Revision history for Test-Harness

3.45_02  2023-06-01
        - Avoid using Errno::EINTR directly for platforms without it

3.45_01  28-04-2023
        - Color the "ok"s as well.
        - Skip symlink tests on msys2
        - Use use absolute path for executable tests
        - Space-quote executable if has spaces

3.44     17-04-2022
        - Release 3.43_06 as non-dev

3.43_06  11-04-2022
        - Fix skipping SEGV test

3.43_05  07-04-2022
        - Fix failing SEGV test on Windows

3.43_04 22-01-2022
        - Remove ASCII-isms to better work on EBCDIC

3.43_03 30-07-2021
        - Move timer initialization
        - Fix YAMLish behaviour with empty values
        - fix eintr error handling in TAP::Parser::Multiplexer
        - Parse out signal name & core dump.

3.43_02 25-03-2020
        - Let the aggregator finish gracefully after bailout
        - Make prove respect HARNESS_VERBOSE if no verbosity flags are passed

3.42    19-03-2018
        - Enable rulesfile.t to run in core

3.41    27-02-2018
        - Released 3.40_01 without code modifications

3.40_01 23-07-2017
        - Return handle for pipes and sockets #58 (Erik Huelsmann)
        - TAP v13 plan allows trailing whitespace (Steffen Schwigon)
        - prove: add a --statefile=<path> option to customize the .prove file
          (Ævar Arnfjörð Bjarmason)
        - Avoid non-deterministic source handling, make a SourceHandler tie an
          error. (Michael Schwern, Leon Timmermans)
        - Fix and simplify MSWin32 colorization (Roy Ivy III)
        - Fix file source handler to accept single extensions option (Tomoki Aonuma)
        - Spelling fixes (Brian Wightman)

3.39    06-04-2017
        - Make tests pass when PERL_USE_UNSAFE_INC=0

3.38    13-03-2017
        - Released 3.37_01 without changes

3.37_01
        - Set PERL_USE_UNSAFE_INC when running tests using Test::Harness (Leon Timmermans)
        - Avoid loading optional modules from . in prove

3.36    30-12-2015
        - Accept YAML with trailing whitespace in header (Maik Hentsche)
        - Stop bundling Test::More for testing

3.35    2015-01-14
        - Fix prove --version to actually print the version (Leon Timmermans, #101216)
        - Add --version to usage message (Leon Timmermans, #101215)

3.34    2014-11-02
        - Enable printing CPU times spent per test (Jarkko Hietaniemi)

3.33    2014-08-16
        - Various documentation fixes (Leon Timmermans, Justin Cook)

3.32    2014-06-11
        - Remove harness_class from argument hash in T::H::E (Leon Timmermans)

3.31    2014-06-07
        - Implement external rulesfile for TAP::Harness (David Golden)
        - Add harness_class argument to TAP::Harness::Env (Leon Timmermans)
        - Make prove respect environmental variables #28 (Leon Timmermans)

3.30    2013-11-12
        - Fix missing parent prereq in META.{yml,json} and NotBuild.PL
          (Dagfinn Ilmari Mannsåker, #89650)
        - Respect PERL5LIB in tainting source handler test (Dagfinn Ilmari Mannsåker, 
          Leon Timmermans)
        - Use base instead of parent:

          This dist is used for testing all other modules, so it should avoid
          having any non-core prerequisites.  Having parent as a prereq leads to a
          circular dependency of parent -> Test::More -> Test::Harness. (Graham Knop)
        - Various POD fixes (Nathan Gary Glenn)
        - Don't localize all of %ENV in harness.t (Craig Berry)
        - Give TAP::Harness::Beyond a unique NAME (Leon Timmermans)

3.29    2013-08-10
        - Get rid of use vars in favor of our in all modules (Leon Timmermans)
          and tests (Karen Etheridge)
        - Added use warnings to all modules (Leon Timmermans) and tests (Karen
          Etheridge)
        - Use parent instead of @ISA in all modules (Leon Timmermans) and
          tests (Karen Etheridge)
        - Fix failing test on VMS (Craig Berry)
        - Improve error message on loading failure (Leon Timmermans, #77730)
        - Use Text::ParseWords, deprecate TAP::Parser::Utils

3.28    2013-05-02
        - Bugfix: Fix taint failures on Windows (Jan Dubois)

3.27    2013-04-30
        - Dramatically reduce memory usage (Nick Clark, RT #84939)
        - Store test_num (in Grammar.pm) as a number instead of a string.
          Reduces memory usage (Nick Clark, RT #84939)
        - PERL5LIB is always propogated to a test's @INC, even with taint more
          (Schwern, RT #84377)
        - restore "always add -w to switches" behavior

3.26    2013-01-16
        - Renamed env.opts.t to env_opts.t (for VMS)
        - Skipped some TAP::Formatter::HTML tests due to this bug: #82738

3.26    2012-06-05
        - Rereleased to fix CPAN permission problem. No functional change.

