test
 574  
 575  A framework for writing test suites.
 576  
 577  =item Tie::Array
 578  
 579  Base class for tied arrays.
 580  
 581  =item Tie::Handle
 582  
 583  Base class for tied handles.
 584  
 585  =item Thread
 586  
 587  Perl thread creation, manipulation, and support.
 588  
 589  =item attrs
 590  
 591  Set subroutine attributes.
 592  
 593  =item fields
 594  
 595  Compile-time class fields.
 596  
 597  =item re
 598  
 599  Various pragmata to control behavior of regular expressions.
 600  
 601  =back
 602  
 603  =head2 Changes in existing modules
 604  
 605  =over 4
 606  
 607  =item Benchmark
 608  
 609  You can now run tests for I<x> seconds instead of guessing the right
 610  number of tests to run.
 611  
 612  Keeps better time.
 613  
 614  =item Carp
 615  
 616  Carp has a new function cluck(). cluck() warns, like carp(), but also adds
 617  a stack backtrace to the error message, like confess().
 618  
 619  =item CGI
 620  
 621  CGI has been updated to version 2.42.
 622  
 623  =item Fcntl
 624  
 625  More Fcntl constants added: F_SETLK64, F_SETLKW64, O_LARGEFILE for
 626  large (more than 4G) file access (the 64-bit support is not yet
 627  working, though, so no need to get overly excited), Free/Net/OpenBSD
 628  locking behaviour flags F_FLOCK, F_POSIX, Linux F_SHLCK, and
 629  O_ACCMODE: the mask of O_RDONLY, O_WRONLY, and O_RDWR.
 630  
 631  =item Math::Complex
 632  
 633  The accessors methods Re, Im, arg, abs, rho, theta, methods can
 634  ($z->Re()) now also act as mutators ($z->Re(3)).
 635  
 636  =item Math::Trig
 637  
 638  A little bit of radial trigonometry (cylindrical and spherical) added,
 639  for example the great circle distance.
 640  
 641  =item POSIX
 642  
 643  POSIX now has its own platform-specific hints files.
 644  
 645  =item DB_File
 646  
 647  DB_File supports version 2.x of Berkeley DB.  See C<ext/DB_File/Changes>.
 648  
 649  =item MakeMaker
 650  
 651  MakeMaker now supports writing empty makefiles, provides a way to
 652  specify that site umask() policy should be honored.  There is also
 653  better support for manipulation of .packlist files, and getting
 654  information about installed modules.
 655  
 656  Extensions that have both architecture-dependent and
 657  architecture-independent files are now always installed completely in
 658  the architecture-dependent locations.  Previously, the shareable parts
 659  were shared both across architectures and across perl versions and were
 660  therefore liable to be overwritten with newer versions that might have
 661  subtle incompatibilities.
 662  
 663  =item CPAN
 664  
 665  See L<perlmodinstall> and L<CPAN>.
 666  
 667  =item Cwd
 668  
 669  Cwd::cwd is faster on most platforms.
 670  
 671  =back
 672  
 673  =head1 Utility Changes
 674  
 675  C<h2ph> and related utilities have been vastly overhauled.
 676  
 677  C<perlcc>, a new experimental front end for the compiler is available.
 678  
 679  The crude GNU C<configure> emulator is now called C<configure.gnu> to
 680  avoid trampling on C<Configure> under case-insensitive filesystems.
 681  
 682  C<perldoc> used to be rather slow.  The slower features are now optional.
 683  In particular, case-insensitive searches need the C<-i> switch, and
 684  recursive searches need C<-r>.  You can set these switches in the
 685  C<PERLDOC> environment variable to get the old behavior.
 686  
 687  =head1 Documentation Changes
 688  
 689  Config.pm now has a glossary of variables.
 690  
 691  F<Porting/patching.pod> has detailed instructions on how to create and
 692  submit patches for perl.
 693  
 694  L<perlport> specifies guidelines on how to write portably. 
 695  
 696  L<perlmodinstall> describes how to fetch and install modules from C<CPAN>
 697  sites.
