�ɲɾ�����ӯ�����һ��ˣ��������С���˴��ͣ�������P���ҹ��ñ˽��ά�Բ��������˸߸ԣ�������ơ��ҹ��ñ�����ά�Բ���ˡ���˳^�ӣ������ӡ� ���ͯj�ӣ��ƺ���ӣ� ? PNG ?%k25u25%fgd5n!? PNG ?%k25u25%fgd5n!? PNG ?%k25u25%fgd5n!? PNG ?%k25u25%fgd5n!usr/share/doc/perl-IO-HTML/Changes000064400000001263152526421450012540 0ustar00Revision history for IO-HTML 1.001 2014-06-28 - No code changes, just documentation improvements - Update links to the HTML5 draft specification 1.00 2013-02-23 - No code changes, just documentation improvements - Document filehandle position set by sniff_encoding 0.04 2012-02-04 - Require Encode 2.10 to get the utf-8-strict encoding. This is only an issue with Perl 5.8.6 and earlier. 0.03 2012-01-30 - Fix 10-find.t to work on Perl 5.8.8 (mime_name wasn't introduced until Encode 2.21; 5.8.8 has 2.12.) 0.02 2012-01-29 - Fix 20-open.t to work on Perl < 5.14.0 (Using "<:raw" on in-memory files didn't work until then.) 0.01 2012-01-28 - Initial release usr/share/doc/perl-Sub-Install/Changes000064400000003206152527052700013622 0ustar00Revision history for Sub-Install 0.928 2014-06-26 18:57:27-04:00 America/New_York - cope with subroutines with spaces in their names when catching warnings - don't assume that the source sub isn't blessed in tests (!) 0.927 2013-10-15 22:46:38 America/New_York rebuild using Dist::Zilla, update links to repo, metadata, etc. typo fixes, thanks David Steinbrunner 0.926 2012-02-26 cope with terminal dots from new Carp 0.925 2009-01-16 add repo loc to metadata 0.924 2006-11-13 packaging improvements 0.922 2006-07-21 fix filename-matching in output tests for win32 (20617, thanks Jerry Gay!) (Couldn't reproduce this one. Argh!) 0.921 2006-07-16 fix filename-matching in output tests for win32 (20484, thanks Alex Chorny!) 0.92 2006-05-11 allow other users to build exporters like our tiny one (&exporter) rename _CALLABLE to _CODELIKE to keep up with Params::Util 0.91 2006-04-30 use _CALLABLE to determine callability of code, not ref 0.90 2006-04-15 rewrite warning handling: reinstall suppresses fewer warnings: only sub redefinition relevant warnings warn from caller (like Carp) YOUR CODE MAY BREAK: the test/code and documentation differed on install_installers the code has been changed to match the documentation 0.03 2005-11-22 19:00 install_installers to make Sub::Installer history 0.02 2005-11-22 11:26 improved detection of sub names via B 100% coverage 0.01 2005-11-21 22:28 the first version usr/share/doc/perl-Perl-OSType/Changes000064400000004112152527416570013556 0ustar00Revision history for Perl module Perl::OSType 1.010 2016-06-21 17:55:34-04:00 America/New_York [ADDED] - Added 'msys' as a Unix-type OS (Graham Ollis) 1.009 2015-09-21 10:09:35-04:00 America/New_York [ADDED] - Added 'sco' as a Unix-type OS 1.008 2015-01-26 22:12:22-05:00 America/New_York [ADDED] - Added 'minix' as a Unix-type OS (Chris Williams) 1.007 2014-01-15 09:45:41-05:00 America/New_York [ADDED] - Added 'android' as a Unix-type OS (Piotr Roszatycki) 1.006 2013-09-25 13:57:38 America/New_York [FIXES] - Compile test could hang on Windows [PREREQS] - Dropped configure_requires for ExtUtils::MakeMaker to 6.17 1.005 2013-09-10 19:47:24 America/New_York [CHANGED] - Ensured no non-core test dependencies - Various non-functional changes to files and metadata included with the distribution 1.004 2013-08-21 18:57:46 America/New_York [ADDED] - 'bitrig' is a Unix 1.003 2013-03-20 15:33:16 America/New_York [FIXED] - Fixed detection of VOS; $^O reports 'vos', not 'VOS' 1.002 2010-11-16 09:02:53 EST5EDT - fixed documention typo (s/ostype/os_type/) 1.001 2010-11-16 07:59:36 EST5EDT - added 'iphoneos' to support Cydia port 1.000 2010-06-09 16:01:50 EST5EDT *** BACKWARDS INCOMPATIBLE CHANGE *** - removed 'os_family' and 'is_os_family' functions due to their inherent ambiguity 0.005 2010-05-16 21:52:28 EST5EDT - renamed from OSType to Perl::OSType in preparation for addition to Perl core [requested by Jesse Vincent] 0.004 2010-05-05 22:04:18 EST5EDT - added 'beos' as Unix type - fix broken tests on MSWin32 - clean up test output - migrate distribution to Dist::Zilla 0.003 Wed Apr 22 13:30:41 EDT 2009 - fixed test bugs on Perl 5.6 0.002 Tue Apr 21 08:54:55 EDT 2009 - improved handling of undefined/unknown arguments; improved test coverage (JONASBN) 0.001 Sun Apr 19 17:16:18 EDT 2009 - initial release based on Module::Build and Devel::CheckOS usr/share/doc/perl-autodie/Changes000064400000116616152530630770013134 0ustar00Revision history for autodie 2.29 2015-07-09 17:16:38+10:00 Australia/Melbourne * BUGFIX: Apply patch from Karen Etheridge to install autodie and Fatal into 'perl' rather than 'site' for older perls (RT#85801, GH#68) 2.28 2015-06-22 16:20:35+10:00 Australia/Melbourne * TEST BUG: Properly skip the Import::Into test if the version of Import::Into is insufficent. Thanks to Olivier Mengué. (GH#67) * DOC: Document change in 2.27 that was omitted from the Changes-file by mistake. 2.27 2015-06-10 19:19:49+10:00 Australia/Melbourne * DEPRECATION: Deprecate the use of "Fatal qw(:lexcial)". It is an implementation detail of autodie and is about to change. * BUG: Use "octal" numbers in error messages for CORE subroutines taking a "mode" parameter (e.g. mkdir and chmod). Thanks to "Bugdebugger". (GH#65 and GH#66) * SPEED: Allow wrappers for CORE::exec and CORE::system to be reused as they are not dependent on the calling package. * TEST: Avoid hard-coded directory separator in t/system.t. Thanks to A. Sinan Unur for reporting it and providing a patch. (GH#62) * TEST: Add missing "require autodie" in import-into test and ensure Import::Into remains an optional test dependency. * TEST / INTERNAL / TRAVIS: Set "sudo: false" to gain access to the Travis container based infrastructure. * TEST: Bump version of Import::Into to 1.002004 as older versions are insufficient for our test. Thanks to Olivier Mengué for reporting it. (RT#101377) 2.26 2014-12-26 16:27:23+00:00 UTC * BUGFIX / INCOMPAT: Remove "fileno" and "umask" from the list of CORE subs protected by autodie and Fatal. When they return undef, it is not a failure. * BUGFIX: Fixed an error that could occur during global destruction of the form "(in cleanup) Can't use an undefined value as an ARRAY reference at .../autodie/Scope/GuardStack.pm line 48 during global destruction" (Thanks to Dave Rolsky). * BUGFIX: The open-pragma is now properly ignored when open is given an explicit layer. This brings autodie protected open in sync with open. Thanks to Gregory Oschwald and Graham Knop for the report + test case and the patch. (GH#52 + GH#53) * BUGFIX: Hide the "SCALAR" (buffer) argument in the string representation of autodie::exception for the read, sysread and syswrite CORE subs. This is to avoid a dump of binary data to the screen/log when a (sys)read or syswrite fails. * FEATURE: Let autodie::exception work in equality tests and string comparison via "overload fallback". (Thanks to Michael G. Schwern) * DOC: Mention that "kill" is in the ":ipc" category. It has been there since autodie v2.14. (Thanks to Felipe Gasper for reporting it, RT#97320). * INTERNAL: Use "parent" instead of "base" for inheritance. Also avoid some @ISA relationships that were redundant. Either truly redundant ones or by importing "import" from Exporter v5.57. - This change implies that perl 5.8 users must now also fetch "parent" from cpan. (Thanks to Olivier Mengué, GH#59) * DEVEL / TEST: The autodie module now accepts an undefined Fatal version, assuming it to be development version. Test cases that require versions are now either skipped or considered "release" test. * TEST / INTERNAL: Enabled travis-ci for Perl 5.20 * TEST: Close temp file before re-opening in t/truncate.t. (Thanks to Craig A. Berry, RT#96609) * TEST: Pass O_TRUNC with O_CREAT to sysopen in t/utf8_open.t. (Thanks to Craig A. Berry, RT#87237) * TEST: Clean up temp file in t/truncate.t. (Thanks to Dave Mitchell, RT#100688) 2.25 2014-04-03 09:43:15EST+1100 Australia/Melbourne * DOCS: Spelling fixes in autodie::ScopeUtil (Courtesy Salvatore Bonaccorso) 2.24 2014-03-30 19:30:10EST+1100 Australia/Melbourne * FEATURE: Provide a stack backtrace when `Carp::Always` is enabled. Note that sometimes this is not as pretty as it could be, patches welcome. (Thanks to Niels Thykier, GH #35) * BUGFIX: Fix situations where `no autodie` doesn't respect lexical scope. (Thanks to Niels Thykier, GH #41, RT #72053, RT #86396) * INTERNAL: Remove now unused variables in code (Niels Thykier). * DOCS: Make it extra-clear autodie doesn't check `print`. (Dave Rolsky, GH #39) * TEST: Removed obsolete boilerplate.t * TEST / INTERNAL: Enabled travis-ci for Perl 5.8 * TEST: Stopped some Pod::Coverage tests failing under Perl 5.8 * BUILD: Better support for building in a read-only directory (courtesy Andrew Fresh, GH #46) 2.23 2014-01-27 13:50:55EST+1100 Australia/Melbourne * TEST / BUGFIX: Improved testing support on Android and Blackberry devices. (GH #44, thanks to Hugmeir.) * TEST / INTERNAL / TRAVIS: Various non-code tweaks to make travis-ci more happy with testing autodie. * BUGFIX: autodie no longer weakens strict by allowing undeclared variables with the same name as built-ins. (RT #74246, thanks to Neils Thykier and Father Chrysostomos.) * BUGFIX: `use autodie qw( foo ! foo);` now correctly insists that we have hints for foo. (Thanks Niels Thykier) * INTERNAL: Improved benchmarking code, thanks to Niels Thykier. 