perl589delta - what is new for perl v5.8.9


NAME

perl589delta - what is new for perl v5.8.9


DESCRIPTION

This document describes differences between the 5.8.8 release and the 5.8.9 release.


Notice

The 5.8.9 release will be the last significant release of the 5.8.x series. Any future releases of 5.8.x will likely only be to deal with security issues, and platform build failures. Hence you should look to migrating to 5.10.x, if you have not started already. See Known Problems for more information.


Incompatible Changes

A particular construction in the source code of extensions written in C++ may need changing. See Changed Internals for more details. All extensions written in C, most written in C++, and all existing compiled extensions are unaffected. This was necessary to improve C++ support.

Other than this, there are no changes intentionally incompatible with 5.8.8. If any exist, they are bugs and reports are welcome.


Core Enhancements

Unicode Character Database 5.1.0.

The copy of the Unicode Character Database included in Perl 5.8 has been updated to 5.1.0 from 4.1.0. See http://www.unicode.org/versions/Unicode5.1.0/#NotableChanges for the notable changes.

stat and -X on directory handles

It is now possible to call stat and the -X filestat operators on directory handles. As both directory and file handles are barewords, there can be ambiguities over which was intended. In these situations the file handle semantics are preferred. Both also treat *FILE{IO} filehandles like *FILE filehandles.

Source filters in @INC

It's possible to enhance the mechanism of subroutine hooks in @INC by adding a source filter on top of the filehandle opened and returned by the hook. This feature was planned a long time ago, but wasn't quite working until now. See require in the perlfunc manpage for details. (Nicholas Clark)

Exceptions in constant folding

The constant folding routine is now wrapped in an exception handler, and if folding throws an exception (such as attempting to evaluate 0/0), perl now retains the current optree, rather than aborting the whole program. Without this change, programs would not compile if they had expressions that happened to generate exceptions, even though those expressions were in code that could never be reached at runtime. (Nicholas Clark, Dave Mitchell)

no VERSION

You can now use no followed by a version number to specify that you want to use a version of perl older than the specified one.

Improved internal UTF-8 caching code

The code that caches calculated UTF-8 byte offsets for character offsets for a string has been re-written. Several bugs have been located and eliminated, and the code now makes better use of the information it has, so should be faster. In particular, it doesn't scan to the end of a string before calculating an offset within the string, which should speed up some operations on long strings. It is now possible to disable the caching code at run time, to verify that it is not the cause of suspected problems.

Runtime relocatable installations

There is now Configure support for creating a perl tree that is relocatable at run time. see Relocatable installations.

New internal variables

${^CHILD_ERROR_NATIVE}
This variable gives the native status returned by the last pipe close, backtick command, successful call to wait or waitpid, or from the system operator. See the perlvar manpage for details. (Contributed by Gisle Aas.)

${^UTF8CACHE}
This variable controls the state of the internal UTF-8 offset caching code. 1 for on (the default), 0 for off, -1 to debug the caching code by checking all its results against linear scans, and panicking on any discrepancy.

readpipe is now overridable

The built-in function readpipe is now overridable. Overriding it permits also to override its operator counterpart, qx// (also known as ``).

simple exception handling macros

Perl 5.8.9 (and 5.10.0 onwards) now provides a couple of macros to do very basic exception handling in XS modules. You can use these macros if you call code that may croak, but you need to do some cleanup before giving control back to Perl. See Exception Handling in the perlguts manpage for more details.

-D option enhancements

XS-assisted SWASHGET

Some pure-perl code that the regexp engine was using to retrieve Unicode properties and transliteration mappings has been reimplemented in XS for faster execution. (SADAHIRO Tomoyuki)

Constant subroutines

The interpreter internals now support a far more memory efficient form of inlineable constants. Storing a reference to a constant value in a symbol table is equivalent to a full typeglob referencing a constant subroutine, but using about 400 bytes less memory. This proxy constant subroutine is automatically upgraded to a real typeglob with subroutine if necessary. The approach taken is analogous to the existing space optimisation for subroutine stub declarations, which are stored as plain scalars in place of the full typeglob.

