GammaLib

Currently the calibration database handling is not very clean.

GammaLib defines a GAMMALIB_CALDB environment variable that is setup by the GammaLib configuration script. ctools defines a CALDB environment variable that is setup by the ctools configuration script. The GammaLib GCaldb class handles GAMMALIB_CALDB and CALDB and gives precedence for GAMMALIB_CALDB over CALDB.

ctools should however not use CALDB, as this may result in conflicts with the Fermi ScienceTools.

ctools could define an own calibration database environment variable CTOOLS_CALDB, but then some specific code needs to be added to GammaLib to catch this. This isn’t very clean.

Alternatively, ctools could use the GAMMALIB_CALDB environment variable, yet there may be situations where ctools cannot install into the directory (for example when GammaLib was installed by an administrator, but ctools is installed by a user).

Maybe at the end we’re left with having only a single CALDB environment variable (an no GAMMALIB_CALDB anymore), as this is the standard anyways. The conflict with the Fermi ScienceTools arises when Fermi response functions are needed but not available in the CALDB directory. Once we package the relevant response functions with GammaLib, this should no longer be a serious problem.

Another issue is that the caldb path is not set properly so far. Actually, the expected structure

$CALDB/data/<mission>
$CALDB/data/<mission>/<instrument>
$GAMMALIB_CALDB/data/<mission>
$GAMMALIB_CALDB/data/<mission>/<instrument>

is not really used (the data directory is actually missing), and the responses classes do not use the appropriate GCaldb::path() method to access the data. This should be made compatible.