...
Resolution: Done, but there still is a problem finding data/caldb/software/tools/caldb.config. Program looks relative to supersede root only. File exists under base (ScienceTools installation) directory.
More run-time issues
commonUtilities
The caldb problem above is actually more pervasive. commonUtilities routines need to be enhanced for support of installations with a supersede directory.
Proposed actions: Rework commonUtilities with as little impact on clients as possible. Unfortunately it's likely that client code will also need to be changed if it is to work properly in all situations.
Wrong python in path
At run time python was resolving to default installation rather than "our" python. Jim tracked this down to unfriendly behavior of one of the set-up files asp_setup.sh calls: it resets PATH.
Resolution: Split asp_setup.sh into two pieces, one run before invoking _setup.sh and the other after. Committed to CVS.
Over-resolution of nfs path
In case an installation at SLAC included a supersede directory, environment variable for base directory if in nfs space was being resolved to something including a server name like "sulkey35". At a later date the server might no longer be serving that particular file so the path should be represented by something starting /nfs...
Resolution: Use utility supplied by Jim to translate back to /nfs... path. Committed to CVS.
PFILES behavior
ASP jobs may run without an afs token in which case they can't write to $HOME. Old PFILES-handling code would make a pfiles subdirectory in $HOME if it didn't already exist. It's bad form for programs run in batch to access $HOME anyway, so Jim proposed eliminating all PFILES handling in _setup.sh for batch processes.
Resolution: Compromised by getting rid of references to $HOME for batch jobs. The PFILES path in this case will include pfiles directory of base installation, pfiles directory of supersede area (if there is one), prepended with process's original value of PFILES (if any). Soon to be committed to CVS.