However, to aid backwards compatibility of existing code, which (wrongly) does not expect anything other than typeglobs in symbol tables, nothing in core uses this feature, other than the regression tests.

Stubs for prototyped subroutines have been stored in symbol tables as plain strings, and stubs for unprototyped subroutines as the number -1, since 5.005, so code which assumes that the core only places typeglobs in symbol tables has been making incorrect assumptions for over 10 years.


New Platforms

Compile support added for:


Modules and Pragmata

New Modules

Updated Modules


Utility Changes

debugger upgraded to version 1.31

perlthanks

Perl 5.8.9 adds a new utility perlthanks, which is a variant of perlbug, but for sending non-bug-reports to the authors and maintainers of Perl. Getting nothing but bug reports can become a bit demoralising - we'll see if this changes things.

perlbug

perlbug now checks if you're reporting about a non-core module and suggests you report it to the CPAN author instead.

h2xs

h2ph


New Documentation

As usual, the documentation received its share of corrections, clarifications and other nitfixes. More tags were added for indexing.

the perlunitut manpage is a tutorial written by Juerd Waalboer on Unicode-related terminology and how to correctly handle Unicode in Perl scripts.

the perlunicode manpage is updated in section user defined properties.

the perluniintro manpage has been updated in the example of detecting data that is not valid in particular encoding.

the perlcommunity manpage provides an overview of the Perl Community along with further resources.

the CORE manpage documents the pseudo-namespace for Perl's core routines.


Changes to Existing Documentation

the perlglossary manpage adds deprecated modules and features and to be dropped modules.

the perlhack manpage has been updated and added resources on smoke testing.

The Perl FAQs (perlfaq1..perlfaq9) have been updated.

the perlcheat manpage is updated with better details on \w, \d, and \s.

the perldebug manpage is updated with information on how to call the debugger.

the perldiag manpage documentation updated with subroutine with an ampersand on the argument to exists and delete and also several terminology updates on warnings.

the perlfork manpage documents the limitation of exec inside pseudo-processes.

the perlfunc manpage:

the perllocale manpage documentation is adjusted for number localization and POSIX::setlocale to fix Debian bug #379463.

perlmodlib is updated with CPAN::API::HOWTO and Sys::Syslog::win32::Win32

the perlre manpage documentation updated to reflect the differences between [[:xxxxx:]] and \p{IsXxxxx} matches. Also added section on /g and /c modifiers.

the perlreguts manpage describe the internals of the regular expressions engine. It has been contributed by Yves Orton.

the perlrebackslash manpage describes all perl regular expression backslash and escape sequences.

the perlrecharclass manpage describes the syntax and use of character classes in Perl Regular Expressions.

the perlrun manpage is updated to clarify on the hash seed PERL_HASH_SEED. Also more information in options -x and -u.

the perlsub manpage example is updated to use a lexical variable for opendir syntax.

the perlvar manpage fixes confusion about real GID $( and effective GID $).

Perl thread tutorial example is fixed in section Queues: Passing Data Around in the perlthrtut manpage and the perlthrtut manpage.

the perlhack manpage documentation extensively improved by Jarkko Hietaniemi and others.

the perltoot manpage provides information on modifying @UNIVERSAL::ISA.

the perlport manpage documentation extended to include different kill(-9, ...) semantics on Windows. It also clearly states dump is not supported on Win32 and cygwin.

INSTALL has been updated and modernised.


Performance Enhancements


Installation and Configuration Improvements

Relocatable installations

There is now Configure support for creating a relocatable perl tree. If you Configure with -Duserelocatableinc, then the paths in @INC (and everything else in %Config) can be optionally located via the path of the perl executable.

At start time, if any paths in @INC or Config that Configure marked as relocatable (by starting them with ".../"), then they are prefixed the directory of $^X. This allows the relocation can be configured on a per-directory basis, although the default with -Duserelocatableinc is that everything is relocated. The initial install is done to the original configured prefix.