3.24    2012-06-03
        - RT #74393: corrected typo in M::B integration docs.
        - RT #63473: fix typo.
        - RT #49732: Attempt to load File::Glob::Windows to get correct
          glob semantics on Win32.
        - RT #47890: Don't use Win32::GetShortPathName.
        - RT #64404: Ignore textness ('-T') of script when reading shebang.
        - Handle the case where we don't know the wait status of the
          test more gracefully.
        - Make the test summary 'ok' line overrideable so that it can be
          changed to a plugin to make the output of prove idempotent.
        - Stop adding '-w' to perl switches by default
        - Apply upstream patch:

            http://perl5.git.perl.org/perl.git/commit \
              /6359c64336d99060952232e7e300bd3c31afead8

          In testargs.t in Test::Harness, don't run a world-writable file.

          The test writes a file, then changes the mode, then executes it. The file needs
          to be +x to be executable (on many platforms). The file will need to be +w to
          be deletable on some platforms. But setting the file world writable just before
          running it feels like a bad idea, given that the file's name is as predictable
          as process IDs, as there's a race condition to break into the account running
          perl's tests.


3.23    2011-02-20
        - Merge in changes from core. Thanks BinGOs.
        - Made SourceHandler understand that an executable binary file
          is probably an executable.
        - Added workaround for Getopt::Long 2.25 handling of
          multivalue options. Fixes test failure on stock perl 5.6.2.

3.22    2010-08-14
        - Allow TAP::Parser to recognize a nested BAIL_OUT directive.
        - Add brief HOWTO for creating and running pgTAP tests to
          TAP::Parser::SourceHandler::pgTAP.
        - Fix trailing plan + embedded YAML + TAP 13 case. Thanks to
          Steffen Schwigon. #54518.
        - Numerous spelling fixes. Thanks to Ville Skyttä.
        - Add new option --tapversion for prove to set the default
          assumed TAP version. Thanks to Steffen Schwigon.
        - Fixed tests to run successfully under Devel::Cover. Thanks to
          Phillipe Bruhat.
        - Fixed injection of test args to work with general executables
          as well as Perl scripts (#59186).
        - Allow multiple --ext=.foo arguments to prove, to allow running
          different types of tests in the same prove run.
        - App::Prove::extension() is now App::Prove::extensions(), and
          returns an arrayref of extensions, rather than a single scalar.
          The same change has been made to App::Prove::State::extension().
        - Preserve old semantics for test scripts with a shebang line
          by favouring Perl as the intepreter for any file with a
          shebang (#59457).
        - Add --trap (summary on Ctrl-C) option to prove (#59427).
        - Removed TAP::Parser::SourceHandler::pgTAP. Find it in its own
          distribution on CPAN.
        - Source options to prove can now be specified so as to be passed to
          the source as a hash reference, eg:

              prove --source XYZ --xyz-option pset=foo=bar

          Ths "pset" option will be passed as a hash reference with the key
          "foo" and the value "bar".

3.21    2010-01-30
        - Add test to ensure we're not depending on a module we no
          longer ship.
        - Fix up skip counts for Windows case - tests were failing
          on Windows.

3.20    2010-01-22
        - Remove references / dependency on TAP::Parser::Source::Perl

3.19    2010-01-20
        - Avoid depending on Module::Build. The resulting circular
          dependency made it impossible to install Test::Harness and/or
          Module::Build in some cases.

3.18    2010-01-19
        - Handle the case where the filename of the perl executable
          contains space. Thanks to kmx.
        - Various documentation fixes.

3.17_04 2010-01-04
        - Fix failures due to unknown location of Perl in t/source_handler.t.
        - Use EUMM style shebang magic to produce an executable 'psql'
          for t/source_handler.t.

3.17_03 2009-11-19
        - Fix failures due to over-strict assertions in t/source.t.

3.17_02 2009-11-17
        - Merge in Steve's missing changes. Oops.

3.17_01 2009-11-17
        - Re-engineered source handling API to allow users to configure how
          TAP is sourced by the parser.  Introduced a new 'sources' param to
          TAP::Harness, and new options to prove, eg:

              prove --source XYZ --xyz-option foo=bar

          The new TAP::Parser::SourceHandler API makes it much easier to
          write plugins.  This breaks backwards compatibility for plugins &
          extenstions that rely on the following APIs:

              TAP::Parser::Source
              TAP::Parser::SourceFactory
              TAP::Parser::IteratorFactory
              TAP::Parser, specifically:
                new: 'source' & 'tap' params
                source_class
                perl_source_class
                iterator_factory_class
                make_source
                make_perl_source
                make_iterator

          Please see the TAP::Parser docs for more details.
          [Steve Purkis & David Wheeler]
        - Removed dependency on File::Spec [Schwern]
        - Made it possible to pass different args to each test [Lee Johnson]
        - Added HARNESS_SUBCLASS option to Test::Harness
        - Added TAP::Parser::SourceHandler::File which lets you to stream TAP
          from a text file (eg: *.tap).
        - Added TAP::Parser::SourceHandler::pgTAP. All the source handlers are
          new, but this is the only one to add major new functioality: the
          ability to run pgTAP tests (http://pgtap.projects.postgresql.org/).