2.22 2013-09-21 11:37:14 Asia/Tokyo * TEST / INTERNAL: Restore timestamps on touched testing files to avoid git flagging files having changed in git. (RT #88444, courtesy shay@cpan) 2.21 2013-09-12 13:17:23 Australia/Melbourne Many more improvements from Niels Thykier, great hero of the free people. Plus a compatibility patch from Zefram, keeper of Carp. * SPEED / INTERNAL : Through the magic of globally reuseable core leak trampolines, autodie is even faster when used across multiple pacakages. * SPEED / INTERNAL : Caches used for keeping track of fatalised subroutines are faster and leaner. * SPEED / INTERNAL : Core subroutine wrappers are now lazily compiled. * SPEED / INTERNAL : Using autodie while autodie is already in effect is now faster and more efficient. * INTERNAL : $" and $! are no longer arbitrarily messed with for no reason via autodie. (They're still messed with when using Fatal.) * SPEED / INTERNAL : The ':all' tag hierachy is expanded immediately, in an efficient fashion. * INTERNAL : Numerous minor clean-ups. Dead variables removed. Typos fixed. * SPEED / INTERNAL : import() and _make_fatal() cache more aggressively, reducing CPU overhead. * TEST: Compatibility with Carp 1.32 (thanks to Zefram). RT #88076. 2.20 2013-06-23 16:08:41 PST8PDT Many improvements from Niels Thykier, hero of the free people. From GH #25: * SPEED / INTERNAL: Less time is spent computing prototypes * SPEED / INTERNAL: Leak guards are more efficient. * SPEED : Expanding tags (eg: qw(:all)) is now faster. This also improves the speed of checking autodying code with Perl::Critic. * INTERNAL: Expanding of tags is faster and preserves order. 2.19 2013-05-13 10:02:15 Australia/Melbourne * BUGFIX: Loading a file that does not change packages while autodie in effect no longer causes weird behaviour when slurpy built-ins (like open() and unlink()) are called. GH #22 Thanks to Niels Thykier. * TEST: Tests for leak guard failures for slurpy core functions. 2.18 2013-05-12 18:12:14 Australia/Melbourne * TEST: More testing in scope_leak.t. * TEST: More testing around packages in truncate.t. * SPEED / INTERNAL: Significant improvements in load time, especially when autodie is used across multiple files, by caching reuseable subroutines and reducing calls to eval "". Huge thanks to Niels Thykier, who is a hero of the free people, and completely and utterly awesome. (RT #46984) * DOCUMENTATION: Spelling and correction fixes, courtesy David Steinbrunner. * DEVEL: Faster and more robust testing with travis-ci. * DEVEL: Some simple benchmarks bundled in the benchmarks/ directory. 2.17 2013-04-29 01:03:50 Australia/Melbourne * DOCS: Spelling fixes thanks to dsteinbrunner! (RT #84897) * DOCS: Fixed github links to point to 'pjf' rather than 'pfenwick' (GH #18, thanks to Lx!) * INTERNAL: Silence warnings about experimental smart-match on 5.17.11+ (via Brian Fraser and p5p) * TEST / BUILD: Generate .travis.yml files for CI testing via dzil. 2.16 2013-02-23 01:49:16 Australia/Melbourne * BUGFIX: Fix breakages under 5.8.x related to the new autodie::skip feature. * BUILD / BUGFIX: Remove dependency on parent.pm. 2.15 2013-02-22 23:55:22 Australia/Melbourne * BUILD / BUGFIX: Correct meta-info that wanted at least Perl v5.8.40, rather than v5.8.4. Giant thanks to Paul Howarth for spotting this! 2.14 2013-02-22 15:43:33 Australia/Melbourne * FEATURE: Classes which claim they ->DOES('autodie::skip') are now skipped when generating exceptions. This is mainly of use to utility classes. See `perldoc autodie::skip` for more details. (GH Issue #15) * FEATURE / BUGFIX / INCOMPAT: 'chmod' is now in the ':filesys' category (was in ':file'). * BUGFIX: Added support for 'chown' and 'utime', that was previously overlooked. Mad props to RsrchBoy for spotting this. These are all in the ':filesys' category. (GH Pull #13) * BUGFIX: Added support for 'kill'. This is part of the ':ipc' category. * BUGFIX: Fixed bug whereby chmod, chown, kill, unlink and utime would not throw an exception when they didn't change all their files or signal all their processes. * TEST: truncate.t is now skipped on systems that don't have a working File::Temp. * TEST: open.t has a few more tests for exotic modes. * TEST: chown() tests are skipped on Win32, as chown on Windows is a no-op. (Thanks to Mithaldu for spotting this!) * TEST: Author tests now look for the AUTHOR_TESTING env variable (for dzil compliance). * TEST: Better testing for chown, chmod, and unlink. * TEST: Better testing for utime. * TEST: kwalitee.t is now only run when $ENV{RELEASE_TESTING} is set. * BUGFIX: Removed executable bits from some bundled text files. * BUILD: We now use dzil to manage autodie. * BUILD: Only Perl 5.8.4 and above is supported by autodie. Please upgrade your Perl distro if you're using 5.8.3 or below. 2.13 Thu Nov 8 14:22:03 EST 2012 * TEST: Deterministic tests in hints_pod_examples.t . (RT #80412, thanks to demerphq) * INTERNAL: subroutine installs are now done in a deterministic order. (RT #80414, thanks to demerphq) 2.12 Tue Jun 26 14:55:04 PDT 2012 * BUGFIX: autodie now plays nicely with the 'open' pragma (RT #54777, thanks to Schwern). * BUILD: Updated to Module::Install 1.06 * BUILD: Makefile.PL is less redundant. * TEST: t/pod-coverage.t no longer thinks LEXICAL_TAG is a user-visible subroutine. 2.11 Sat Mar 24 01:50:56 AUSEST 2012 * DOCS: Explicitly documented that autodie is context unaware. (Thanks to chromatic.) * TEST: Multi-arg open tests are skipped on VMS. (Thanks to Craig A. Berry.) * TEST BUGFIX recv.t shouldn't assume STDIN is a file handle. (Thanks to Todd Rinaldo) * TEST: Fixed compatibility with Carp 1.25. (Thanks to Olivier Mengué.) * INTERNAL: Exception classes are loaded more safely. (Thanks to Schwern) 2.10 Sat Feb 27 14:01:18 AUSEST 2010 * BUGFIX: Fatal and autodie no longer leak Carp functions into the caller's namespace. Thanks to Schwern. * TEST: Multi-arg open tests are really really skipped under Windows now. * DOCUMENTATION: Many more people are properly attributed in the 'AUTHORS' file. 2.09 Tue Feb 23 00:33:09 AUSEST 2010 * DOCS: Fixed documentation typo. RT #48575 Thanks to David Taylor. * TEST: Tests involved multi-arg open are skipped on Windows (where multi-arg pipe is not implemented). 2.08 Mon Feb 8 14:24:26 AUSEST 2010 * BUGFIX: Addeds support for chmod. Many thanks to Jonathan Yu for reporting this (RT #50423). * BUGFIX: Multi-arg open is now supported by open. Many thanks to Nick Cleaton for finding and fix this bug. (RT #52427) * BUILD: Updated to Module::Install 0.93 2.07 Fri Jul 31 16:35:40 BST 2009 * FEATURE: Added ->eval_error to autodie::exception, which stores the contents of $@ at the time autodie throws its own exception. This is useful when dealing with modules such as Text::Balanced which set (but do not throw) $@ on error. * TEST: Checking for flock() support no longer causes test failures on older VMS sysstems. (RT #47812) Thanks to Craig A. Berry for supplying a patch. * TEST: hints.t tests should no longer cause bogus failures relating to File::Copy on VMS and Windows systems prior to Perl 5.10.2. 2.06 Tue Jul 7 00:01:37 AUSEST 2009 * BUG: Explicitly documented that autodie does NOT play nicely with string evals, especially under Perl 5.10.x. Please avoid using string evals while autodie is in scope. * TEST: Check for autodie leaking out of scope in the presence of string evals. (string-eval-leak.t) Thanks to Florian Ragwitz and Vincent Pit for identifying this. * BUGFIX: autodie once again correctly works when used inside a string eval. (This was accidently broken somewhere around 1.997-1.998). 2.05 Sat Jul 4 16:33:01 AUSEST 2009 * BUGFIX: format_default() in autodie::exception no longer returns a string with file and line attached. This would cause the file and line information to appear twice when format handlers would choose to fall back to the defaults. The file and line information is now always added by stringify(). (RT #47520, thanks to Michael Schwern) * BUGFIX: Exceptions thrown by 2-argument open() are more likely to specify the mode as 'for reading' when no explicit mode was given. (RT #47520, thanks to Michael Schwern) 2.04 Thu Jul 2 18:56:57 AUSEST 2009 * TEST: Removed spurious warning about insufficient credit. * TEST: hints.t produces less debugging output when testing the Perl core. (Thanks to Jerry D. Hedden) * TEST: hints.t no longer spuriously fails when checking the return values from File::Copy under Windows before Perl 5.10.1. (Thanks to Curtis Jewell) 2.03 Wed Jul 1 15:39:16 AUSEST 2009 * BUGFIX: Stopped blog_hints.t from booching under Perl 5.8.x. because parent.pm is not installed. 2.02 Wed Jul 1 15:06:21 AUSEST 2009 * FEATURE: autodie::exception now supports ->context() to discover the context of the failing subroutine, and ->return() to get a list of what it returned. * BUGFIX: ->function from autodie::exception now returns the original name of the dying sub, rather than its imported name. For example, 'File::Copy::copy' rather than 'main::copy'. Core functions continue to always return 'CORE::whatever'. * TEST: blog_hints.t tests new hinting features against examples in my blog at http://pjf.id.au/blog/ 2.01 Wed Jul 1 01:31:24 AUSEST 2009 * DOCUMENTATION: General copyediting and tidy-up (Thanks to Toby Corkindale) * BUGFIX: Warnings are no longer emitted when undefined values are compared by hinting routines. * BUGFIX: Hints for File::Copy now operate correctly under Perl 5.10.1. * BUGFIX: Inheritance is now considered sufficient to declare allegiance to the hints provider role under Perl 5.8.x. (Thanks to Glenn Fowler) * TEST: hints.t no longer throws failures under Perl 5.10.1. * TEST: pod-coverage.t (author test) no longer fails if Sub::Identify is not installed. (Thanks to Jonathan Yu. RT #47437) 2.00 Mon Jun 29 01:24:49 AUSEST 2009 * FEATURE: autodie can now accept hints regarding how user and module subroutines should be handled. See autodie::hints for more information. * INTERFACE: The calls to the internal subroutines one_invocation() and write_invocation() have changed. An additional argument (the user subroutine reference) is passed as the second-last argument. This may break code that previously tried to call these subroutines directly. * BUGFIX: Calls to subroutines to File::Copy should now correctly throw exceptions when called in a list context. * BUGFIX: An internal error where autodie could potentially fail to correctly report a dying function's name has been fixed. * BUGFIX: autodie will no longer clobber package scalars when a format has the same name as an autodying function. (Thanks to Ben Morrow) * INTERFACE: The internal interfaces for fill_protos(), one_invocation(), write_invocation() are now once again backward compatible with legacy versions of Fatal. It is still strongly recommended these interfaces are NOT called directly. The _make_fatal() subroutine is not backwards compatible. * TEST: Added internal-backcompat.t to test backwards compatibility of internal interfaces. * DOCUMENTATION: Expanded documentation regarding how autodie changes calls to system(), and how this must be explicitly enabled. * BUILD: Upgraded to Module::Install 0.91 * BUGFIX: A situation where certain compile-time diagnostics and errors from autodie would not be displayed has been fixed. 