 698  
 699  Some more Perl traps are documented now.  See L<perltrap>.
 700  
 701  L<perlopentut> gives a tutorial on using open().
 702  
 703  L<perlreftut> gives a tutorial on references.
 704  
 705  L<perlthrtut> gives a tutorial on threads.
 706  
 707  =head1 New Diagnostics
 708  
 709  =over 4
 710  
 711  =item Ambiguous call resolved as CORE::%s(), qualify as such or use &
 712  
 713  (W) A subroutine you have declared has the same name as a Perl keyword,
 714  and you have used the name without qualification for calling one or the
 715  other.  Perl decided to call the builtin because the subroutine is
 716  not imported.
 717  
 718  To force interpretation as a subroutine call, either put an ampersand
 719  before the subroutine name, or qualify the name with its package.
 720  Alternatively, you can import the subroutine (or pretend that it's
 721  imported with the C<use subs> pragma).
 722  
 723  To silently interpret it as the Perl operator, use the C<CORE::> prefix
 724  on the operator (e.g. C<CORE::log($x)>) or by declaring the subroutine
 725  to be an object method (see L<attrs>).
 726  
 727  =item Bad index while coercing array into hash
 728  
 729  (F) The index looked up in the hash found as the 0'th element of a
 730  pseudo-hash is not legal.  Index values must be at 1 or greater.
 731  See L<perlref>.
 732  
 733  =item Bareword "%s" refers to nonexistent package
 734  
 735  (W) You used a qualified bareword of the form C<Foo::>, but
 736  the compiler saw no other uses of that namespace before that point.
 737  Perhaps you need to predeclare a package?
 738  
 739  =item Can't call method "%s" on an undefined value
 740  
 741  (F) You used the syntax of a method call, but the slot filled by the
 742  object reference or package name contains an undefined value.
 743  Something like this will reproduce the error:
 744  
 745      $BADREF = 42;
 746      process $BADREF 1,2,3;
 747      $BADREF->process(1,2,3);
 748  
 749  =item Can't check filesystem of script "%s" for nosuid
 750  
 751  (P) For some reason you can't check the filesystem of the script for nosuid.
 752  
 753  =item Can't coerce array into hash
 754  
 755  (F) You used an array where a hash was expected, but the array has no
 756  information on how to map from keys to array indices.  You can do that
 757  only with arrays that have a hash reference at index 0.
 758  
 759  =item Can't goto subroutine from an eval-string
 760  
 761  (F) The "goto subroutine" call can't be used to jump out of an eval "string".
 762  (You can use it to jump out of an eval {BLOCK}, but you probably don't want to.)
 763  
 764  =item Can't localize pseudo-hash element
 765  
 766  (F) You said something like C<< local $ar->{'key'} >>, where $ar is
 767  a reference to a pseudo-hash.  That hasn't been implemented yet, but
 768  you can get a similar effect by localizing the corresponding array
 769  element directly -- C<< local $ar->[$ar->[0]{'key'}] >>.
 770  
 771  =item Can't use %%! because Errno.pm is not available
 772  
 773  (F) The first time the %! hash is used, perl automatically loads the
 774  Errno.pm module. The Errno module is expected to tie the %! hash to
 775  provide symbolic names for C<$!> errno values.
 776  
 777  =item Cannot find an opnumber for "%s"
 778  
 779  (F) A string of a form C<CORE::word> was given to prototype(), but
 780  there is no builtin with the name C<word>.
 781  
 782  =item Character class syntax [. .] is reserved for future extensions
 783  
 784  (W) Within regular expression character classes ([]) the syntax beginning
 785  with "[." and ending with ".]" is reserved for future extensions.
 786  If you need to represent those character sequences inside a regular
 787  expression character class, just quote the square brackets with the
 788  backslash: "\[." and ".\]".
 789  
 790  =item Character class syntax [: :] is reserved for future extensions
 791  
 792  (W) Within regular expression character classes ([]) the syntax beginning
 793  with "[:" and ending with ":]" is reserved for future extensions.
 794  If you need to represent those character sequences inside a regular
 795  expression character class, just quote the square brackets with the
 796  backslash: "\[:" and ":\]".
 797  
 798  =item Character class syntax [= =] is reserved for future extensions
 799  
 800  (W) Within regular expression character classes ([]) the syntax
 801  beginning with "[=" and ending with "=]" is reserved for future extensions.