3.17    2009-05-05
        - Changed the 'failures' so that it is overridden by verbosity rather
          than the other way around.
        - Added the 'comments' option, most useful when used in conjunction
          with the 'failures' option.
        - Deprecated support for Perls earlier than 5.6.0.
        - Allow '-I lib' as well as '-Ilib' in $Test::Harness::Switches
          (regression).
        - Restore old skip parsing semantics for TAP < v13. Refs #39031.
        - Numerous small documentation fixes.
        - Remove support for fork-based parallel testing. Multiplexed
          parallel testing remains.

3.16    2009-02-19
        - Fix path splicing on platforms where the path separator
          is not ':'.
        - Fixes/skips for failing Win32 tests.
        - Don't break with older CPAN::Reporter versions.

3.15    2009-02-17
        - Refactor getter/setter generation into TAP::Object.
        - The App::Prove::State::Result::Test now stores the parser object.
        - After discussion with Andy, agreed to clean up the test output
          somewhat.  t/foo.....ok becomes t/foo.t ... ok
        - Make Bail out! die instead of exiting. Dies with the same
          message as 2.64 for (belated) backwards compatibility.
        - Alex Vaniver's patch to refactor TAP::Formatter::Console into
          a new class, TAP::Formatter::File and a common base class: 
          TAP::Formatter::Base.
        - Fix a bug where PERL5LIB might be put in the wrong spot in @INC.
          #40257
        - Steve Purkis implemented a plugin mechanism for App::Prove.

3.14    2008-09-13
        - Created a proper (ha!) API for prove state results and tests.
        - Added --count and --nocount options to prove to control X/Y display
          while running tests.
        - Added 'fresh' state option to run test scripts that have been
          touched since the test run.
        - fixed bug where PERL5OPT was not properly split  
        - fixed regex in _filtered_inc. Fixes #39248. Thanks Slaven.

3.13    2008-07-27
        - fixed various closure related leaks
        - made prove honour HARNESS_TIMER
        - Applied patches supplied by Alex Vandiver
          - add 'rules' switch to prove: allows parallel execution rules
            to be specified on the command line.
          - allow '**' (any path) wildcard in parallel rules
          - fix bug report address
          - make tprove_gtk example work again.

3.12    2008-06-22
        - applied Steve Purkis' huge refactoring patch which adds
          configurable factories for most of the major internal classes.
        - applied David Wheeler's patch to allow exec to be a code
          reference.
        - made tests more robust in the presence of -MFoo in PERL5OPT.

3.11    2008-06-09
        - applied Jim Keenan's patch that makes App::Prove::run return a
          rather than exit (#33609)
        - prove -r now recurses cwd rather than 't' by default (#33007)
        - restored --ext switch to prove (#33848)
        - added ignore_exit option to TAP::Parser and corresponding
          interfaces to TAP::Harness and Test::Harness. Requested for
          Parrot.
        - Implemented rule based parallel scheduler.
        - Moved filename -> display name mapping out of formatter. This
          prevents the formatter's strip-extensions logic from stripping
          extensions from supplied descriptions.
        - Only strip extensions from test names if all tests have the
          same extension. Previously we stripped extensions if all names
          had /any/ extension making it impossible to distinguish tests
          whose name differed only in the extension.
        - Removed privacy test that made it impossible to subclass
          TAP::Parser.
        - Delayed initialisation of grammar making it easier to replace
          the TAP::Parser stream after instantiation.
        - Make it possible to supply import parameters to a replacement
          harness with prove.
        - Make it possible to replace either _grammar /or/ _stream
          before reading from a TAP::Parser.

3.10    2008-02-26
        - fix undefined value warnings with bleadperl.
        - added pragma support.
        - fault unknown TAP tokens under strict pragma.

3.09    2008-02-10
        - support for HARNESS_PERL_SWITCHES containing things like 
          '-e "system(shift)"'.
        - set HARNESS_IS_VERBOSE during verbose testing.
        - documentation fixes.

3.08    2008-02-08
        - added support for 'out' option to
          Test::Harness::execute_tests. See #32476. Thanks RENEEB.
        - Fixed YAMLish handling of non-alphanumeric hash keys.
        - Added --dry option to prove for 2.64 compatibility.

3.07    2008-01-13
        - prove now supports HARNESS_PERL_SWITCHES.
        - restored TEST_VERBOSE to prove.

3.06    2008-01-01
        - Skip t/unicode.t if PERL_UNICODE set. Fixes #31731.
          Thanks Lukas.
        - App::Prove::State no longer complains about tests that
          are deleted.
        - --state=new and --state=old now consider the modification time
          of test scripts.
        - Made test suite core-compatible.

3.05    2007-12-09
        - Skip unicode.t if Encode unavailable
        - Support for .proverc files.
        - Clarified prove documentation.

3.04    2007-12-02
        - Fixed output leakage with really_quiet set.
        - Progress reports for tests without plans now show
          "143/?" instead of "143/0".
        - Made TAP::Harness::runtests support aliases for test names.
        - Made it possible to pass command line args to test programs
          from prove, TAP::Harness, TAP::Parser.
        - Added --state switch to prove.