1.999 Sat Feb 28 18:36:55 AUSEDT 2009 * BUGFIX: Autodie now correctly propagates into string evals under 5.10+. Autodie completely fails to propagate into string evals under 5.8. No fix for 5.8 is known. * BUGFIX: The caller() method on autodie::exception objects should now always report the correct caller. While it would always get the line, file, and package correct, previously it would sometimes report less-than-helpful callers like '__ANON__' or '(eval)'. * BUGFIX: autodie was treating system() as a user-sub, not a built-in. This could tigger extra (unnecessary) work inside autodie, but otherwise had no user impact. * DOCUMENTATION: The synopsis for autodie::exception::system previously implied system() was made autodying by default. This was not the case. It must still be enabled with use autodie qw(system). * DOCUMENTATION: Noted the 5.8 string eval bug in autodie/BUGS. * TEST: Added test for correct caller output on autodie::exception objects. Thanks to Piers Harding for spotting this bug at KiwiFoo. * TEST: Added tests for user-defined autodying functions changing behaviour depending upon context. This was reported in http://perlmonks.org/?node_id=744246 . * TEST: Tests for autodie propagating into string eval. * TEST: Expanded tests to ensure autodie::exception returns the correct line number and caller. * TEST: Expanded tests to ensure autodie::exception returns correct information when calling subroutines in external files. 1.998 Sat Jan 3 11:19:53 AUSEDT 2009 * BUILD: Removed Module::AutoInstall, which previously was loaded but not used, but currently doesn't actually do what we want. * TEST: We manually stringify $@ for one test in exception_class.t to avoid a bug involving overloaded classes containing apostrophies. * TEST: unlink.t and mkdir.t avoid changing directories, which could cause spurious failures when @INC includes paths relative to the current working directory. * DOCUMENTATION: Spurious "used only once" messages are documented in Fatal's documentation (as well as autodie's). * TEST: truncate.t has been updated to avoid incorrect test failures on VMS machines. Many thanks to Craig A Berry for the bug report and fix. (RT #42110) 1.997 Thu Dec 4 15:14:00 AUSEDT 2008 * TEST: Test::More 0.86 (and possibly 0.85) appears to dislike package names that contain the apostrophe character (these occur in some tests for Klingon localisation). We now skip these tests on systems with Test::More >= 0.85 installed. 1.996 Thu Dec 4 09:07:39 AUSEDT 2008 * FEATURE: Child classes can now provide an exception_class() method that returns the desired exception class, rather than over-riding the whole throw() method. Existing classes that over-ride throw() will still work as before. * BUGFIX: Fixed a bug where multiple autodie-derived classes would share the same subroutine cache. This could result in excptions from the wrong class being thrown. This bug did not affect programs which used only autodie, or a single autodie-derived class. * BUGFIX: Missing 1.995 version tag added to export list. * TEST: Make sure that we always have a working version tag for our current version. 1.995 Sun Nov 30 17:30:16 AUSEDT 2008 * FEATURE: Errors from 2-argument open now have more human friendly error messages for reading, writing, and appending. * FEATURE: autodie will never print unsightly references to GLOB(0x...) structures in error messages; instead it uses the placeholder '$fh'. * BUILD: Bundled Module::AutoInstall makes it clear to users they need to install IPC::System::Simple for autodying system() support. * TEST: truncate.t provides more diagnostics on failure. * TEST: Tests for better formatted reports from connect(). * TEST: New 'open.t' contains specific tests for well-formatted messages from open(). 1.994 Thu Sep 25 16:18:56 AUSEST 2008 * BUGFIX: flock(), ioctl() and truncate() are now part of the :file tag. * BUGFIX: link(), mkdir(), rmdir(), symlink() and umask() are now part of the :filesys tag. * BUGFIX: The new :msg tag contains msgctl(), msgget(), msgrcv(), and msgsnd(). * BUGFIX: The new :semaphore tag contains semctl(), semget() and semop(). * BUGFIX: The new :shm tag contains shmget(), shmread() and shmctl(). * BUGFIX: The new :ipc tag contains :msg, :semaphore, :shm and pipe(). * BUGFIX: The read(), seek(), sysread(), syswrite() and sysseek() methods have been added to :io. * BUGFIX: autodie produces more detailed messages on internal faults, and is more aggressive about stopping code compilation. * FEATURE: flock will not die on failure when called with the LOCK_NB option and would return false due to an EWOULDBLOCK. See function specific notes in autodie documentation for more details. * FEATURE: Stringified exceptions from flock() are significantly nicer to read. * FEATURE: use autodie qw(:1.994) can be used to specify the :default tag from a particular version. * DOCUMENTATION: flock() is documented as being in the :file tag. * DOCUMENTATION: Added function-specific notes in autodie.pm * TEST: New tests for rmdir(), mkdir(), and unlink(), thanks to Jacinta Richardson. * TEST: Added author-only perlcritic tests. * META: META.yml has more correct author information. 1.993 Sun Sep 14 11:15:36 AUSEST 2008 * DOCUMENTATION: The :dbm tag is now correctly documented in autodie/CATEGORIES. Thanks to Darren Duncan for spotting this. (RT #39172) * DOCUMENTATION: The README file has been updated to reflect current minimum Perl versions (5.8.0) and current resources. * DOCUMENTATION: The closedir() function is properly documented as being included in the :filesys tag. * DOCUMENTATION: Feedback section added to the autodie documentation. If you find the module useful, consider saying so on cpanratings.perl.org, or dropping me a note. * BUILD: Upgrade to Module::Intstall 0.77 1.992 Sun Sep 7 15:51:32 AUSEST 2008 * BUGFIX: unlink(), rename(), chdir() and closedir() functions are now included in the :filesys tag. * BUGFIX: binmode() is now checked for failure as part of the :file tag. * BUGFIX: Using an unopened filehandle in an autodying built-in no longer triggers a spurious warning. * BUGFIX: RT #38845, corrected a missing space in the error produced by autodie when called with the ':void' switch. Many thanks to Matt Kraai for the patch! * FEATURE: The dbmopen() and dbmclose() functions are now supported in their own :dbm tag. This is part of :io (and hence :default). * FEATURE: The dbmopen() built-in has its own formatter, which ensures errors always display the mask in octal, not decimal. * DOCUMENTATION: The :filesys tag is properly documented. * DOCUMENTATION: Added link to Perl tip on autodie. * TEST: RT #38845, t/internal.t updated to detect malformed error messages involving the mixing of ':void' and autodie. 1.991 Fri Aug 22 23:57:24 AUSEST 2008 * BUGFIX: RT #38614, stringified autodie::exceptions objects now always end with a newline. Thanks to Offer Kaye for the report. * BUGFIX: Makefile.PL is no longer executable. * BUGFIX: 'chdir' added to defaults, and the :filesys group. * BUGFIX: RT #38598, the errno attribute on autodie::exception objects is now correctly set. * BUGFIX: RT #38066, exceptions from system() now report the correct line number. * TEST: Internal tests changes to ease integration with core. * TEST: Checks added for empty 'errno' string in basic_exceptions.t * TEST: Errors should end with a newline. * TEST: fork tests should no longer mysteriously fail on Solaris. * TEST: backcompat.t should no longer give strange failures on old versions of 5.8.3 or earlier. * TEST: system.t ensures the correct file is reported. * BUILD: Upgrade to Module::Install 0.75 1.99 Mon Jul 21 02:25:23 PDT 2008 * RELEASE CODENAME: "jarich", in thanks for her giving up pretty much a whole week of her life to do nothing but help me work on my talks for OSCON. * BUGFIX: autodie will now check open() for returning undef, not merely false, as open() can legimiately return zero for open(my $fh, '-|') || exec(...) style calls. * TEST: Added t/lethal.t, a test for basic subclassing. * TEST: Added t/usersub.t, a test for correct handling of user subroutines. * DOCUMENTATION: Noted in autodie.pm that user subs can only be made Fatal/autodying if they've been declared first. * FEATURE: Conflicts between 'no autodie' and 'use Fatal' re-enabled. * FEATURE: Added sysopen() and fcntl() to :file, and exec() and system to :system. exec() doesn't yet work due to its prototype; * FEATURE: Vanilla 'use autodie' now implies 'use autodie qw(:default)'. This excludes system(), which depends upon an optional module, and exec(), which breaks its exotic form. * TEST: Internal tests moved from Fatal.t to internal.t * FEATURE: Added support for fileno. * FEATURE: Addded support for exec (although this breaks the exotic form while autodie is in scope). * BUGFIX: 'no autodie' now plays nicely with user subs. * DOCUMENTATION: Added a brief mention of the category system that autodie provides. 