 802  If you need to represent those character sequences inside a regular
 803  expression character class, just quote the square brackets with the
 804  backslash: "\[=" and "=\]".
 805  
 806  =item %s: Eval-group in insecure regular expression
 807  
 808  (F) Perl detected tainted data when trying to compile a regular expression
 809  that contains the C<(?{ ... })> zero-width assertion, which is unsafe.
 810  See L<perlre/(?{ code })>, and L<perlsec>.
 811  
 812  =item %s: Eval-group not allowed, use re 'eval'
 813  
 814  (F) A regular expression contained the C<(?{ ... })> zero-width assertion,
 815  but that construct is only allowed when the C<use re 'eval'> pragma is
 816  in effect.  See L<perlre/(?{ code })>.
 817  
 818  =item %s: Eval-group not allowed at run time
 819  
 820  (F) Perl tried to compile a regular expression containing the C<(?{ ... })>
 821  zero-width assertion at run time, as it would when the pattern contains
 822  interpolated values.  Since that is a security risk, it is not allowed.
 823  If you insist, you may still do this by explicitly building the pattern
 824  from an interpolated string at run time and using that in an eval().
 825  See L<perlre/(?{ code })>.
 826  
 827  =item Explicit blessing to '' (assuming package main)
 828  
 829  (W) You are blessing a reference to a zero length string.  This has
 830  the effect of blessing the reference into the package main.  This is
 831  usually not what you want.  Consider providing a default target
 832  package, e.g. bless($ref, $p || 'MyPackage');
 833  
 834  =item Illegal hex digit ignored
 835  
 836  (W) You may have tried to use a character other than 0 - 9 or A - F in a
 837  hexadecimal number.  Interpretation of the hexadecimal number stopped
 838  before the illegal character.
 839  
 840  =item No such array field
 841  
 842  (F) You tried to access an array as a hash, but the field name used is
 843  not defined.  The hash at index 0 should map all valid field names to
 844  array indices for that to work.
 845  
 846  =item No such field "%s" in variable %s of type %s
 847  
 848  (F) You tried to access a field of a typed variable where the type
 849  does not know about the field name.  The field names are looked up in
 850  the %FIELDS hash in the type package at compile time.  The %FIELDS hash
 851  is usually set up with the 'fields' pragma.
 852  
 853  =item Out of memory during ridiculously large request
 854  
 855  (F) You can't allocate more than 2^31+"small amount" bytes.  This error
 856  is most likely to be caused by a typo in the Perl program. e.g., C<$arr[time]>
 857  instead of C<$arr[$time]>.
 858  
 859  =item Range iterator outside integer range
 860  
 861  (F) One (or both) of the numeric arguments to the range operator ".."
 862  are outside the range which can be represented by integers internally.
 863  One possible workaround is to force Perl to use magical string
 864  increment by prepending "0" to your numbers.
 865  
 866  =item Recursive inheritance detected while looking for method '%s' %s
 867  
 868  (F) More than 100 levels of inheritance were encountered while invoking a
 869  method.  Probably indicates an unintended loop in your inheritance hierarchy.
 870  
 871  =item Reference found where even-sized list expected
 872  
 873  (W) You gave a single reference where Perl was expecting a list with
 874  an even number of elements (for assignment to a hash). This
 875  usually means that you used the anon hash constructor when you meant 
 876  to use parens. In any case, a hash requires key/value B<pairs>.
 877  
 878      %hash = { one => 1, two => 2, };   # WRONG
 879      %hash = [ qw/ an anon array / ];   # WRONG
 880      %hash = ( one => 1, two => 2, );   # right
 881      %hash = qw( one 1 two 2 );                 # also fine
 882  
 883  =item Undefined value assigned to typeglob
 884  
 885  (W) An undefined value was assigned to a typeglob, a la C<*foo = undef>.