3.03    2007-11-17
        - Fixed some little bugs-waiting-to-happen inside
          TAP::Parser::Grammar.
        - Added parser_args callback to TAP::Harness.
        - Made @INC propagation even more compatible with 2.64 so that
          parrot still works *and* #30796 is fixed.

3.02    2007-11-15
        - Process I/O now unbuffered, uses sysread, plays better with
          select. Fixes #30740.
        - Made Test::Harness @INC propagation more compatible with 2.64.
          Was breaking Parrot's test suite.
        - Added HARNESS_OPTIONS (#30676)

3.01    2007-11-12
        - Fix for RHEL incpush.patch related failure.
        - Output real time of test completion with --timer
        - prove -b adds blib/auto to @INC
        - made SKIP plan parsing even more liberal for pre-v13 TAP

3.00    2007-11-06
        - Non-dev release. No changes since 2.99_09.

2.99_09 2007-11-05
        - Implemented TODO-in-PLAN syntax for TAP version 12 and earlier.

2.99_08 2007-11-04
        - Tiny changes. New version pushed to get some smoke coverage.

2.99_07 2007-11-01
        - Fix for #21938: Unable to handle circular links
        - Fix for #24926: prove -b and -l should use absolute paths
        - Fixed prove switches. Big oops. How the hell did we miss that?
        - Consolidated quiet, really_quiet, verbose into verbosity.
        - Various VMS related fixes to tests

2.99_06 2007-10-30
        - Added skip_all method to TAP::Parser.
        - Display reason for skipped tests.
        - make test now self tests.

2.99_05 2007-10-30
        - Fix for occasional rogue -1 exit code on Windows.
        - Fix for @INC handling under CPANPLUS.
        - Added real time to prove --timer output
        - Improved prove error message in case where 't' not found and
          no tests named.

2.99_04 2007-10-11
        - Fixed bug where 'All tests successful' would not be printed if bonus
          tests are seen.
        - Fixed bug where 'Result: FAIL' would be printed at the end of a test
          run if there were unexpectedly succeeding tests.
        - Added -M, -P switches to allow arbitrary modules to be loaded
          by prove. We haven't yet defined what they'll do once they
          load but it's a start...
        - Added testing under simulated non-forking platforms.

2.99_03 2007-10-06
        - Refactored all display specific code out of TAP::Harness.
        - Relaxed strict parsing of skip plan for pre v13 TAP.
        - Elapsed hi-res time is now displayed in integer milliseconds
          instead of fractional seconds.
        - prove stops running if any command-line switches are invalid.
        - prove -v would try to print an undef.
        - Added support for multiplexed and forked parallel tests. Use
          prove -j 9 to run tests in parallel and prove -j 9 --fork to
          fork. These features are experimental and currently
          unavailable on Windows.
        - Rationalized the management of the environment that we give to
          test scripts (PERL5LIB, PERL5OPT, switches).
        - Fixed handling of STDIN (we no longer close it) for test
          scripts.
        - Performance enhancements. Parser is now 30% - 40% faster.

2.99_02 2007-09-07
        - Ensure prove (and App::Prove) sort any recursively
          discovered tests
        - It is now possible to register multiple callback handlers for
          a particular event.
        - Added before_runtests, after_runtests callbacks to
          TAP::Harness.
        - Moved logic of prove program into App::Prove.
        - Added simple machine readable summary.
        - Performance improvement: The processing pipeline within
          TAP::Parser is now a closure which speeds up access to the
          various attribtes it needs.
        - Performance improvement: Test count spinner now updates
          exponentially less frequently as the count increases which
          saves a lot of I/O on big tests.
        - More improvements in test coverage from Leif.
        - Fixes to TAP spooling - now captures YAML blocks correctly.
        - Fix YAMLish handling of empty arrays, hashes.
        - Renamed TAP::Harness::Compatible to Test::Harness,
          runtests to prove.
        - Fixes to @INC handling. We didn't always pass the correct path
          to subprocesses.
        - We now observe any switches in HARNESS_PERL_SWITCHES.
        - Changes to output formatting for greater compatibility with
          Test::Harness 2.64.
        - Added unicode test coverage and fixed a couple of
          unicode issues.
        - Additions to documentation.
        - Added support for non-forking Perls. If forking isn't
          available we fall back to open and disable stream merging.
        - Added support for simulating non-forking Perls to improve our
          test coverage.

========================================================================
Version numbers below this point relate to TAP::Parser - which was the
name of this version of Test::Harness during its development.
========================================================================

0.54
    - Optimized I/O for common case of 'runtests -l'
    - Croak if supplied an empty (0 lines) Perl script.
    - Made T::P::Result::YAML return literal input YAML correctly.
    - Merged speed-ups from speedy branch.

