=head1 Math::GSL Changes

=head1 v0.42 - June 2020
    - Disabled two tests in SF.t and GSL.t that failed on some platforms.

=head1 v0.41 - June 2020
    - Added support for GSL 2.6 (hakonhagland)
    - Added docker file support for building a CPAN distribution
    - Added support for gsl_odeiv_system (hakonhagland)
    - Add support for Debian-specific version 1.16.1, which caused test failures

=head1 v0.40 - December 2018
    - added support for GSL 2.5 (djerius)
    - vast improvements to Travis CI test infrastructure (djerius)
    - added random method in Math::GSL::Matrix (Rui Meira)
    - added Math::GSL::Multiset (standard gsl API and Perl OO)
    - gsl_linalg_givens now works
    - gsl_vector_const_subvector is now exported properly
    - Support for Sparse Matrices was added
    - Documentation for Math::GSL::Rstat, Math::GSL::SparseMatrix,
    - Math::GSL::Multifit now requires GSL 2.1 and higher
    - Math::GSL::Rstat now requires GSL 2.1 and higher

=head1 v0.39 - September 2016
    - Fix Debian Bug reported by Niko Tyni
    - Extensive fixes to allow GSL 1.15 and GSL 1.16
      to continue to work with GSL 2.x
    - GSL 1.11 - 1.14 are no longer supported

=head1 v0.38 - September 2016
    - Fixes for compiling against various GSL
      versions (thanks CPANtesters!)

=head1 v0.37 - September 2016
    - Support for GSL 2.0, 2.1 and 2.2
    - Multilarge subsystem added for GSL >= 2.1

=head1 v0.36 - Fri Mar 25 23:32:03 PDT 2016
    - Fixed error messages on Matrix set_col and set_row
      methods (Jon Schutz)
    - Implemented array-based Bessel functions:

=head1 v0.35 - April 2015

    - Remotion of trailing whitespaces (Shlomi Fish)
    - Documentation fixes (Shlomi Fish)
    - Remotion of PkgConfig from runtime requirements
      (Bartosz Jakubski)
    - Indentation fixes (Bartosz Jakubski)
    - Installation feedback improvements (Bartosz Jakubski)

=head1 v0.34 - December 2014

    - Fixed bug on uncleared error variable (5.8.9)
    - Added OO transpose method for matrices (Ivan Baidakou)

=head1 v0.33 - September 2014
    - Added OO code for RNG->shuffle, RNG->choose and RNG->sample.
    - Complain when opening (fopen) a file from a tainted string.
    - Complex now has operator overloading for +,-, ==, !=, *, /
    - Added write/load methods to save a matrix to file.

=head1 v0.32 - June 2014
    - Correctly split the compiler name.
    - Added OO code to the QRNG module.
    - Fixed QRNG code that was failing on some machines.

=head1 v0.31 - June 2014
    - Fixed Matrix multiplication that was forcing square matrices.

=head1 v0.30 - June 2014
    - Fixed generated C files with wrong GSL version numbers.
    - Refactored Build.PL in order to use either gsl-config or PkgConfig.
    - Change Build.PL to return 0 when GSL library not found.
    - Use PkgConfig instead of ExtUtils::PkgConfig.
    - Math::GSL::Matrix constructor from a GSL matrix.
    - Reorganization on the Math::GSL::Matrix documentation.
	- Add symbols introduced on gsl 1.15.
	- Fixed in the correct way the prove/TAP::Harness problem.

=head1 v0.29 - June 2014
	- Fix issue when prove/TAP::Harness uses lib/ modules instead of blib/ ones.
    - Add functions introduced in gsl 1.16.
    - Fix issue when compiling on some Linuxes.
	- Math::GSL ->each method.
    - Math::GSL ->max and ->min methods.
	- Math::GSL ->dim method.
    - Math::GSL ->set_elem and ->get_elem methods.
    - Math::GSL concat_vertically and concat_horizontally.