 886  This does nothing.  It's possible that you really mean C<undef *foo>.
 887  
 888  =item Use of reserved word "%s" is deprecated
 889  
 890  (D) The indicated bareword is a reserved word.  Future versions of perl
 891  may use it as a keyword, so you're better off either explicitly quoting
 892  the word in a manner appropriate for its context of use, or using a
 893  different name altogether.  The warning can be suppressed for subroutine
 894  names by either adding a C<&> prefix, or using a package qualifier,
 895  e.g. C<&our()>, or C<Foo::our()>.
 896  
 897  =item perl: warning: Setting locale failed.
 898  
 899  (S) The whole warning message will look something like:
 900  
 901         perl: warning: Setting locale failed.
 902         perl: warning: Please check that your locale settings:
 903                 LC_ALL = "En_US",
 904                 LANG = (unset)
 905             are supported and installed on your system.
 906         perl: warning: Falling back to the standard locale ("C").
 907  
 908  Exactly what were the failed locale settings varies.  In the above the
 909  settings were that the LC_ALL was "En_US" and the LANG had no value.
 910  This error means that Perl detected that you and/or your system
 911  administrator have set up the so-called variable system but Perl could
 912  not use those settings.  This was not dead serious, fortunately: there
 913  is a "default locale" called "C" that Perl can and will use, the
 914  script will be run.  Before you really fix the problem, however, you
 915  will get the same error message each time you run Perl.  How to really
 916  fix the problem can be found in L<perllocale/"LOCALE PROBLEMS">.
 917  
 918  =back
 919  
 920  
 921  =head1 Obsolete Diagnostics
 922  
 923  =over 4
 924  
 925  =item Can't mktemp()
 926  
 927  (F) The mktemp() routine failed for some reason while trying to process
 928  a B<-e> switch.  Maybe your /tmp partition is full, or clobbered.
 929  
 930  Removed because B<-e> doesn't use temporary files any more.
 931  
 932  =item Can't write to temp file for B<-e>: %s
 933  
 934  (F) The write routine failed for some reason while trying to process
 935  a B<-e> switch.  Maybe your /tmp partition is full, or clobbered.
 936  
 937  Removed because B<-e> doesn't use temporary files any more.
 938  
 939  =item Cannot open temporary file
 940  
 941  (F) The create routine failed for some reason while trying to process
 942  a B<-e> switch.  Maybe your /tmp partition is full, or clobbered.
 943  
 944  Removed because B<-e> doesn't use temporary files any more.
 945  
 946  =item regexp too big
 947  
 948  (F) The current implementation of regular expressions uses shorts as
 949  address offsets within a string.  Unfortunately this means that if
 950  the regular expression compiles to longer than 32767, it'll blow up.
 951  Usually when you want a regular expression this big, there is a better
 952  way to do it with multiple statements.  See L<perlre>.
 953  
 954  =back
 955  
 956  =head1 Configuration Changes
 957  
 958  You can use "Configure -Uinstallusrbinperl" which causes installperl
 959  to skip installing perl also as /usr/bin/perl.  This is useful if you
 960  prefer not to modify /usr/bin for some reason or another but harmful
 961  because many scripts assume to find Perl in /usr/bin/perl.
 962  
 963  =head1 BUGS
 964  
 965  If you find what you think is a bug, you might check the headers of
 966  recently posted articles in the comp.lang.perl.misc newsgroup.
 967  There may also be information at http://www.perl.com/perl/ , the Perl
 968  Home Page.
 969  
 970  If you believe you have an unreported bug, please run the B<perlbug>
 971  program included with your release.  Make sure you trim your bug down
 972  to a tiny but sufficient test case.  Your bug report, along with the
 973  output of C<perl -V>, will be sent off to <F<perlbug@perl.com>> to be
 974  analysed by the Perl porting team.
 975  
 976  =head1 SEE ALSO
 977  
 978  The F<Changes> file for exhaustive details on what changed.
 979  
 980  The F<INSTALL> file for how to build Perl.
 981  
 982  The F<README> file for general stuff.
 983  
 984  The F<Artistic> and F<Copying> files for copyright information.
 985  
 986  =head1 HISTORY
 987  
 988  Written by Gurusamy Sarathy <F<gsar@activestate.com>>, with many contributions
 989  from The Perl Porters.
 990  
 991  Send omissions or corrections to <F<perlbug@perl.com>>.
 992  
 993  =cut