0.53  18 August 2007
    - Fixed a few docs nits.
    - Added -V (--version) switch to runtests. Suggested by markjugg on
      Perlmonks.
    - Fixed failing t/030-grammer.t under 5.9.5. Exact cause still
      unknown; something to do with localisation of $1 et all I think.
    - Fixed use of three arg open in t/compat/test-harness-compat; was
      failing on 5.6.2.
    - Fixed runtests --exec option. T::H wasn't passing the exec option
      to T::P.
    - Merged Leif Eriksen's coverage enhancing changes to 
      t/080-aggregator.t, t/030-grammar.t
    - Made various changes so that we test cleanly on 5.0.5.
    - Many more coverage enhancements by Leif.
    - Applied Michael Peters' patch to add an EOF callback to
      TAP::Parser.
    - Added --reverse option to runtests to run tests in reverse order.
    - Made runtests exit with non-zero status if the test run had
      problems.
    - Stopped TAP::Parser::Iterator::Process from trampling on STDIN.

0.52  14 July 2007
    - Incorporate Schwern's investigations into TAP versions.
      Unversioned TAP is now TAP v12. The lowest explicit version number
      that can be specified is 13.
    - Renumbered tests to eliminate gaps.
    - Killed execrc.  The '--exec' switch to runtests handles all of this for
      us.
    - Refactored T::P::Iterator into
      T::P::Iterator::(Array|Process|Stream) so that we have a
      process specific iterator with which to experiment with
      STDOUT/STDERR merging.
    - Removed vestigial exit status handling from T::P::I::Stream.
    - Removed unused pid interface from T::P::I::Process.
    - Fixed infinite recursion in T::P::I::Stream and added regression
      coverage for same.
    - Added tests for T::P::I::Process.
    - TAP::Harness now displays the first five TAP syntax errors and
      explains how to pass the -p flag to runtests to see them all.
    - Added merge option to TAP::Parser::Iterator::Process,
      TAP::Parser::Source, TAP::Parser and TAP::Harness.
    - Added --merge option to runtests to enable STDOUT/STDERR merging.
      This behaviour used to be the default.
    - Made T::P::I::Process use open3 for both merged and non-merged
      streams so that it works on Windows.
    - Implemented Eric Wilhelm's IO::Select based multiple stream
      handler so that STDERR is piped to us even if stream merging is
      turned off. This tends to reduce the temporal skew between the
      two streams so that error messages appear closer to their
      correct location.
    - Altered the T::P::Grammar interface so that it gets a stream
      rather than the next line from the stream in preparation for
      making it handle YAML diagnostics.
    - Implemented YAML syntax. Currently YAML may only follow a
      test result. The first line of YAML is '---' and the last
      line is '...'.
    - Made grammar version-aware. Different grammars may now be selected
      depending on the TAP version being parsed.
    - Added formatter delegate mechanism for test results.
    - Added prototype stream based YAML(ish) parser.
    - Added more tests for T::P::YAMLish
    - Altered T::P::Grammar to use T::P::YAMLish
    - Removed T::P::YAML
    - Added raw source capture to T::P::YAMLish
    - Added support for double quoted hash keys
    - Added TAP::Parser::YAMLish::Writer and renamed T::P::YAMLish as
      T::P::YAMLish::Reader.
    - Added extra TAP::Parser::YAMLish::Writer output options
    - Inline YAML documents must now be indented by at least one space
    - Fixed broken dependencies in bin/prove
    - Make library paths absolute before running tests in case tests
      chdir before loading modules.
    - Added libs and switches handling to T::H::Compatible. This and the
      previous change fix [24926]
    - Added PERLLIB to libraries stripped in _default_inc [12030]
    - Our version of prove now handles directories containing circular
      links correctly [21938]
    - Set TAP_VERSION env var in Parser [11595]
    - Added setup, teardown hooks to T::P::I::Process to facilitate the
      setup and cleanup of the test script's environment
    - Any additional libs added to the command line are also added to
      PERL5LIB for the duration of a test run so that any Perl children
      of the test script inherit the same library paths.
    - Fixed handling of single quoted hash keys in T::P::Y::Reader
    - Made runtests return the TAP::Parser::Aggregator
    - Fixed t/120-harness.t has failures if TAP::Harness::Color cannot
      load optional modules [27125] - thanks DROLSKY
    - Fixed parsing of \# in test description  
0.51 12 March 2007
    - 'execrc' file now allows 'regex' matches for tests.
    - rename 'TAPx' --> 'TAP'
    - Reimplemented the parse logic of TAP::Parser as a state machine.
    - Removed various ad-hoc state variables from TAP::Parser and moved
      their logic into the state machine.
    - Removed now-unused is_first / is_last methods from Iterator and
      simplified remaining logic to suit.
    - Removed now-redundant t/140-varsource.t.
    - Implemented TAP version syntax.
    - Tidied TAP::Harness::Compatible documentation
    - Removed redundant modules below TAP::Harness::Compatible
    - Removed unused compatibility tests