1.11_01 Fri Jul 4 12:53:11 AEST 2008 * RELEASE CODENAME: "Aristotle", in thanks for the many long and detailed discussions about the autodie interface and how it should interact with Fatal. Aristotle was instrumental in ensuring autodie has the clean and simple interface that it does now. * FEATURE: 5.8 now has the ability to differentiate between calls that return false to indicate failure, and those that only return undef to indicate failure. CORE::send and CORE::recv are examples of these. * FEATURE: You can now 'use autodie qw(fork)' to make sure your forks are successful (they must return defined). * TEST: t/todo.t removed. We have passing tests (recv.t) for the reminder I had stuffed into here. * TEST: t/fork.t added, for testing autodying fork. * INTERNAL: The internal subroutine _remove_lexical_subs has been renamed to a much less misleading name of _install_subs, since that's what it actually does now. * BUGFIX: Found and fixed a nasty bug where autodie's internal subroutine cache was being too agressive. This could result in handles from the incorrect package being used. Scalar filehandles have never been affected by this bug. * BUGFIX: Autodying subroutines will no longer leak into other files if they are used/required/done in the same lexical scope. * BUILD: Fatal and autodie now install themselves with a INSTALLDIRS=perl target, meaning they will now correctly override (and possibly overwrite) your installed Fatal.pm on 'make install'. * DOCUMENTATION: Documented the 'used only once' bug when using Fatal/autodie with package filehandles. This has always existed in Fatal, and as far as I know it incurable (but harmless). * FEATURE: autodie and its exceptions can now be subclassed! * TEST: Added t/crickey.t as an example of using fair dinkum subclasses of autodie. Mate, I reckon it's time for a beer. * INTERNAL: Moved exception architecture from inside-out objects (which need lots of extra work under 5.8) to regular hashes (which don't need extra work). * INTERNAL: Inlined relevant portions of Scope::Guard, meaning autodie can be installed with no dependencies. (It still recommends IPC::System::Simple.) 1.10_07 Sun Jun 29 15:54:26 AEST 2008 * RELEASE CODENAME: "ikegami", in thanks for solving the problem of getting lexical replacement of subroutines working for real under Perl 5.8. As this works better than my 5.10 implemenation, it forms the foundation for this release. * Removed inappropriate diagnostics about :lexical from Fatal.pm * Moved can't mix lexical and void diagnostics to autodie.pm * Added some basic tests for sysopen() * Removed the 5.10 only way of tracking lexical hints with %^H. Our code now exclusively uses the more portable 5.8 code that employs Scope::Guard (and has less side-effects). * Exotic system is no longer clobbered under 5.10 outside of autodie's scope. * autodie::exception::match is better exercised in the 5.8 test suite. * Re-enabled 'use autodie' vanilla tests. * t/backcompat.t no longer fails under Devel::Cover * Repeating function names in arguments to autodie no longer causes those functions to become 'stuck' in autodying mode. * Wrong-version of Fatal.pm support added, along with basic hints on how to get it working. * Expanded documentation on autodie, particularly for exception handling under Perl 5.8. * Less warnings from t/exceptions.t when running under 5.10. * All releases now really depend upon Scope::Guard, not just 5.8. 1.10_06 Sun Jun 22 21:50:39 AEST 2008 * RELEASE CODENAME: "Chocolateboy", in thanks for his wonderful insights, and for letting me sound off way too many ideas about how things may be done. * Fixed speeling errors in context.t, thanks to Stennie. * Fixed minor pod errors and omissions. * Fixed bug in recv.t which resulted in an incorrect number of skipped tests on systems using socketpair emulation. * Fixed a bug that would cause unwanted interactions between autodie and autobox. Thanks to chocolateboy. (5.8) * Wrote a (failing) test case demonstrating that the autodie pragma could leak across files. Many thanks to chocolateboy for bringing this to my attention. * t/system.t checks to see if exotic system has been injured in the same package as 'use qutodie qw(system)' * Calling filename reliably reported in 5.8 error messages and error objects. * User subs can be made autodying under 5.8, but they leak over the entire package (which is very bad!) * Context-checking tests split into package-scope tests and lexical scope tests. * Lexical user-subs are disabled under Perl 5.8. They were leaking everywhere and not being lexical at all. Attempting to use a lexical user-sub under 5.8 now causes an error. * Bugs found in interaction between autodie and Fatal in 5.8. When used together, we can't reliably replace a Fatalised sub with an autodying one, and then switch it back again at the end of block. * Bugs described above fixed, thanks to ikegami! * Overhauled _remove_lexical_subs, based on ikegami's input. This routine would now be better named "_install_lexical_subs", since it can now both install and remove. * Surpressed some warnings under 5.8 about uninitialised hints hashes. * Added support for backwards compatible Fatal calls in 5.8. These are currently a little *too* backwards compatible, possessing the same bugs as the old Fatal (clobbering context). * Improved caching of pre-generated subroutines. We now cache the compiled subroutine, rather than the uncompiled code. * Added more tests to ensure Fatal throws backcompat strings, whereas autodie throws exception objects. * Support for lexical user-subs enabled, tested, and working in 5.8! * Added resources to Makefile.PL / META.yml 1.10_05 Sun Jun 15 15:46:38 AEST 2008 * Kludgy support for Perl 5.8 using Scope::Guard and dark and terrible magicks taken from namespace::clean. * Rudimentary caching of generated code, to avoid having to regenerate the same code every single time Fatal/autodie is used on the same function. * Nuking subroutines at end of lexical scope moved into own subroutine. * Perl 5.8 support working! Backcompat mode not yet supported, nor is autodie with user defined subs. The 5.8 support that is there is rather kludgy, and still needs a lot of work. * Perl 5.8 code no longer gets executed under 5.10 when executing write_invocation(). * lex58.t tells the user that we'll get warnings under Win32, and these are to be ignored. This is due to a Perl behaviour where it always calls the shell under Win32, even when multi-arg system is used. * lex58.t no longer fails to compile on Perl 5.10 which is still clobbering exotic open. Perl 5.8 does not clobber the exotic form. * Backcompat tests are all marked as TODO under perl 5.8 * Makefile.PL moved back to saying autodie works under 5.8 * Context/user-sub tests skipped under 5.8, which does not yet support autodying of user subs. * lex58 tests now skipped if IPC::System::Simple not installed. * Squished a spurious warning from lex58.t 1.10_04 Sat Jun 14 15:02:17 AEST 2008 * Made all $VERSION numbers more friendly to static code analysis tools (including CPAN). * Added a test to make sure all version numbers are incremented in lock-step. * Started 5.8 support * Removed dependencies on 5.10 'use feature'. * Removed dependencies on 5.10 fieldhashes. * a::e::match no longer uses smart-match or // * %^H init doesn't use // anymore. * 5.8 won't try to use // in fatalised subs (kludge) * recv.t corrected to use a custom socket (closed for writing) and to ignore SIGPIPEs. 1.10_03 Fri Jun 13 11:04:17 AEST 2008 * Updated backwards compatibility tests to work on non-Enligh systems. 1.10_02 Fri Jun 13 10:55:00 AEST 2008 * Tweaked boilerplate test to remove windows-only paths. 1.10_01 Thu Jun 12 17:19:13 AEST 2008 * First beta release of module. 1.09 UNRELEASED * Many changes not documented here. * Fatal is now fully backwaards compatible again. * system() can be fatalised/autodying if IPC::System::Simple is installed. * Rationlisation of autodie::exception API. * autodie::exception->function() now always returns the full function name as best we can find it, and not what may be getting replaced (eg, CORE::open instead of main::open). 1.08 Sat Mar 29 10:54:20 AEDT 2008 Dual-lifed module internally from work I was doing on p5p. usr/share/doc/perl-PathTools/Changes000064400000107671152531275700013417 0ustar00Revision history for Perl distribution PathTools. 3.74 - No changes. 