Release 0.5-2:
  * Import "utils" to NAMESPACE.

Release 0.5-1:
  * Fixed build issue with x64 windows.
  * Update to latest pbdMPI/pbdSLAP flags/linking.
  * Large internal re-organization.
  * Added base.det().
  * Fixed some minor Fortran issues.
  * Deprecated dallreduce() R bindings.
  * Better string portability with pdlaprnt().
  * Add FC_LEN_T and USE_FC_LEN_T for C/Fortran char* length via size_t.
  * Add examples and return values.

Release 0.5-0:
  * Generalize pdgemm() wrapper.
  * Update to RNACI v0.5.0.
  * Added base.pdgelqf().
  * Added base.rpdorglq().
  * Exported native routines.
  * Fixed base.matexp() bug.
  * Fixed some compiler warnings.
  * Cleaned up some internal issues.
  * Initial support for BLACS contexts with multiple MPI communicators.
  * Fixed Windows build issue (Jeroen Ooms).
  * Fix OMP flags issue.

Release 0.4-6:
  * Fix indexing problem in pdsweep()

Release 0.4-5:
  * Rewrote sweeper.
  * Remove -w and fix compiler warnings.
  * Update internal RNACI.
  * Fixed several serious memory issues when computing eigenvalues.
  * Switch pdsyev() to pdsyevr().

Release 0.4-4:
  * Add custom distribution function to replace pdgemr2d() for most purposes.
  * Fixed pdgemr2d() bug.
  * Fix several issues with condition numbers.
  * Some small internal cleanup.

Release 0.4-3:
  * Add pigemr2d() wrapper.
  * Added g2lcoord()
  * Fixed indxg2p bug indexing bug.
  * Swap floor() to as.integer() in indexing functions.

Release 0.4-2:
  * Fixed major problem with internal BLACS grid management with MKL.
  * Switched to a header-only RNACI.
  * Fixed build issues on Solaris (wcc).
  * Added flag to disable src/export_blacs (needed for Intel mpi).

Release 0.4-1:
  * Fixed mkgblmat bug when using non-full contexts.
  * Fixed build issue with Intel compilers.
  * Updated README.md.

Release 0.4-0:
  * Changed base api.
  * Numerous improvements to documentation.
  * Switched to roxygen2.
  * Numerous source organization improvements.

Release 0.3-0
  * Fixed compiler warnings generated from calling Fortran routines.
  * Cleaned up R wrappers.
  * Fixed memory/gc issue in R_PDGELS().
  * Nasty ICTXT bug in descinit (notably affecting Cpdgemr2d) fixed.
  * Modified matrix exponential kernels to take p as argument.
  * Fixed OpenMP issues in Intel compiler suite.
  * Fixed portability issue in Makevars.in.
  * Stopgap fixed regression in MKSUBMAT.
  * Stopgap fixed regression in MKGBLMAT.
  * Updated vignette format.
  * Fixed author affiliation in vignette.
  * Linked with SEXPtools.
  * Refactored internal wrappers to use SEXPtools.
  * Fixed ownany() bug fixed that affected some grid shapes.
  * Vignette copyright updated.
  * PDGELS bug fixes.

Release 0.2-3:
  * Fixed errors in manual.
  * Fixed matrix exponental bugs.
  * Added matrix exponentiation drivers.
  * Added experimental pdsyevx wrapper.
  * Added driver for (non-symmetric) eigenvalue calculator.

Release 0.2-2:
  * Nasty init.grid() bugfix.
  * Bugfix for changing between BLACS and MPI process numbering.
  * Added experimental methods for level 2 distributed-distributed R-like BLAS.
  * Fixed initialization bug in pdgemr2d.

Release 0.2-1:
  * Added PDSYEV wrappers.

Release 0.2-0:
  * Created better separation between pbdBASE and pbdDMAT packages
  * Completely redid the internals.  Much less stupid/wasteful now.
  * Expose BLACS C extension; BLACS pointers now exportable.
  * Added pdsyrk wrappers.
  * Modified pdgemm wrappers.
  * Added new matrix norm and condition number routines.
  * Fix .Last.lib problem.
  * Fixed bug in SEXP wrappers of ScaLAPACK functions; passing 0-byte Ralloc'd vectors.
  * Fix multi-arch problem.
  * Add copyright to vignette.

Release 0.1-1:
  * Changed functions to write to new env rather than .GlobalEnv.
  * Made PBLAS/ScaLAPACK wrappers less stupid.
  * Fixed some nasty redistribution bugs.
  * Made lm.fit-related foreign language wrappers more efficient.
  * Significantly cleaned up all foreign language wrappers.
  * Cleaned up wrappers for singular value decomposition.
  * Distributed QR added.
  * Distributed linear least squares solver added for n>p case.

Release 0.1-0:
  * Added demos.
  * Modify "PDLAPRNT" function.
  * Added insertion methods.
  * Split 'pbdDMAC' into 'pbdBASE' and 'pbdDMAC'.
  * Delete configure.ac and use RmpiSPMD:::LdFlags() and others.
  * Rename to 'pbdDMAC'.
  * Link with RmpiSLAP.
  * Add guide, man pages, installation, ...
  * Add BLACS, PBLAS, ScaLAPACK.