0.50_07 5 March 2007
    - Fixed bug where we erroneously checked the test number instead of number
      of tests run to determine if we've run more tests than we planned.
    - Add a --directives switch to 'runtests' which only shows test results
      with directives (such as 'TODO' or 'SKIP').
    - Removed some dead code from TAPx::Parser.
    - Added color support for Windows using Win32::Console.
    - Made Color::failure_output reset colors before printing
      the trailing newline.
    - Corrected some issues with the 'runtests' docs and removed some
      performance notes which no longer seem accurate.
    - Fixed bug whereby if tests without file extensions were included then
      the spacing of the result leaders would be off.
    - execrc file is now a YAML file.
    - Removed white background on the test failures.  It was too garish for
      me.  Just more proof that we need better ways of overriding color
      support.
    - Started work on TAPx::Harness::Compatible. Right now it's mainly just
      a direct lift of Test::Harness to make sure the tests work.
    - Commented out use Data::Dumper::Simple in T::Harness.pm - it's not
      a core module.
    - Added next_raw to TAPx::Parser::Iterator which skips any fixes for
      quirky TAP that are implemented by next. Used to support
      TAPx::Harness::Compatible::Iterator
    - Applied our version number to all T::H::Compatible modules
    - Removed T::H::C::Assert. It's documented as being private to
      Test::Harness and we're not going to need it.
    - Refactored runtests to call aggregate_tests to expose the
      interface we need for the compatibility layer.
    - Make it possible to pass an end time to summary so that it needn't
      be called immediately after the tests complete.
    - Moved callback handling into TAPx::Base and altered TAPx::Parser
      to use it.
    - Made TAPx::Harness into a subclass of TAPx::Base and implemented
      made_parser callback.
    - Moved the dispatch of callbacks out of run and into next so that
      they're called when TAPx::Harness iterates through the results.
    - Implemented PERL_TEST_HARNESS_DUMP_TAP which names a directory
      into which the raw TAP of any tests run via TAPx::Harness will
      be written.
    - Rewrote the TAPx::Grammar->tokenize method to return a
      TAPx::Parser::Result object.  Code is much cleaner now.
    - Moved the official grammar from TAPx::Parser to TAPx::Parser::Grammar,
      provided a link and updated the grammar.
    - Fixed bug where a properly escaped '# TODO' line in a test description
      would still be reported as a TODO test.
    - Added patches/ExtUtils-MakeMaker-6.31.patch - a patch against EUMM
      that makes test_harness use TAPx::Harness instead of Test::Harness
      if PERL_EUMM_USE_TAPX is true and TAPx::Harness is installed. In
      other words cause 'make test' for EUMM based models to use
      TAPx::Harness.
    - Added support for timer option to TAPx::Harness which causes the
      elapsed time for each test to be displayed.
    - Setup tapx-dev@hexten.net mailing list.
    - Fixed accumulating @$exec bug in TAPx::Harness.
    - Made runtests pass '--exec' option as an array.
    - (#24679) TAPx::Harness now reports failure for tests that die
      after completing all subtests.
    - Added in_todo attribute on TAPx::Parser which is true while the
      most recently seen test was a TODO.
    - (#24728) TAPx::Harness now supresses diagnostics from failed
      TODOs. Not sure if the semantics of this are correct yet.
      
0.50_06 18 January 2007
    - Fixed doc typo in examples/README [rt.cpan.org #24409]
    - Colored test output is now the default for 'runtests' unless
      you're running under windows or -t STDOUT is false.
      [rt.cpan.org #24310]
    - Removed the .t extension from t/source_tests/*.t since those are
      'test tests' which caused false negatives when running recursive
      tests. [Adrian Howard]
    - Somewhere along the way, the exit status started working again.
      Go figure.
    - Factored color output so that disabling it under Windows is
      cleaner.
    - Added explicit switch to :crlf layer after open3 under Windows.
      open3 defaults to raw mode resulting in spurious \r characters input
      parsed input.
    - Made Iterator do an explicit wait for subprocess termination.
      Needed to get process status correctly on Windows.
    - Fixed bug which didn't allow t/010-regression.t to be run directly
      via Perl unless you specified Perl's full path.
    - Removed SIG{CHLD} handler (which we shouldn't need I think because
      we explicitly waitpid) and made binmode ':crlf' conditional on
      IS_WIN32. On Mac OS these two things combined to expose a problem
      which meant that output from test scripts was sometimes lost.
    - Made t/110-source.t use File::Spec->catfile to build path to
      test script.
    - Made Iterator::FH init is_first, is_last to 0 rather than undef
      for consistency with array iterator.
    - Added t/120-varsource.t to test is_first and is_last semantics
      over files with small numbers of lines.
    - Added check for valid callback keys.
    - Added t/130-results.t for Result classes.

0.50_05 15 January 2007
    - Removed debugging code accidentally left in bin/runtests.
    - Removed 'local $/ = ...' from the iterator.  Hopefully that will fix the
      line ending bug, but I don't know about the wstat problem.

0.50_04 14 January 2007
    - BACKWARDS IMCOMPATIBLE:  Renamed all '::Results' classes to '::Result'
      because they represent a single result.
    - Fixed bug where piping would break verbose output.
    - IPC::Open3::open3 now takes a @command list rather than a $command
      string.  This should make it work under Windows.
    - Added 'stdout_sterr' sample test back to regression tests.  IPC::Open3
      appears to make it work.
    - Bug fix:  don't print 'All tests successful' if no tests are run.
    - Refactored 'runtests' to make it a bit easier to follow.
    - Bug fix:  Junk and comments now allowed before a leading plan.
    - HARNESS_ACTIVE and HARNESS_VERSION environment variables now set.
    - Renamed 'problems' in TAPx::Parser and TAPx::Aggregator to
      'has_problems'.