Configuration improvements

Configure is now better at removing temporary files. Tom Callaway (from RedHat) also contributed patches that complete the set of flags passed to the compiler and the linker, in particular that -fPIC is now enabled on Linux. It will also croak when your /dev/null isn't a device.

A new configuration variable d_pseudofork has been to Configure, and is available as $Config{d_pseudofork} in the Config module. This distinguishes real fork support from the pseudofork emulation used on Windows platforms.

Config.pod and config.sh are now placed correctly for cross-compilation.

$Config{useshrplib} is now 'true' rather than 'yes' when using a shared perl library.

Compilation improvements

Parallel makes should work properly now, although there may still be problems if make test is instructed to run in parallel.

Many compilation warnings have been cleaned up. A very stubborn compiler warning in S_emulate_eaccess() was killed after six attempts. g++ support has been tuned, especially for FreeBSD.

mkppport has been integrated, and all ppport.h files in the core will now be autogenerated at build time (and removed during cleanup).

Installation improvements.

installman now works with -Duserelocatableinc and DESTDIR.

installperl no longer installs:

Platform Specific Changes

There are improved hints for AIX, Cygwin, DEC/OSF, FreeBSD, HP/UX, Irix 6 Linux, MachTen, NetBSD, OS/390, QNX, SCO, Solaris, SunOS, System V Release 5.x (UnixWare 7, OpenUNIX 8), Ultrix, UMIPS, uts and VOS.

FreeBSD

Solaris

VMS

Windows


Selected Bug Fixes

Unicode

Many many bugs related to the internal Unicode implementation (UTF-8) have been fixed. In particular, long standing bugs related to returning Unicode via tie, overloading or $@ are now gone, some of which were never reported.

unpack will internally convert the string back from UTF-8 on numeric types. This is a compromise between the full consistency now in 5.10, and the current behaviour, which is often used as a ``feature'' on string types.

Using :crlf and UTF-16 IO layers together will now work.

Fixed problems with split, Unicode /\s+/ and / \0/.

Fixed bug RT #40641 - encoding of Unicode characters in regular expressions.

Fixed a bug where using certain patterns in a regexp led to a panic. [RT #45337]

Perl no longer segfaults (due to infinite internal recursion) if the locale's character is not UTF-8 [RT #41442]:

    use open ':locale';
    print STDERR "\x{201e}"; # „

PerlIO

Inconsistencies have been fixed in the reference counting PerlIO uses to keep track of Unix file descriptors, and the API used by XS code to manage getting and releasing FILE *s

Magic

Several bugs have been fixed in Magic, the internal system used to implement features such as tie, tainting and threads sharing.

undef @array on a tied array now correctly calls the CLEAR method.

Some of the bitwise ops were not checking whether their arguments were magical before using them. [RT #24816]

Magic is no longer invoked twice by the expression \&$x

A bug with assigning large numbers and tainting has been resolved. [RT #40708]

A new entry has been added to the MAGIC vtable - svt_local. This is used when copying magic to the new value during local, allowing certain problems with localising shared variables to be resolved.

For the implementation details, see Magic Virtual Tables in the perlguts manpage.

Reblessing overloaded objects now works

Internally, perl object-ness is on the referent, not the reference, even though methods can only be called via a reference. However, the original implementation of overloading stored flags related to overloading on the reference, relying on the flags being copied when the reference was copied, or set at the creation of a new reference. This manifests in a bug - if you rebless an object from a class that has overloading, into one that does not, then any other existing references think that they (still) point to an overloaded object, choose these C code paths, and then throw errors. Analogously, blessing into an overloaded class when other references exist will result in them not using overloading.

The implementation has been fixed for 5.10, but this fix changes the semantics of flag bits, so is not binary compatible, so can't be applied to 5.8.9. However, 5.8.9 has a work-around that implements the same bug fix. If the referent has multiple references, then all the other references are located and corrected. A full search is avoided whenever possible by scanning lexicals outwards from the current subroutine, and the argument stack.