=head1 v0.28 - Jan 2014
    - Fix build errors with clang
    - Improve POD to play nice with Perl 5.18
    - Teach Math::GSL about new functions added in GSL 1.16
    - Added 1D minimization tests

=head1 v0.27 - July 2012
    - Mathieu functions are now accessible
    - Improved detection of gsl
    - Fix bug in calling ->get() method on Math::GSL::Vectors
    - Greatly improved HTML formatting of documentation

=head1 v0.26 - September 2011
    - The Math::GSL::Min and Math::GSL::Roots are now functional - Yury Zavarin
    - Dirichlet random distribution support and greatly increased docs by Tom Nishimura <tnish@fastmail.jp>
    - gsl_version() function which returns the version of GSL being used by Tom Nishimura <tnish@fastmail.jp>
    - Fix '-bundle not allowed with -dynamiclib' build error on Mac OS X - Yury Zavarin
    - Build swig wrappers for different versions of GSL (1.11 - 1.15) - Yury Zavarin
    - Minimum SWIG version is now 2.x (only required for developers using the git repo)
    - Removed deprecated stuff from Build.PL and added Module::Build to configure_requires
    - Many small documentation improvements
    - Fix build on Windows - Sisyphus

=head1 v0.25 - April 2011
    - TODO RT#66882 properly so that a stable release with a passing test suite can be used.

=head1 v0.24 - April 2011
    - Fix https://rt.cpan.org/Ticket/Display.html?id=66882 (Tests fail on Perl's with DUSELONGDOUBLE)

=head1 v0.23 - March 2011
    - Fixed the tolerance of some tests that were causing CPANtesters failures
    - Fixed some doc typos

=head1 v0.22 - June 2010
    - Improved tests
    - == and != operators for Vectors
    - Fixed the export of hypergeometric functions (Thierry Moisan)
    - In instantiation of GSLBuilder in Build.PL, the value for the parameter
      include_dirs were corrected to an anonymous empty array, instead of an
      empty string.  (Jakob Ilves)
    - Operator overloading of vector equality (Thierry Moisan)
    - Various doc fixes (Pjotr Prins)
    - Typemap fixes (Vincent Danjean)

=head1 v0.20 - May 19 2009
    - $rng->get($n) will now return the next $n values of the random number generator
    - Fixed + added test for RT#45044 math-gsl eigenpair bug, reported with patch by Ian Malone
    - Added more tests for Randist
    - Fixed some incorrectly written tests for BLAS and Linalg
    - Improved NTuple support and tests
    - Improvements to callbacks, thanks to patches from piotrj
    - The following improvements were merged from Vincent Danjean's git repo:
    - Fixed some memory leaks caused by in-typemaps without argout-typemaps
    - Improved portable callback implementation
    - Correct handling of double, float and size_t on x64
    - Removed duplicate definition of fopen/fclose in VectorComplex
    - Fix bug where proper flags to SWIG were not passed

=head1 v0.18 - March 23 2009

    - Make Math::GSL play nice with GSL 1.12
    - Added swap() to Vector objects with tests and docs
    - Added p-norms to Vector objects via norm() and normalize()
    - Added operator overloading so that
            abs $vector == $vector->norm
    - Added as_vector() to Matrix and MatrixComplex objects
    - Added inverse(), is_square(), det(),  lndet(), zero() and identity()
        to Matrix objects
    - Added inverse(), is_square(), det(),  lndet(), zero(), identity()
        and hermitian() to MatrixComplex objects
    - Added dot product to Matrix objects
    - Fixed various typos in documentation
    - Fixed warnings about overloaded operators in Matrix and BLAS
    - Overloaded '==' and '!=' for MatrixComplex and Matrix objects
    - Fixed amd64 -fPIC compile failure
    - Added tests to Monte and refactor Sort tests
    - Refactored and improve error checking in callback interface
    - Fixed 'NaN' test failures (thanks CPANtesters!)

=head1 v0.16 - December 14 2008 (Happy Snowpocalypse PDX!)