0.50_03 08 January 2007

    - Fixed bug where '-q' or '-Q' with colored tests weren't suppressing all
      information.
    - Fixed an annoying MANIFEST nit.
    - Made '-h' for runtests now report help.  Using a new harness requires
      the full --harness switch.
    - Added 'problems' method to TAPx::Parser and TAPx::Parser::Aggregator.
    - Deprecatd 'todo_failed' in favor of 'todo_passed'
    - Add -I switch to runtests.
    - Fixed runtests doc nit (smylers)
    - Removed TAPx::Parser::Builder.
    - A few more POD nits taken care of.
    - Completely removed all traces of C<--merge> as IPC::Open3 seems to be
      working.
    - Moved the tprove* examples to examples/bin in hopes of them no longer
      showing up in CPAN's docs.
    - Made the 'unexpectedly succeeded' message clearer (Adam Kennedy)

0.50_02 06 January 2007
    - Added some files I left out of the manifest (reported by Florian
      Ragwitz).
    - Added strict to Makefile.PL and changed @PROGRAM to @program (reported
      Florian Ragwitz).

0.50_01 06 January 2007
    - Added a new example which shows to how test Perl, Ruby, and URLs all at
      the same time using 'execrc' files.
    - Fixed the diagnostic format mangling bug.
    - We no longer override Test::Builder to merge streams.  Instead, we go
      ahead and use IPC::Open3.  It remains to be seen whether or not this is
      a good idea.
    - Fixed vms nit:  for failing tests, vms often has the 'not' on a line by
      itself.
    - Fixed bugs where unplanned tests were not reporting as a failure (test
      number greater than tests planned).
    - TAPx::Parser constructor can now take an 'exec' option to tell it what
      to execute to create the stream (huge performance boost).
    - Added TAPx::Parser::Source.  This allows us to run tests in just about
      any programming language.
    - Renamed the filename() method to source() in TAPx::Parser::Source::Perl.
    - We now cache the @INC values found for TAPx::Parser::Source::Perl.
    - Added two test harnesses, TAPx::Harness and TAPx::Harness::Color.
    - Removed references to manual stream construction from TAPx::Parser
      documentation.  Users should not (usually) need to worry about streams.
    - Added bin/runtests utility.  This is very similar to 'prove'.
    - Renumbered tests to make it easier to add new ones.
    - Corrected some minor documentation nits.
    - Makefile.PL is no longer auto-generated (it's built by hand).
    - Fixed regression test bug where driving tests through the harness I'm
      testing caused things to break.
    - BUG:  exit() values are now broken.  I don't know how to capture them
      with IPC::Open3.  However, since no one appears to be using them, this
      might not be an issue.

0.41  12 December 2006
    - Fixed (?) 10-regression.t test which failed on Windows.  Removed the
      segfault test as it has no meaning on Windows.  Reported by PSINNOTT
      <link@redbrick.dcu.ie> and fix recommended by Schwern based on his
      Test::Harness experience.
      http://rt.cpan.org/Ticket/Display.html?id=21624

0.40  05 December 2006
    - Removed TAPx::Parser::Streamed and folded its functionality into
      TAPx::Parser.
    - Fixed bug where sometimes is_good_plan() would return a false positive
      (exposed by refactoring).
    - A number of tiny performance enhancements.

0.33  22 September 2006
    - OK, I'm getting ticked off by some of the comments on Perl-QA so I
      rushed this out the door and broke it :(  I'm backing out one test and
      slowing down a bit.

0.32  22 September 2006
    - Applied patch from Schwern which fixed the Builder package name (TAPx::
      instead of TAPX:: -- stupid case-insensitive package names!).
      [rt.cpan.org #21605]

0.31  21 September 2006
    - Fixed bug where Carp::croak without parens could cause Perl to fail to
      compile on some platforms. [Andreas J. Koenig]
    - Eliminated the non-portable redirect of STDERR to STDOUT (2>&1) and
      fixed the synchronization issue.  This involves overridding
      Test::Builder::failure_output() in a very sneaky way.  I may have to
      back this out.
    - Renamed boolean methods to begin with 'is_'.  The methods they replace
      are documented, deprecated, and will not be removed prior to version
      1.00.