A certain well known Linux vendor applied incomplete versions of this bug fix to their /usr/bin/perl and then prematurely closed bug reports about performance issues without consulting back upstream. This not being enough, they then proceeded to ignore the necessary fixes to these unreleased changes for 11 months, until massive pressure was applied by their long-suffering paying customers, catalysed by the failings being featured on a prominent blog and Slashdot.

strict now propagates correctly into string evals

Under 5.8.8 and earlier:

    $ perl5.8.8 -e 'use strict; eval "use foo bar" or die $@'
    Can't locate foo.pm in @INC (@INC contains: ... .) at (eval 1) line 2.
    BEGIN failed--compilation aborted at (eval 1) line 2.

Under 5.8.9 and later:

    $ perl5.8.9 -e 'use strict; eval "use foo bar" or die $@'
    Bareword "bar" not allowed while "strict subs" in use at (eval 1) line 1.

This may cause problems with programs that parse the error message and rely on the buggy behaviour.

Other fixes

Platform Specific Fixes

Darwin / MacOS X

OS/2

Tru64

RedHat Linux

Solaris/i386

VMS

Windows

Smaller fixes


New or Changed Diagnostics

panic: sv_chop %s

This new fatal error occurs when the C routine Perl_sv_chop() was passed a position that is not within the scalar's string buffer. This is caused by buggy XS code, and at this point recovery is not possible.

Maximal count of pending signals (%s) exceeded

This new fatal error occurs when the perl process has to abort due to too many pending signals, which is bound to prevent perl from being able to handle further incoming signals safely.

panic: attempt to call %s in %s

This new fatal error occurs when the ACL version file test operator is used where it is not available on the current platform. Earlier checks mean that it should never be possible to get this.

FETCHSIZE returned a negative value

New error indicating that a tied array has claimed to have a negative number of elements.

Can't upgrade %s (%d) to %d

Previously the internal error from the SV upgrade code was the less informative Can't upgrade that kind of scalar. It now reports the current internal type, and the new type requested.

%s argument is not a HASH or ARRAY element or a subroutine

This error, thrown if an invalid argument is provided to exists now correctly includes ``or a subroutine''. [RT #38955]

Cannot make the non-overridable builtin %s fatal

This error in Fatal previously did not show the name of the builtin in question (now represented by %s above).

Unrecognized character '%s' in column %d

This error previously did not state the column.

Offset outside string

This can now also be generated by a seek on a file handle using PerlIO::scalar.

Invalid escape in the specified encoding in regexp; marked by <-- HERE in m/%s/

New error, introduced as part of the fix to RT #40641 to handle encoding of Unicode characters in regular expression comments.

Your machine doesn't support dump/undump.

A more informative fatal error issued when calling dump on Win32 and Cygwin. (Given that the purpose of dump is to abort with a core dump, and core dumps can't be produced on these platforms, this is more useful than silently exiting.)


Changed Internals

The perl sources can now be compiled with a C++ compiler instead of a C compiler. A necessary implementation details is that under C++, the macro XS used to define XSUBs now includes an extern "C" definition. A side effect of this is that C++ code that used the construction

    typedef XS(SwigPerlWrapper);

now needs to be written

    typedef XSPROTO(SwigPerlWrapper);

using the new XSPROTO macro, in order to compile. C extensions are unaffected, although C extensions are encouraged to use XSPROTO too. This change was present in the 5.10.0 release of perl, so any actively maintained code that happened to use this construction should already have been adapted. Code that needs changing will fail with a compilation error.

set magic on localizing/assigning to a magic variable will now only trigger for container magics, i.e. it will for %ENV or %SIG but not for $#array.

The new API macro newSVpvs() can be used in place of constructions such as newSVpvn("ISA", 3). It takes a single string constant, and at C compile time determines its length.