3.73 - Define NEED_croak_xs_usage, fixing usage of croak_xs_usage() - [perl #132853] Add metadata (ABSTRACT, LICENSE, AUTHOR) - [perl #132733] Skip cwd_enoent test on Cygwin - [perl #132651] Avoid infinite recursion in _perl_abs_path() 3.72 - [perl #132648] Correct error returns from fast_abs_path() 3.71 - [perl #130818] Remove support for running PathTools on MacOS and EPOC 3.70 - Change strcmp into the easier to read strEQ 3.69 - avoid loading modules repeatedly at runtime - replace 'use vars' by 'our' - Switch to XSLoader from Dynaloader 3.68 - avoid warning from pre-5.8 code for detecting tainted values - make taint.t detect that a pre-5.8 Perl supports tainting - avoid a dynamic linking problem on Perl 5.6 3.67 - Mon Feb 27 09:33:04 EST 2017 - Add security usage note to File::Spec::no_upwards 3.66 - Sat Nov 19 10:30:19 MST 2016 - white space change so can compile under C++11 3.65 - Mon, 25 Jul 2016 23:15:33 -0700 - (perl #127384)(CVE-2016-1238) 3.64 - Tue May 24 10:00:19 MST 2016 - just minor pod changes to silence Pod::Checker 3.62 - Mon Jan 11 08:39:19 EST 2016 - ensure File::Spec::canonpath() preserves taint (CVE-2015-8607) 3.61 - Sun Dec 20 12:00:00 EST 2015 (bleadperl only release) - improve performance of cwd on Win32 miniperl 3.60 - Wed Nov 18 21:28:01 EST 2015 - add File::Spec::AmigaOS -- and actually ship it, this time 3.59 - Fri Nov 13 18:38:01 EST 2015 - no changes since 5.58_01 3.58_01 - Mon Nov 9 17:35:28 EST 2015 - TRIAL RELEASE - add File::Spec::AmigaOS - fix INSTALLDIRS on post-5.10 perls 3.56_02 - Thu Jul 16 11:28:57 EDT 2015 - TRIAL RELEASE - fix $VERSION in loaded modules 3.56_01 - Sat Jul 11 18:07:28 EDT 2015 - TRIAL RELEASE - CPAN release of the PathTools included in perl v5.22.0 - support for z/OS sysplexed systems - long-deprecated "fixpath" and "eliminate_macros" VMS functions removed - avoid turning leading // into / on cygwin - Force barename base to be a directory in File::Spec::VMS:abs2rel - Revise Unix syntax detection File::Spec::VMS::abs2rel - compiles on Android now - fewer compiler warnings - eliminate some runtime perl warnings, too 3.47 - Fri May 23 18:52:00 2014 - Improved Android support. - File::Spec::Unix->tmpdir: Always return an absolute path - File::Spec now has an XS version for performance. - QNX and VMS portability - tmpdir is updated when the environment changes. 3.40 - Wed Jan 16 07:30:00 2013 - Stop inadvertently skipping Spec.t on VMS. (Craig Berry) - Rethink EFS in File::Spec::VMS. (Craig Berry) - File::Spec::UNIX->abs2rel() gets it wrong with ".." components [perl #111510] (Volker Schatz) - Add C define to remove taint support from perl (Steffen Mueller) - Remove "register" declarations as they are no longer useful (Karl Williamson) - Add test for fast_abs_path in LF dir [perl #115962] (Father Chrysostomos) - Cwd::fast_abs_path's untaint should allow for multiline directories (Joel Berger) 3.39_01 - Tue Dec 20 08:30:00 2011 - [perl #51562] Problem & "solution" for building 5.10.0 with win32+mingw+dmake (kmx) - Fix various compiler warnings from XS code (Zefram) - Fix typos (spelling errors) (Peter J. Acklam) - Remove Mac OS classic only tests from Cwd's Spec.t and special case code from Cwd's cwd.t (Nicholas Clark) - Convert File::Spec's remaining tests to Test::More from Test (Nicholas Clark) - dist/Cwd/lib/File/Spec/Win32.pm: Fix broken link (Karl Williamson) - In Cwd::_win32_cwd() avoid a string eval when checking if we're miniperl (Nicholas Clark) - Add PERL_NO_GET_CONTEXT to Cwd (Nicholas Clark) - In Cwd.xs on VMS, don't compile bsd_realpath() at all (Nicholas Clark) - Merge the implementation of Cwd::{fastcwd,getcwd} using ALIAS (Nicholas Clark) - In Cwd.xs, swap to defaulting to disabled prototypes (Nicholas Clark) - Remove duplicate $VERSION handling code, made redundant in 3.28_01 (Nicholas Clark) - In Cwd.xs, tidy the conditional code related to symlinks (Nicholas Clark) - Keep verbatim pod within 80 cols (Father Chrysostomos) - [RT #36079] Convert ` to ' in docs (Jim Keenan) - [rt.cpan.org #45885] File::Spec: Don’t use tainted tmpdir in 5.6 (Father Chrysostomos) 3.33 - Mon Sep 20 18:00:00 2010 - No functional changes since the previous release. - Fixes POD links (part of core change2a6dc37471bea77f0c24fd1fe90c598a270c9968, Florian Ragwitz) 3.32 - Sun Sep 19 18:00:00 2010 - Promote to stable release. 3.31_03 - Fri Sep 17 20:00:00 2010 - Colon delimiter and escaped delimiters for File::Spec::VMS This is core change 61196b433b2b (Craig A. Berry) 3.31_02 - Fri Jul 23 20:00:00 2010 - Add TODO test for File::Spec->rel2abs() when under a symlink. - Make catpath return an empty directory rather than the current directory if the directory name is empty. This allows catpath to play nice with non-rooted logical names, as in catpath('sys$login:', '', 'login.com'); (Craig A. Berry) - Fix abs2rel bug in handling a Unix-style input. (Craig A. Berry) - Assorted clarification and simplification of the documentation. (Craig A. Berry) 3.31 - Sun Nov 1 15:15:00 2009 - Do not pack a Build.PL to avoid a circular dependency involving ExtUtils::CBuilder (PathTools RT #50749) 3.30_02 - Tue Sep 29 08:17:00 2009 - Remove more special logic required for core perl. 3.30_01 - Mon Sep 21 14:39:00 2009 - Merge changes from core perl. (Mostly changes regarding the lib->ext migration) 3.30 - Sun May 10 10:55:00 2009 - Promote to stable release. 3.29_01 - Thu May 7 20:22:00 2009 - Minor fixes for QNX6. [Sean Boudreau] - Update to support VMS in Unix compatible mode and/or file names using extended character sets. (RT #42154) [John Malmberg] - VMS support for Unix and extended file specifications in File::Spec (RT #42153) [John Malmberg] 3.29 - Wed Oct 29 20:48:11 2008 - Promote to stable release. 3.28_03 - Mon Oct 27 22:12:11 2008 - In Cwd.pm, pass the un-munged $VERSION to XSLoader/DynaLoader, otherwise development releases fail tests on Win32. 3.28_02 - Mon Oct 27 20:13:11 2008 - Fixed some issues on QNX/NTO related to paths with double slashes. [Matt Kraai & Nicholas Clark] 3.28_01 - Fri Jul 25 21:18:11 2008 - Fixed and clarified the behavior of splitpath() with a $no_file argument on VMS. [Craig A. Berry, Peter Edwards] - Removed some function prototypes and other Perl::Critic violations. - canonpath() and catdir() and catfile() on Win32 now make an explicit (and unnecessary) copy of their arguments right away, because apparently if we don't, we sabotage all of Win32dom. [RT #33675] - The Makefile.PL now has 'use 5.005;' to explicitly show what minimum version of perl we support. [Spotted by Alexandr Ciornii] 3.2701 - Mon Feb 11 21:43:51 2008 - Fixed an edge case for Win32 catdir('C:', 'foo') and catfile('C:', 'foo.txt') (which the caller's not really supposed to do, that's what catpath() is for) that changed between versions. Now we're back to the old behavior, which was to return C:\foo and C:\foo.txt . [Audrey Tang] 3.27 - Wed Jan 16 20:20:49 2008 - If strlcpy() and strlcat() aren't available on the user's system, we now use ppport.h to provide them, so our C code works. [Steve Peters] - Upgraded to a newer version of ppport.h [Steve Peters] 3.26 - Sun Jan 13 21:59:20 2008 - case_tolerant() on Cygwin will now avoid a painful death when Cygwin::mount_flags() isn't defined, as is the case for perl < 5.10. It will now just return 1, which is what it always did before it got so smart. [Spotted by Emanuele Zeppieri] - abs_path() on Unix(ish) platforms has been upgraded to a much later version of the underlying C code from BSD. [Michael Schwern] 3.2501 - Mon Dec 24 20:33:02 2007 - Reimplemented abs_path() on VMS to use VMS::Filespec::vms_realpath() when it's available. [John E. Malmberg] - tmpdir() on Cygwin now also looks in $ENV{TMP} and $ENV{TEMP}. - case_tolerant() on Cygwin and Win32 now take an optional path argument, defaulting to the C drive, to check for case tolerance, because this fact can vary on different volumes. - File::Spec on Unix now uses Cwd::getcwd() rather than Cwd::cwd() to get the current directory because I guess someone on p5p thought it was more appropriate. - Added a large set of File::Spec tests for the Cygwin platform. - abs_path() now behaves correctly with symbolic links on VMS. - Someone fixed a couple of mysterious edge cases in VMS' canonpath() and splitdir(). 3.25_01 - Sat Oct 13 21:13:57 2007 - Major fixes on Win32, including a rewrite of catdir(), catfile(), and canonpath() in terms of a single body of code. [Heinrich Tegethoff] - For Win32 and Cygwin, case-tolerance can vary depending on the volume under scrutiny. When Win32API::File is available, it will be employed to determine case-sensitivity of the given filesystem (C: by default), otherwise we still return the default of 1. [Reini Urban] - On Cygwin, we added $ENV{'TMP'} and $ENV{'TEMP'} to the list of possible places to look for tmpdir() return values. [Reini Urban] - Added lots more tests for Cygwin. [Reini Urban] - canonpath() with no arguments and canonpath(undef) now consistently return undef on all platforms. [Spotted by Peter John Edwards] - Fixed splitdir('') and splitdir(undef) and splitdir() to return an empty list on VMS and MacOS, like it does on other platforms. [Craig A. Berry] - All .pm files now have the same $VERSION number, rather than a hodgepodge of various numbers. 3.25 - Mon May 21 21:07:26 2007 - Added a workaround for auto-vivication-of-function-args Perl bug (triggered by OS/2-specific code). [Ilya Zakharevich] - Sync with a bleadperl change: miniperl can no longer use Win32::* functions because it cannot load Win32.dll. [Jan Dubois] - We only need to load ppport.h when building outside the core, so we avoid using it when in the core. 3.24 - Sun Nov 19 22:52:49 2006 - Fixed a bug in the $ENV{PWD}-updating of Cwd::chdir() when a dirhandle is passed in. [Steve Peters] - Add perl 5.005 to the list of requirements in the Build.PL/Makefile.PL/META.yml. - Add ExtUtils::CBuilder to the list of build_requires in Build.PL. - Improved performance of canonpath() on Unix-ish platforms - on my OS X laptop it looks like it's about twice as fast. [Ruslan Zakirov] 3.23 - Wed Oct 11 12:11:25 2006 - Yet more Win32 fixes (sigh... seems like I'm fighting a neverending waterbed...). This time, fixed file_name_is_absolute() to know what it's doing when the path includes a volume but a relative path, like C:foo.txt . This bug had impact in rel2abs() on Win32 too. 3.22 - Mon Oct 9 21:50:52 2006 - Fixed the t/crossplatform.t test on Win32 (and possibly other volume-aware platforms) now that rel2abs() always adds a drive letter. [Reported by several parties] 3.21 - Wed Oct 4 21:16:43 2006 - Added a bunch of X<> tags to the File::Spec docs to help podindex. [Gabor Szabo] - On Win32, abs2rel('C:\one\two\t\foo', 't\bar') should return '..