    - Fixed RT#40947: configure_requires ExtUtils::PkgConfig bug
    - Added reverse() and swap() methods to VectorComplex objects with tests and docs
    - as_list() and get() now work on VectorComplex objects
        as_list() returns a list of Math::Complex objects
        get() returns a list of Math::GSL::Complex objects
    - Math::GSL::Test::ok_status() now takes an optional message argument
    - Added KNOWN_BUGS
    - Added Math::GSL::Test::is_status_ok()
    - New subsytems: VectorComplex and MatrixComplex
    - Operator overloading for addition and subtraction with vectors
    - Added example: examples/matrix/nonsymmetric_eigen
        Shows how to find the eigenvalues of a nonsymmetric matrix
    - Added copy() method to vectors
    - Improved Vector docs
    - Sped up vector dot products by 15x using BLAS function gsl_blas_ddot
        NOTE: using gsl_blas_ddot($x->raw,$y->raw) directly is still about 10 times faster
              than $x * $y due to error checking and function call overhead
    - Separated POD and Perl code out of SWIG interface files, yay for proper syntax
    - Added stub test files for Heapsort, IEEEUtils, Multimin, Siman, Wavelet2D
    - All test files now keep track of how many tests should run
    - Fix an include bug in Histogram2D which caused the build to fail (Thierry)
    - Operator overloading for addition, multiplication and subtraction on Matrix objects (Thierry)

=head1 v0.14 - October 21 2008

    - Chebyshev Series Approximation, with tests and docs
    - Examples and new documentation in Statistics, as well as tests (Thierry)
    - Bugfixes relating to compiling on 64bit Red Hat Linux, reported by Michael Roberge
    - Improved FFT, Deriv, Integration, Chebyshev, Combination and Roots docs
    - Documentation for Math::GSL::Test
    - Improved introduction examples in Math::GSL
    - Improved Minimization subsystem tests, but it is not functional
    - Added examples/fft/forward_real
    - Added example/deriv/basic

        This shows the simple fact that d/dx(sin(x)) = cos(x)

    - Added example/sf/erfc_check (thanks to Keith Lofstrom)

        Script which tests the erfc() special function against computing
        the integral definition with gsl_integration_qagiu()

    - Added example/vector/speed

        This shows a considerable performance boost using Math::GSL::Vectors
        instead of List::Util when searching for the min and max elements of
        large sets of random numbers.

    - Fixed return signature of gsl_deriv_* functions to return a flat list
    - Fixed some failing tests relating to  -Duselongdouble (Sisyphus)
    - Fix location of shared objects (Sisyphus)
    - Added raw() method to RNG objects

=head1 v0.12 - September 14 2008

    - Darwin support!
    - Minimum Perl version changed to 5.8.0
    - Make platform check compile time constants for performance
        aka The Wilhelm Speedup (thanks to Eric Wilhelm <scratchcomputing@gmail.com>)
    - FFT for real data
    - Numerical derivatives fully implemented
    - Integration subsystem, gsl_integrate_qags() and gsl_integrate_qagi(), etc..
        - QNG    : Non-adaptive Gaussian (uses fixed number of sample points in interval)
        - QAG(S) : Adaptive Gaussian (Singluar) (slices interval up based on properties of the function)
        - QAGI   : Adaptive Gaussian with infinite integration range
        - other are supported but not tested
    - Initial Monte tests
    - Initial Multiroots tests
    - More nan/inf fixes on MSWin32 from <sisyphus@cpan.org>
    - Stripping of binaries on Windows, which significantly reduces size
        from <sisyphus@cpan.org>
    - Full support and tests for all functions in the Sort subsystem
    - Added 'examples/benchmark/sort' to show performance of gsl_sort()
    - Added 'examples/clicker/chart' to show graphing of functions
    - Added 'examples/gsl_repl' which is a rapid prototyping tool/interactive interpreter
    - Compile support for GSL's as old as 1.8 (test suite still needs to be properly guarded)

=head1 v0.10

    - Continued porting to MSWin32, fixing many NaN and Inf handling issues

=head1 v0.08

    - Large documentation improvements
    - MSWin32 support thanks to testing and feedback from <sisyphus@cpan.org>

=head1 v0.07

    - Initial CPAN release

=head1 v0.042

    - Started change to Test::Class

=head1 v0.01

    - Added swig interface files and swig-ified build script
    - initial tests for special function library gsl_sf <--> Math::GSL::Sf
    - Proof Of Concept: makegsl and testgsl