The new API function Perl_newSV_type() can be used as a more efficient replacement of the common idiom

    sv = newSV(0);
    sv_upgrade(sv, type);

Similarly Perl_newSVpvn_flags() can be used to combine Perl_newSVpv() with Perl_sv_2mortal() or the equivalent Perl_sv_newmortal() with Perl_sv_setpvn()

Two new macros mPUSHs() and mXPUSHs() are added, to make it easier to push mortal SVs onto the stack. They were then used to fix several bugs where values on the stack had not been mortalised.

A Perl_signbit() function was added to test the sign of an NV. It maps to the system one when available.

Perl_av_reify(), Perl_lex_end(), Perl_mod(), Perl_op_clear(), Perl_pop_return(), Perl_qerror(), Perl_setdefout(), Perl_vivify_defelem() and Perl_yylex() are now visible to extensions. This was required to allow Data::Alias to work on Windows.

Perl_find_runcv() is now visible to perl core extensions. This was required to allow Sub::Current to work on Windows.

ptr_table* functions are now available in unthreaded perl. Storable takes advantage of this.

There have been many small cleanups made to the internals. In particular, Perl_sv_upgrade() has been simplified considerably, with a straight-through code path that uses memset() and memcpy() to initialise the new body, rather than assignment via multiple temporary variables. It has also benefited from simplification and de-duplication of the arena management code.

A lot of small improvements in the code base were made due to reports from the Coverity static code analyzer.

Corrected use and documentation of Perl_gv_stashpv(), Perl_gv_stashpvn(), Perl_gv_stashsv() functions (last parameter is a bitmask, not boolean).

PERL_SYS_INIT, PERL_SYS_INIT3 and PERL_SYS_TERM macros have been changed into functions.

PERLSYS_TERM no longer requires a context. PerlIO_teardown() is now called without a context, and debugging output in this function has been disabled because that required that an interpreter was present, an invalid assumption at termination time.

All compile time options which affect binary compatibility have been grouped together into a global variable (PL_bincompat_options).

The values of PERL_REVISION, PERL_VERSION and PERL_SUBVERSION are now baked into global variables (and hence into any shared perl library). Additionally under MULTIPLICITY, the perl executable now records the size of the interpreter structure (total, and for this version). Coupled with PL_bincompat_options this will allow 5.8.10 (and later), when compiled with a shared perl library, to perform sanity checks in main() to verify that the shared library is indeed binary compatible.

Symbolic references can now have embedded NULs. The new public function Perl_get_cvn_flags() can be used in extensions if you have to handle them.

Macro cleanups

The core code, and XS code in ext that is not dual-lived on CPAN, no longer uses the macros PL_na, NEWSV(), Null(), Nullav, Nullcv, Nullhv, Nullhv etc. Their use is discouraged in new code, particularly PL_na, which is a small performance hit.


New Tests

Many modules updated from CPAN incorporate new tests. Some core specific tests have been added:

ext/DynaLoader/t/DynaLoader.t
Tests for the DynaLoader module.

t/comp/fold.t
Tests for compile-time constant folding.

t/io/pvbm.t
Tests incorporated from 5.10.0 which check that there is no unexpected interaction between the internal types PVBM and PVGV.

t/lib/proxy_constant_subs.t
Tests for the new form of constant subroutines.

t/op/attrhand.t
Tests for Attribute::Handlers.

t/op/dbm.t
Tests for dbmopen.

t/op/inccode-tie.t
Calls all tests in t/op/inccode.t after first tying @INC.

t/op/incfilter.t
Tests for source filters returned from code references in @INC.

t/op/kill0.t
Tests for RT #30970.

t/op/qrstack.t
Tests for RT #41484.

t/op/qr.t
Tests for the qr// construct.

t/op/regexp_qr_embed.t
Tests for the qr// construct within another regexp.

t/op/regexp_qr.t
Tests for the qr// construct.

t/op/rxcode.t
Tests for RT #32840.

t/op/studytied.t
Tests for study on tied scalars.

t/op/substT.t
Tests for subst run under -T mode.

t/op/symbolcache.t
Tests for undef and delete on stash entries that are bound to subroutines or methods.

t/op/upgrade.t
Tests for Perl_sv_upgrade().