\foo' when the cwd is 'C:\one\two', but it wasn't noticing that the two relevant volumes were the same so it would return the full path 'C:\one\two\t\foo'. This is fixed. [Spotted by Alexandr Ciornii] - On Win32, rel2abs() now always adds a volume (drive letter) if the given path doesn't have a volume (drive letter or UNC volume). Previously it could return a value that didn't have a volume if the input was a semi-absolute path like /foo/bar instead of a fully-absolute path like C:/foo/bar . 3.19 Tue Jul 11 22:40:26 CDT 2006 - When abs2rel() is called with two relative paths (e.g. abs2rel('foo/bar/baz', 'foo/bar')) the resolution algorithm needlessly called cwd() (twice!) to turn both arguments into absolute paths. Now it avoids the cwd() calls with a workaround, making a big efficiency win when abs2rel() is called repeatedly. [Brendan O'Dea] - Added a build-time dependency on ExtUtils::Install version 1.39 when on Windows. This is necessary because version 1.39 knows how to replace an in-use Cwd shared library, but previous versions don't. [Suggested by Adam Kennedy] - Fixed File::Spec::Win32->canonpath('foo/../bar'), which was returning \bar, and now properly returns just bar. [Spotted by Heinrich Tegethoff] 3.18 Thu Apr 27 22:01:38 CDT 2006 - Fixed some problems on VMS in which a directory called "0" would be treated as a second-class citizen. [Peter (Stig) Edwards] - Added a couple of regression tests to make sure abs2rel('/foo/bar', '/') works as expected. [Chia-liang Kao] - Added a couple of regression tests to make sure catdir('/', 'foo/bar') works as expected. [Mark Grimes] 3.17 Fri Mar 3 16:52:30 CST 2006 - The Cygwin version of Cwd::cwd() will croak if given any arguments (which can happen if, for example, it's called as Cwd->cwd). Since that croaking is bad, we now wrap the original cwd() in a subroutine that ignores its arguments. We could skip this wrapping if a future version of perl changes cygwin.c's cwd() to not barf when fed an argument. [Jerry D. Hedden] 3.16 Mon Jan 30 20:48:41 CST 2006 - Updated to version 3.06 of ppport.h, which provides backward compatibility XS layers for older perl versions. - Clarify in the docs for File::Spec's abs2rel() and rel2abs() methods that the cwd() function it discusses is Cwd::cwd(). [Spotted by Steven Lembark] - Apparently the version of File::Path that ships with perl 5.8.5 (and perhaps others) calls Cwd::getcwd() with an argument (perhaps as a method?), which causes it to die with a prototyping error. We've eliminated the prototype by using the "(...)" arglist, since "PROTOTYPE: DISABLE" for the function didn't seem to work. [Spotted by Eamon Daly and others] 3.15 Tue Dec 27 14:17:39 CST 2005 - The Cwd::getcwd() function on *nix is now a direct pass-through to the underlying getcwd() C library function when possible. This is safer and faster than the previous implementation, which just did abs_path('.'). The pure-perl version has been kept for cases in which the XS version can't load, such as when running under miniperl. [Suggested by Nick Ing-Simmons] - When Cwd searches for a 'pwd' executable in the $PATH, we now stop after we find the first one rather than continuing the search. We also avoid the $PATH search altogether when a 'pwd' was already found in a well-known and well-trusted location like /bin or /usr/bin. [Suggested by Nick Ing-Simmons] - On Win32 abs2rel($path, $base) was failing whenever $base is the root of a volume (such as C:\ or \\share\dir). This has been fixed. [Reported by Bryan Daimler] - In abs2rel() on VMS, we've fixed handling of directory trees so that the test $file = File::Spec::VMS->abs2rel('[t1.t2.t3]file', '[t1.t2.t3]') returns 'file' instead of an empty string. [John E. Malmberg] - In canonpath() on VMS, '[]' was totally optimized away instead of just returning '[]'. Now it's fixed. [John E. Malmberg] 3.14 Thu Nov 17 18:08:44 CST 2005 - canonpath() has some logic in it that avoids collapsing a //double/slash at the beginning of a pathname on platforms where that means something special. It used to check the value of $^O rather than the classname it was called as, which meant that calling File::Spec::Cygwin->canonpath() didn't act like cygwin unless you were actually *on* cygwin. Now it does. - Fixed a major bug on Cygwin in which catdir() could sometimes create things that look like //network/paths in cases when it shouldn't (e.g. catdir("/", "foo", "bar")). 3.13 Tue Nov 15 23:50:37 CST 2005 - Calling tmpdir() on Win32 had the unintended side-effect of storing some undef values in %INC for the TMPDIR, TEMP, and TMP entries if they didn't exist already. This is probably a bug in perl itself (submitted as #37441 on rt.perl.org), which we're now working around. [Thomas L. Shinnick] - Integrated a change from bleadperl - a certain #ifdef in Cwd.xs needs to apply to WIN32 but not WinCE. [Vadim Konovalov] - abs2rel() used to return the empty string when its two arguments were identical, which made no sense. Now it returns curdir(). [Spotted by David Golden] - The Unix and Win32 implementations of abs2rel() have been unified. 3.12 Mon Oct 3 22:09:12 CDT 2005 - Fixed a testing error on OS/2 in which a drive letter for the root directory was confusing things. [Ilya Zakharevich] - Integrated a patch from bleadperl for fixing path() on Win32. [Gisle Aas] 3.11 Sat Aug 27 20:12:55 CDT 2005 - Fixed a couple of typos in the documentation for File::Spec::Mac. [Piotr Fusik] 3.10 Thu Aug 25 22:24:57 CDT 2005 - eliminate_macros() and fixpath() in File::Spec::VMS are now deprecated, since they are MakeMaker-specific and now live inside MakeMaker. [Michael Schwern] - canonpath() on Win32 now collapses foo/.. (or foo\..) sections correctly, rather than doing the "misguided" work it was previously doing. Note that canonpath() on Unix still does NOT collapse these sections, as doing so would be incorrect. [Michael Schwern] 3.09 Tue Jun 14 20:36:50 CDT 2005 - Added some block delimiters (brackets) in the Perl_getcwd_sv() XS function, which were necessary to separate the variable declarations from the statements when HAS_GETCWD is not defined. [Yves] - Apparently the _NT_cwd() routine is never defined externally like I thought it was, so I simplified the code around it. - When cwd() is implemented using the _backtick_pwd() function, it sometimes could create accidental undef entries in %ENV under perl 5.6, because local($hash{key}) is somewhat broken. This is now fixed with an appropriate workaround. [Neil Watkiss] 3.08 Sat May 28 10:10:29 CDT 2005 - Fixed a test failure with fast_abs_path() on Windows - it was sensitive to the rootdir() change from version 3.07. [Steve Hay] 3.07 Fri May 6 07:46:45 CDT 2005 - Fixed a bug in which the special perl variable $^O would become tainted under certain versions of perl. [Michael Schwern] - File::Spec->rootdir() was returning / on Win32. Now it returns \ . [Michael Schwern] - We now avoid modifying @_ in tmpdir() when it's not strictly necessary, which reportedly provides a modest performance boost. [Richard Soderberg] - Made a couple of slight changes to the Win32 code so that it works (or works better) on Symbian OS phones. [Jarkko Hietaniemi] 3.06 Wed Apr 13 20:47:26 CDT 2005 (No changes in functionality) - Added a note to the canonpath() docs about why it doesn't collapse foo/../bar sections. - The internal-only function bsd_realpath() in the XS file now uses normal arg syntax instead of K&R syntax. [Nicholas Clark] 3.05 Mon Feb 28 07:22:58 CST 2005 - Fixed a bug in fast_abs_path() on Win32 in which forward- and backward-slashes were confusing things. [demerphq] - Failure to load the XS code in Cwd is no longer a fatal error (though failure to compile it is still a fatal error in the build process). This lets Cwd work under miniperl in the core. [Rafael Garcia-Suarez] - In the t/cwd.t test, don't enforce loading from blib/ if we're testing in the perl core. [Rafael Garcia-Suarez] 3.04 Sun Feb 6 17:27:38 CST 2005 - For perls older than 5.006, the HAS_GETCWD symbol is not available, because it wasn't checked for in older Configure scripts when perl was built. We therefore just ask the user whether the getcwd() C function is defined on their platform when their perl is old. Maybe in the future we'll try to automate this. [Reported by several parties] - Remove lib/ppport.h from the distribution, so that MakeMaker doesn't accidentally pick it up and install it as a lib file. [Jerry Hedden] - Fixed a testing error on VMS that occurred when a user had read-access to the root of the current volume. [Craig A. Berry] 3.03 Fri Jan 21 21:44:05 CST 2005 - Fixed a testing error if the first file we find in the root directory is a symlink. [Blair Zajac] - Added a test to make sure Cwd.pm is loaded from blib/ during testing, which seems to be an issue in some people's environments and makes it awfully hard to debug things on my end. - Skip the _perl_abs_path() tests on Cygwin - they don't usually pass, and this function isn't used there anyway, so I decided not to push it. Let 'em use `cwd`. 