2023-11-01  Jaroslav Fowkes
Version 2.8.1
Bug fixes:
   * Fix dropped $FFLAGS in Matlab install scripts

2023-02-13  Jennifer Scott
Version 2.8.0
Update code and documentation:
   * info%maxfront statistic corrected (it holds the max front size, which is equal to
     the max number of rows in a supernode)
   * info%maxsupernode added to hold what was previously in info%maxfront, which is
     the max number of columns in a supernode
Make matlab version compatible with MacOS and Windows.

2022-12-06  Jaroslav Fowkes
Version 2.7.3
Update documentation:
   * Correct advanced features C spec example
   * Add advanced features Fortran spec example

2022-07-08  Niall Bootland
Version 2.7.2
Update documentation:
   * State METIS 5 incompatibility in documentation

2022-01-24  Jaroslav Fowkes
Version 2.7.1
Bug fixes:
   * Correct C header file info struct naming to reflect precision

2021-08-27  Andrew Lister
Version 2.7.0
Bug fixes:
   * Expose matching based scaling through the scale parameter of factor
     methods
   * Display warning for ignored scaling in control%scaling == 4 cases

2020-11-17  Jaroslav Fowkes
Version 2.6.0
Bug fixes:
   * MATLAB interface now compiles without OpenMP by default as Mathworks
     no longer support OpenMP under gfortran and this was causing crashes

2018-11-29  Lukas Schork
Version 2.5.0
Bug fixes:
   * Fix two memory leaks that occur when memory allocation fails.

2016-09-02  Jonathan Hogg
Version 2.4.0
Improved MATLAB support:
   * Provide access to control%solve_blas3 and control%solve_mf from MATLAB
   * Fix bug involving calculation of matching-based orderings from MATLAB

2014-02-12  Jonathan Hogg
Version 2.3.0
Performance improvements:
   * MATLAB interface now compiles with OpenMP enabled by default
Bug fixes:
   * Fix integer overflow issues that occur for some large problems
   * Create work around for ifort bug that affects solve
   * Create work around for ifort bug with OpenMP and vars named 'alloc'
   * Fix bug in integer kinds for matlab interface with 32-bit DI

2013-03-21  Jonathan Hogg
Version 2.2.0
Bug fixes:
   * Fix memory leak that occurs when a large matrix is found not to be +ive
     definite.
   * Fix a bug in the matlab install script that looks for no longer existant
     ddeps.f and zdeps.f files.
   * Correct upper limit for control%u to 0.5.
   * Add some explicit type casting to remove some warnings with g95.
     NB: g95 still fails to compile the code due to a compiler bug.

2012-12-20  Jonathan Hogg
Version 2.1.0
Several New Features + Bug Fixes + MATLAB interface update
   * Add routine ma97_solve_fredholm() to support Fredholm alternative for
     singular matrices.
   * Add routine ma97_lmultiply() to enable users to multiply by L and L^T.
   * Add routine ma97_sparse_fwd_solve() to enable efficient solution of
     Lx = b when b is sparse.
   * Add support for MC30 scaling.
   * Modify MATLAB interface to support above operations, also adding support
     for new scaling and matching operations from previous release.
   * Improve control%ordering=5 heuristic (differentiate it from ordering=6)
   * Bug fixes:
      * Propogate stats structure down to ldsrk() routine allowing correct
        checking for memory allocation error.
      * Fix error reporting bug where info%st != 0 but info%flag was 0.
      * Fix MATLAB install script to support major versions higher than 7.
      * Fix bug with control%ordering=6

2012-06-20  Jonathan Hogg
Version 2.0.0
   * Add support for matching-based orderings. This includes adding additional
     arguments (optional in Fortran, mandatory but optionally NULL in C) to
     ma97_analyse() and ma97_analyse_coord().
   * Change default behaviour of ordering heuristic to give better parallel
     performance for large and sparse problems. Original behaviour still
     available as control%ordering=6.
   * Improved handling of structually singular matrices when using MC64
     scaling option.
   * A structual singularity previously reported under warning +6 is now
     reported under warning +7 for reasons of consistency.
   * MC64 scaling now uses HSL_MC64. This has altered the permutation of
     the matrix that is used by the algorithm, which may alter the runtime
     significantly for some problems (either faster or slower). The same
     scaling should still be found for structually non-singular problems.
   * Fix bug with mc77 scaling.
   * Fix bug involving 64-bit addressing of large supernodes.
   * Fix memory leak when performing repeated factorizations.
   * Fix a bug involving zero pivots in small supernodes.
   * Fix a performance issue with node-level parallelism.
   * Fix a bug due to memory allocation failure that only occurs in parallel.

2012-01-31  Jonathan Hogg
Version 1.2.0
   * Fix incorrect calculation of info%num_two during factorization in real
     codes

2012-01-09  Jonathan Hogg
Version 1.1.0
   * Update documentation to cover:
      * control%factor_min
      * control%solve_mf
      * control%solve_blas3
      * control%solve_min
   * Add above controls to C interface.
   * Change control%solve_min from INTEGER to INTEGER(long)

2011-12-20  Jonathan Hogg
Version 1.0.0
First public release; fix bug in MC77 pivoting.

2011-11-23  Jonathan Hogg
Version 1.0.0-pre2
Significantly improved memory managment plus some moderate performance
enhancements.

2011-11-01  Jonathan Hogg and Jennifer Scott
Version 1.0.0-pre1
Initial version of code