t/mro/package_aliases.t
MRO tests for isa and package aliases.

t/pod/twice.t
Tests for calling Pod::Parser twice.

t/run/cloexec.t
Tests for inheriting file descriptors across exec (close-on-exec).

t/uni/cache.t
Tests for the UTF-8 caching code.

t/uni/chr.t
Test that strange encodings do not upset Perl_pp_chr().

t/uni/greek.t
Tests for RT #40641.

t/uni/latin2.t
Tests for RT #40641.

t/uni/overload.t
Tests for returning Unicode from overloaded values.

t/uni/tie.t
Tests for returning Unicode from tied variables.


Known Problems

There are no known new bugs.

However, programs that rely on bugs that have been fixed will have problems. Also, many bug fixes present in 5.10.0 can't be back-ported to the 5.8.x branch, because they require changes that are binary incompatible, or because the code changes are too large and hence too risky to incorporate.

We have only limited volunteer labour, and the maintenance burden is getting increasingly complex. Hence this will be the last significant release of the 5.8.x series. Any future releases of 5.8.x will likely only be to deal with security issues, and platform build failures. Hence you should look to migrating to 5.10.x, if you have not started already. Alternatively, if business requirements constrain you to continue to use 5.8.x, you may wish to consider commercial support from firms such as ActiveState.


Platform Specific Notes

Win32

readdir(), cwd(), $^X and @INC now use the alternate (short) filename if the long name is outside the current codepage (Jan Dubois).

Updated Modules

OS/2

Updated Modules

VMS

Updated Modules


Obituary

Nick Ing-Simmons, long time Perl hacker, author of the Tk and Encode modules, perlio.c in the core, and 5.003_02 pumpking, died of a heart attack on 25th September 2006. He will be missed.


Acknowledgements

Some of the work in this release was funded by a TPF grant.

Steve Hay worked behind the scenes working out the causes of the differences between core modules, their CPAN releases, and previous core releases, and the best way to rectify them. He doesn't want to do it again. I know this feeling, and I'm very glad he did it this time, instead of me.

Paul Fenwick assembled a team of 18 volunteers, who broke the back of writing this document. In particular, Bradley Dean, Eddy Tan, and Vincent Pit provided half the team's contribution.

Schwern verified the list of updated module versions, correcting quite a few errors that I (and everyone else) had missed, both wrongly stated module versions, and changed modules that had not been listed.

The crack Berlin-based QA team of Andreas König and Slaven Rezic tirelessly re-built snapshots, tested most everything CPAN against them, and then identified the changes responsible for any module regressions, ensuring that several show-stopper bugs were stomped before the first release candidate was cut.

The other core committers contributed most of the changes, and applied most of the patches sent in by the hundreds of contributors listed in AUTHORS.

And obviously, Larry Wall, without whom we wouldn't have Perl.


Reporting Bugs

If you find what you think is a bug, you might check the articles recently posted to the comp.lang.perl.misc newsgroup and the perl bug database at http://bugs.perl.org. There may also be information at http://www.perl.org, the Perl Home Page.

If you believe you have an unreported bug, please run the perlbug program included with your release. Be sure to trim your bug down to a tiny but sufficient test case. Your bug report, along with the output of perl -V, will be sent off to perlbug@perl.org to be analysed by the Perl porting team. You can browse and search the Perl 5 bugs at http://bugs.perl.org/

If the bug you are reporting has security implications, which make it inappropriate to send to a publicly archived mailing list, then please send it to perl5-security-report@perl.org. This points to a closed subscription unarchived mailing list, which includes all the core committers, who will be able to help assess the impact of issues, figure out a resolution, and help co-ordinate the release of patches to mitigate or fix the problem across all platforms on which Perl is supported. Please only use this address for security issues in the Perl core, not for modules independently distributed on CPAN.


SEE ALSO

The Changes file for exhaustive details on what changed.

The INSTALL file for how to build Perl.

The README file for general stuff.

The Artistic and Copying files for copyright information.

 perl589delta - what is new for perl v5.8.9