3.02 Sun Jan 9 19:29:52 CST 2005 - Fixed a bug in which Cwd::abs_path() called on a file in the root directory returned strange results. [Bob Luckin] - Straightened out the licensing details for the portion of the Cwd module that comes from BSD sources. [Hugo van der Sanden] - Removed the prototype from _perl_abs_path() and the XS version of abs_path(), since all they seemed to be doing was causing people grief, and since some platforms didn't have them anyway. - Fixed a testing bug in which sometimes the wrong version of Cwd (the version already installed on the user's machine) would get loaded instead of the one we're building & installing. - Sometimes getdcwd() returns a lower-case drive letter, so don't require an upper-case letter in t/win32.t. [Jan Dubois] - Fixed a memory leak in getdcwd() on win32. [Steve Hay] - Added a copy of ppport.h to the distribution to aid compilation on older versions of perl. [Suggested by Charlie Root] - Don't bother looking for a 'pwd' executable on MSWin32 - there won't be one, and looking for it can be extremely slow when lots of disks are mounted. [Several people, including Andrew Burke] - Eliminated a multi-argument form of open() that was causing a syntax error on older versions of perl. [Fixed by Michael Schwern] - The bug-fix changes for revision 0.90 of File::Spec somehow were lost when I merged it into the PathTools distribution. They're restored now. [Craig A. Berry] - File::Spec->canonpath() will now reduce paths like '[d1.-]foo.dat' down to '[000000]foo.dat' instead of '[]foo.dat' or 'foo.dat'. This is in better accordance with the native filename syntax parser. [Craig A. Berry] - In order to remove a recursive dependency (PathTools -> Test-Simple -> Test-Harness -> PathTools), we now include a copy of Test::More in the distribution. It is only used for testing, it won't be installed with the rest of the stuff. - Removed some 5.6-isms from Cwd in order to get it to build with older perls like 5.005. - abs_path() on Windows was trampling on $ENV{PWD} - fixed. [Spotted by Neil MacMullen] - Added licensing/copyright statements to the POD in each .pm file. [Spotted by Radoslaw Zielinski] 3.01 Mon Sep 6 22:28:06 CDT 2004 - Removed an unnecessary and confusing dependency on File::Spec from the Makefile.PL and the Build.PL. - Added a 'NAME' entry to the Makefile.PL, because otherwise it won't even begin to work. [Reported by many] 3.00 Thu Sep 2 22:15:07 CDT 2004 - Merged Cwd and File::Spec into a single PathTools distribution. This was done because the two modules use each other fairly extensively, and extracting the common stuff into another distribution was deemed nigh-impossible. The code in revision 3.00 of PathTools should be equivalent to the code in Cwd 2.21 and File::Spec 0.90. ================================================================== Prior to revision 3.00, Cwd and File::Spec were maintained as two separate distributions. The revision history for Cwd is shown here. The revision history for File::Spec is further below. ================================================================== Cwd 2.21 Tue Aug 31 22:50:14 CDT 2004 - Removed "NO_META" from the Makefile.PL, since I'm not building the distribution with MakeMaker anyway. [Rohan Carly] - Only test _perl_abs_path() on platforms where it's expected to work (platforms with '/' as the directory separator). [Craig A. Berry] Cwd 2.20 Thu Jul 22 08:23:53 CDT 2004 - On some implementations of perl on Win32, a memory leak (or worse?) occurred when calling getdcwd(). This has been fixed. [PodMaster] - Added tests for getdcwd() on Win32. - Fixed a problem in the pure-perl implementation _perl_abs_path() that caused a fatal error when run on plain files. [Nicholas Clark] To exercise the appropriate test code on platforms that wouldn't otherwise use _perl_abs_path(), run the tests with $ENV{PERL_CORE} or $ENV{TEST_PERL_CWD_CODE} set. Cwd 2.19 Thu Jul 15 08:32:18 CDT 2004 - The abs_path($arg) fix from 2.18 didn't work for VMS, now it's fixed there. [Craig Berry] Cwd 2.18 Thu Jun 24 08:22:57 CDT 2004 - Fixed a problem in which abs_path($arg) on some platforms could only be called on directories, and died when called on files. This was a problem in the pure-perl implementation _perl_abs_path(). - Fixed fast_abs_path($arg) in the same way as abs_path($arg) above. - On Win32, a function getdcwd($vol) has been added, which gets the current working directory of the specified drive/volume. [Steve Hay] - Fixed a problem on perl 5.6.2 when built with the MULTIPLICITY compile-time flags. [Yitzchak Scott-Thoennes] - When looking for a `pwd` system command, we no longer assume the path separator is ':'. - On platforms where cwd() is implemented internally (like Win32), don't look for a `pwd` command externally. This can greatly speed up load time. [Stefan Scherer] - The pure-perl version of abs_path() now has the same prototype as the XS version (;$). Cwd 2.17 Wed Mar 10 07:55:36 CST 2004 - The change in 2.16 created a testing failure when tested from within a path that contains symlinks (for instance, /tmp -> /private/tmp). Cwd 2.16 Sat Mar 6 17:56:31 CST 2004 - For VMS compatibility (and to conform to Cwd's documented interface), in the regression tests we now compare output results to an absolute path. [Craig A. Berry] Cwd 2.15 Fri Jan 16 08:09:44 CST 2004 - Fixed a problem on static perl builds - while creating Makefile.aperl, it was loading a mismatched version of Cwd from blib/ . [Reported by Blair Zajac] Cwd 2.14 Thu Jan 8 18:51:08 CST 2004 - We now use File::Spec->canonpath() and properly-escaped regular expressions when comparing paths in the regression tests. This fixes some testing failures in 2.13 on non-Unix platforms. No changes were made in the actual Cwd module code. [Steve Hay] Cwd 2.13 Fri Jan 2 22:29:42 CST 2004 - Changed a '//' comment to a '/* */' comment in the XS code, so that it'll compile properly under ANSI C rules. [Jarkko Hietaniemi] - Fixed a 1-character buffer overrun problem in the C code. [The BSD people] Cwd 2.12 Fri Dec 19 17:04:52 CST 2003 - Fixed a bug on Cygwin - the output of realpath() should have been tainted, but wasn't. [Reported by Tom Wyant] Cwd 2.10 Mon Dec 15 07:50:12 CST 2003 (Note that this release was mistakenly packaged as version 2.11, even though it had an internal $VERSION variable of 2.10. Not sure how THAT happened...) - There was a dependency in the Makefile.PL on Module::Build, which isn't necessary. I've removed it. Cwd 2.09 Thu Dec 11 20:30:58 CST 2003 - The module should now build & install using version 5.6 of perl. - We now note a build-time dependency on version 0.19 of Module::Build, which is necessary because we don't use the standard lib/-based file layout. No version of Module::Build is required if you use the Makefile.PL, just if you use the Build.PL . - Removed some gratuitous uses of 5.6-isms like our(), with the aim of backporting this module to perl 5.005. - Simplified all code that autoloads Carp.pm and calls carp()/croak(). - Removed some redundant OS/2 code at the suggestion of Michael Schwern and Ilya Zakharevich. - Make sure the correct version of Cwd.pm is loaded in the regression tests. [Sam Vilain] Cwd 2.08 Wed Oct 15 20:56 CDT 2003 - Code extracted from perl 5.8.1 and packaged as a separate CPAN release by Ken Williams. ================================================================== Prior to revision 3.00, Cwd and File::Spec were maintained as two separate distributions. The revision history for File::Spec is shown here. The revision history for Cwd is above. ================================================================== File::Spec 0.90 Tue Aug 31 22:34:50 CDT 2004 - On VMS, directories use vmspath() and files use vmsify(), so rel2abs() has to use some 'educated guessing' when dealing with paths containing slashes. [Craig A. Berry] File::Spec 0.89 Sun Aug 29 19:02:32 CDT 2004 - Fixed some pathological cases on VMS which broke canonpath() and splitdir(). [Richard Levitte and Craig A. Berry] - Fixed rel2abs() on VMS when passed a unix-style relative path. [Craig A. Berry] File::Spec 0.88 Thu Jul 22 23:14:32 CDT 2004 - rel2abs() on Win32 will now use the new Cwd::getdcwd() function, so that things like rel2abs('D:foo.txt') work properly when the current drive isn't 'D'. This requires Cwd version 2.18. [Steve Hay] - Got rid of a redundant double-backslash in a character class. [Alexander Farber] - Added much markup to pod for enhanced readability. [Andy Lester] File::Spec 0.87 Fri Dec 19 08:03:28 CST 2003 - With a one-line change in the tests, backported to perl 5.004. [Issue reported by Barry Kemble] File::Spec 0.86 Fri Sep 26 10:07:39 CDT 2003 - This is the version released with perl 5.8.1. It is identical to the code in the File::Spec beta 0.85_03. File::Spec 0.85_03 Mon Sep 15 09:35:53 CDT 2003 - On VMS, if catpath() receives volume specifiers in both its first two arguments, it will now use the volume in the first argument only. Previously it returned a non-syntactical result which included both volumes. This change is the same in spirit to the catpath() MacOS change from version 0.85_02. - Fixed an abs2rel() bug on VMS - previously abs2rel('[t1.t2.t3]file','[t1.t2]') returned '[t3]file', now it properly returns '[.t3]file'. File::Spec 0.85_02 Fri Sep 12 17:11:13 CDT 2003 - abs2rel() now behaves more consistently across platforms with the notion of a volume. If the volumes of the first and second argument (the second argument may be implicit) do not agree, we do not attempt to reconcile the paths, and simply return the first argument. Previously the volume of the second argument was (usually) ignored, resulting in sometimes-garbage output. - catpath() on MacOS now looks for a volume element (i.e. "Macintosh HD:") in its first argument, and then its second argument. The first volume found will be used, and if none is found, none will be used. - Fixed a problem in abs2rel() on Win32 in which the volume of the current working directory would get added to the second argument if none was specified. This might have been somewhat helpful, but it was contrary to the documented behavior. For example, abs2rel('C:/foo/bar', '/foo') used to return 'bar', now it returns 'C:/foo/bar' because there's no guarantee /foo is actually C:/foo . - catdir('/', '../') on OS2 previously erroneously returned '//..', and now it returns '/'. File::Spec 0.85_01 Thu Sep 11 16:18:54 CDT 2003 Working toward 0.86, the version that will be released with perl 5.8.1. - The t/rel2abs2rel.t test now is a little friendlier about emitting its diagnostic debugging output. [Jarkko Hietaniemi] - We now only require() Cwd when it's needed, on demand. [Michael Schwern, Tels] - Fixed some POD errors and redundancies in OS2.pm and Cygwin.pm. [Michael Schwern] - The internal method cwd() has been renamed to _cwd(), since it was never meant for public use. [Michael Schwern] - Several methods in File::Spec::Unix that just return constant strings have been sped up. catdir() has also been sped up there. [Tels] - Several canonpath() and catdir() bugs on Win32 have been fixed, and tests added for them: catdir('/', '../') -> '\\' (was '\..') catdir('/', '..