A bug with pstool, which involved ridiculously low precision of IEEE-754 doubles emmited in the output, has been fixed. Support for FFTW2 libraries compiled with single-precision will be dropped in the next 24 hours.
Currently, ability to perform pre-transform and post-transform filtering (such as windowing and windowing correction) is being added to pstool. "Filters" will be stand-alone shared object which can be loaded by pstool.
The documentation for pstool has reached a usable state. Even though no man pages are available yet, pstool is ready for public consumption.
Project pstool has reached what I would consider stable. Advertised functionality is present and found to be working (just fixed a bug that resulted in odd results). No file releases yet since I've yet to write the documentation. Coming soon!
The problem with the Debian MPICH is that there is no mpiCC/mpicxx/mpic++ - i.e. no wrapper around C++. This doesn't happen if you build MPICH from source.
Building PSTOOL with the Debian mpich and fftw-dev packages fails. The culprit was found in MPICH package.
The Debian MPICH's mpicc command behaves strangely, not like mpicc from stock MPICH from http://www-unix.mcs.anl.gov/mpi/mpich/ - the -c option has odd behavior and the command doesn't deal well with MPI headers. Unlike stock MPICH, this mpicc expects you to hardcode the actual paths to the headers. "#include <mpi.h>" doesn't work, and must be replaced with "#include <mpi/mpi.h>"... read more
Initial commit to CVS has been done. Functionality is provided and working. The calculated power spectrum is normalized to agree with Parseval's Theorem, and at this stage I'm trying to ensure sane results. Documentation is yet to be written. Until then there will be no public releases - please use your favorite CVS client to check out the latest source code.
Once functionality is ensured to be correct and ample documentation is provided, I will post tarballs, IA-32, AMD-64 and PowerPC binaries in the files section.