0.30  17 September 2006
    - Fixed bug where no output would still claim to have a good plan.
    - Fixed bug where no output would cause parser to die.
    - Fixed bug where failing to specify a plan would be two parse errors
      instead of one.
    - Fixed bug where a correct plan count in an incorrect place would still
      report as a 'good_plan'.
    - Fixed bug where comments could accidently be misparsed as directives.
    - Eliminated testing of internal structure of result objects.  The other
      tests cover this.
    - Allow hash marks in descriptions.  This was causing a problem because
      many test suites (Regexp::Common and Perl core) allowed them to exist.
    - Added support for SKIP directives in plans.
    - Did some work simplifying &TAPx::Parser::_initialize.  It's not great,
      but it's better than it was.
    - TODO tests now always pass, regardless of actual_passed status.
    - Removed 'use warnings' and now use -w
    - 'switches' may now be passed to the TAPx::Parser constructor.
    - Added 'exit' status.
    - Added 'wait' status.
    - Eliminated 'use base'.  This is part of the plan to make TAPx::Parser
      compatible with older versions of Perl.
    - Added 'source' key to the TAPx::Parser constructor.  Making new parsers
      is now much easier.
    - Renamed iterator first() and last() methods to is_first() and is_last().
      Credit:  Aristotle.
    - Planned tests != tests run is now a parse error.  It was really stupid
      of me not to do that in the first place.
    - Added massive regression test suite in t/100-regression.t
    - Updated the grammar to show that comments are allowed.
    - Comments are now permitted after an ending plan.

0.22  13 September 2006
    - Removed buggy support for multi-line chunks from streams.  If your
      streams or iterators return anything but single lines, this is a bug.
    - Fixed bug whereby blank lines in TAP would confuse the parser.  Reported
      by Torsten Schoenfeld.
    - Added first() and last() methods to the iterator.
    - TAPx::Parser::Source::Perl now has a 'switches' method which allows
      switches to be passed to the perl executable running the test file.
      This allows tprove to accept a '-l' argument to force lib/ to be
      included in Perl's @INC.

0.21  8 September 2006
    - Included experimental GTK interface written by Torsten Schoenfeld.
    - Fixed bad docs in examples/tprove_color
    - Applied patch from Shlomi Fish fixing bug where runs from one stream
      could leak into another when bailing out.  [rt.cpan.org #21379] 
    - Fixed some typos in the POD.
    - Corrected the grammar to allow for a plan of "1..0" (infinite stream).
    - Started to add proper acknowledgements.

0.20  2 September 2006
    - Fixed bug reported by GEOFFR.  When no tap output was found, an
      "Unitialized value" warning occurred.  [rt.cpan.org #21205]
    - Updated tprove to now report a test failure when no tap output found.
    - Removed examples/tprove_color2 as tprove_color now works.
    - Vastly improved callback system and updated the docs for how to use
      them.
    - Changed TAPx::Parser::Source::Perl to use Symbol::gensym() instead of a
      hard-to-guess filehandle name.

0.12  30 July 2006
    - Added a test colorization script
    - Callback support added.
    - Added TAPx::Parser::Source::Perl.
    - Added TAPx::Parser::Aggregator. 
    - Added version numbers to all classes.
    - Added 'todo_failed' test result and parser.
    - 00-load.t now loads all classes instead of having individual tests load
      their supporting classes.
    - Changed $parser->results to $parser->next

0.11  25 July, 2006
    - Renamed is_skip and is_todo to has_skip and has_todo.  Much less
      confusing since a result responding true to those also responded true to
      is_test.
    - Added simplistic bin/tprove to run tests.  Much harder than I thought
      and much code stolen from Test::Harness.
    - Modified stolen iterator to fix a bug with stream handling when extra
      newlines were encountered.
    - Added TAPx::Parser::Iterator (stolen from Test::Harness::Iterator)
    - Normalized internal structure of result objects.
    - All tokens now have a 'type' key.  This greatly simplifies internals.
    - Copied much result POD info into the main docs.
    - Corrected the bug report URLs.
    - Minor updates to the grammar listed in the POD.

0.10  23 July, 2006
    - Oh my Larry, we gots docs!
    - _parse and _tap are now private methods.
    - Stream support has been added.
    - Moved the grammar into its own class.
    - Pulled remaining parser functionality out of lexer.
    - Added type() method to Results().
    - Parse errors no longer croak().  Instead, they are available through the
      parse_errors() method.
    - Added good_plan() method.
    - tests_planned != tests_run is no longer a parse error.
    - Renamed test_count() to tests_run().
    - Renamed num_tests() to tests_planned().

0.03  17 July, 2006
    - 'Bail out!' is now handled.
    - The parser is now data driven, thus skipping a huge if/else chain
    - We now track all TODOs, SKIPs, passes and fails by test number.
    - Removed all non-core modules.
    - Store original line for each TAP line.  Available through
      $result->raw().
    - Renamed test is_ok() to passed() and added actual_passed().  The former
      method takes into account TODO tests and the latter returns the actual
      pass/fail status.
    - Fixed a bug where SKIP tests would not be identified correctly.

0.02  8 July, 2006
    - Moved some lexer responsibility to the parser.  This will allow us to
      eventually parse streams.
    - Properly track passed/failed tests, even accounting for TODO.
    - Added support for comments and unknown lines.
    - Allow explicit and inferred test numbers to be mixed.
    - Allow escaped hashes in the test description.
    - Renamed to TAPx::Parser.  Will probably rename it again.

0.01  Date/time
    - First version, unreleased on an unsuspecting world.
    - No, you'll never know when ...