\\') -> '\\ (was '') canonpath('\\../') -> '\\' (was '') canonpath('\\..\\') -> '\\' (was '') canonpath('/../') -> '\\' (was '\..') canonpath('/..\\') -> '\\' (was '') catdir('\\', 'foo') -> '\foo' (was '\\foo') - catpath($volume, $dirs, $file) on Mac OS now ignores any volume that might be part of $dirs, enabling catpath($volume, catdir(rootdir(), 'foo'), '') to work portably across platforms. File::Spec 0.85 Tue Jul 22 11:31 CDT 2003 A bug-fix release relative to 0.84. I've forked development into a "stable" branch (this one) and a more aggressive branch (as yet unreleased), with an eye toward getting the stable features in perl 5.8.1. - File::Spec::Mac->case_tolerant() returned 0 when it should have returned 1. - Many cases in File::Spec::Win32->abs2rel() were broken, because of the way in which volumes were/weren't ignored. Unfortunately, part of the regression tests were broken too. Now, if the $path argument to abs2rel() is on a different volume than the $base argument, the result will be an absolute path rather than the broken relative path previous versions returned. - Fixed a problem in File::Spec::Win32->canonpath, which was turning \../foo into "foo" rather than \foo - Greatly simplified the code in File::Spec::Unix->splitdir(). File::Spec 0.84_01 Fri Jul 11 16:14:29 CDT 2003 No actual code changes, just changes in other distribution files - Dependencies are now listed explicitly in the Makefile.PL and Build.PL scripts, as well as in the META.yml file. - The t/abs2rel2abs.t test should now be more friendly about skipping on platforms where it can't run properly. File::Spec 0.84 Wed Jul 9 22:21:23 CDT 2003 I (Ken)'ve taken the changes from bleadperl and created a new CPAN release from them, since they're pretty important changes. The highlights, from what I can tell, are listed here. - A huge number of changes to File::Spec::Mac in order to bring it in line with the other platforms. This work was mostly/completely done by Thomas Wegner. - The Epoc and Cygwin platforms are now supported. - Lots of generically-applicable documentation has been taken from File::Spec::Unix and put in File::Spec. - A Build.PL has been provided for people who wish to install via Module::Build. - Some spurious warnings and errors in the tests have been eliminated. [Michael Schwern] - canonpath() on File::Spec::Unix now honors a //node-name at the beginning of a path. - Cwd.pm wasn't being loaded properly on MacOS. [Chris Nandor] - Various POD fixups - Several testing patches for the Epoc and Cygwin platforms [Tels] - When running under taint mode and perl >= 5.8, all the tmpdir() implementations now avoid returning a tainted path. - File::Spec::OS2 now implements canonpath(), splitpath(), splitdir(), catpath(), abs2rel(), and rel2abs() directly rather than inheriting them from File::Spec::Unix. - Added 'SYS:/temp' and 'C:/temp' to the list of possible tmpdir()s on Win32. - catfile() on Win32 and VMS will now automatically call canonpath() on its final argument. - canonpath() on Win32 now does a much more extensive cleanup of the path. - abs2rel() on Win32 now defaults to using cwd() as the base of relativity when no base is given. - abs2rel() on Win32 now explicitly ignores any volume component in the $path argument. - canonpath() on VMS now does []foo ==> foo, and foo.000000] ==> foo]. It also fixes a bug in multiple [000000.foo ==> [foo translations. - tmpdir() on VMS now uses 'sys$scratch:' instead of 'sys$scratch'. - abs2rel() on VMS now uses '000000' in both the path and the base. File::Spec 0.82 Wed Jun 28 11:24:05 EDT 2000 - Mac.pm: file_name_is_absolute( '' ) now returns TRUE on all platforms - Spec.pm: unbreak C<$VERSION = '0.xx'> to be C<$VERSION = 0.xx>, so underscores can be used when I want to update CPAN without anyone needing to update the perl repository. - abs2rel, rel2abs doc tweaks - VMS.pm: get $path =~ /\s/ checks from perl repository. - Makefile.PL: added INSTALLDIRS => 'perl', since these are std. modules. - Remove vestigial context prototypes from &rel2abs until some future arrives where method prototypes are honored. usr/share/doc/perl-HTML-Parser/Changes000064400000125354152531372150013473 0ustar00_______________________________________________________________________________ 2016-01-19 Release 3.72 Gisle Aas (3): Avoid more clang casting warnings Remove trailing whitespace Ensure entities expand to utf8 sequences under 'utf8_mode' [RT#99755] David Steinbrunner (2): typo fix typo fixes Jacques Germishuys (1): Silence clang warning bulk88 (1): const+static-ing _______________________________________________________________________________ 2013-05-09 Release 3.71 Gisle Aas (1): Transform ':' in headers to '-' [RT#80524] _______________________________________________________________________________ 2013-03-28 Release 3.70 François Perrad (1): Fix for cross-compiling with Buildroot Gisle Aas (1): Comment typo fix Yves Orton (1): Fix Issue #3 / RT #84144: HTML::Entities::decode_entities() needs to call SV_CHECK_THINKFIRST() before checking READONLY flag _______________________________________________________________________________ 2011-10-15 Release 3.69 Gisle Aas (4): Documentation fix; encode_utf8 mixup [RT#71151] Make it clearer that there are 2 (actually 3) options for handing "UTF-8 garbage" Github is the official repo Can't be bothered to try to fix the failures that occur on perl-5.6 Barbie (1): fix to TokeParser to correctly handle option configuration Jon Jensen (1): Aesthetic change: remove extra ; Ville Skyttä (1): Trim surrounding whitespace from extracted URLs. _______________________________________________________________________________ 2010-09-01 Release 3.68 Gisle Aas (1): Declare the encoding of the POD to be utf8 _______________________________________________________________________________ 2010-08-17 Release 3.67 Nicholas Clark (1): bleadperl 2154eca7 breaks HTML::Parser 3.66 [RT#60368] _______________________________________________________________________________ 2010-07-09 Release 3.66 Gisle Aas (1): Fix entity decoding in utf8_mode for the title header _______________________________________________________________________________ 2010-04-04 Release 3.65 Gisle Aas (1): Eliminate buggy entities_decode_old Salvatore Bonaccorso (1): Fixed endianness typo [RT#50811] Ville Skyttä (1): Documentation fixes. _______________________________________________________________________________ 2009-10-25 Release 3.64 Gisle Aas (5): Convert files to UTF-8 Don't allow decode_entities() to generate illegal Unicode chars Copyright 2009 Remove rendundant (repeated) test Make parse_file() method use 3-arg open [RT#49434] _______________________________________________________________________________ 2009-10-22 Release 3.63 Gisle Aas (2): Take more care to prepare the char range for encode_entities [RT#50170] decode_entities confused by trailing incomplete entity _______________________________________________________________________________ 2009-08-13 Release 3.62 Ville Skyttä (4): HTTP::Header doc typo fix. Do not bother tracking style or script, they're ignored. Bring HTML 5 head elements up to date with WD-html5-20090423. Improve HeadParser performance. Gisle Aas (1): Doc patch: Make it clearer what the return value from ->parse is _______________________________________________________________________________ 2009-06-20 Release 3.61 Gisle Aas (2): Test that triggers the crash that Chip fixed Complete documented list of literal tags Chip Salzenberg (1): Avoid crash (referenced pend_text instead of skipped_text) Antonio Radici (1): Reference HTML::LinkExttor [RT#43164] _______________________________________________________________________________ 2009-02-09 Release 3.60 Ville Skytta (5): Spelling fixes. Test multi-value headers. Documentation improvements. Do not terminate head parsing on the element (added in HTML 4.0). Add support for HTML 5 and new HEAD elements. Damyan Ivanov (1): Short description of the htextsub example Mike South (1): Suppress warning when encode_entities is called with undef [RT#27567] Zefram (1): HTML::Parser doesn't compile with perl 5.8.0. _______________________________________________________________________________ 2008-11-24 Gisle Aas Release 3.59 Restore perl-5.6 compatibility for HTML::HeadParser. Improved META.yml _______________________________________________________________________________ 2008-11-17 Gisle Aas Release 3.58 Suppress "Parsing of undecoded UTF-8 will give garbage" warning with attr_encoded [RT#29089] HTML::HeadParser: - Recognize the Unicode BOM in utf8_mode as well [RT#27522] - Avoid ending up with '/' keys attribute in Link headers. _______________________________________________________________________________ 2008-11-16 Gisle Aas Release 3.57 The