• On Linux (and Windows if possible), I would like to have the
    development environment create and install the executables exactly
    as delivered by the installer. This is crucial to be able to write
    and deploy Python scripts that drive the Science Tools for various
    analyses, that others may use, and that can be run as part of the
    Release Manager unit tests (i.e., for the sane package).
    As a developer on Linux, I would like to be able to type 
    make
make install
    and I (hope to) see the single bin directory populated with
    exactly the same executables (whether they be wrapper scripts or
    not) that one gets for that package from the installer.
  • For Python use, at least two environment variables must be set so
    that the shared libraries and Python modules are found. Currently,
    for using Likelihood from Python from an ST distribution obtained
    via the installer, they must be set like this: 
    setenv PYTHONPATH ${INST_DIR}/pyLikelihood/v0r1/rh9_gcc32:${INST_DIR}/pyLikelihood/v0r1/python:\
${INST_DIR}/lib:${GLAST_EXT}/ROOT/v4.02.00/root/lib:${PYTHONPATH}

setenv LD_LIBRARY_PATH ${GLAST_EXT}/ROOT/v4.02.00/root/lib:${GLAST_EXT}/cfitsio/v2470/lib:\
${GLAST_EXT}/cppunit/1.9.14/lib:${GLAST_EXT}/fftw/3.0.1/lib:${GLAST_EXT}/xerces/2.6.0/lib:\
${GLAST_EXT}/CLHEP/1.8.0.0/lib:${INST_DIR}/lib:${INST_DIR}/st_graph/v1r4p3/rh9_gcc32:\
{INST_DIR}/optimizers/v2r4p3/rh9_gcc32:${INST_DIR}/xmlBase/v5r2/rh9_gcc32
    CMT supports this already. This needs to be supported for the end
    user and enabled somehow (e.g., by creating a setup.csh) when I type
    "make install". It would also be good if 
    setenv PATH ${INST_DIR}/bin:${PATH}
setenv CALDB ${INST_DIR}/irfs/caldb/v0/CALDB
setenv CALDBCONFIG ${CALDB}/software/tools/caldb.config
setenv CALDBALIAS ${CALDB}/software/tools/alias_caldb.fits
    were enabled as well.
  • There must be a central python subdirectory, just as we have planned
    for bin, lib, and pfiles, to hold our Python modules.
  • No labels