PASFRG  minutes 3/23/98



Shareware strategy: Eileen Berman

SDSS Software for Data Processing

Combination of freeware, commercial, and home made SW; was a success
~30-40 packages
6 different pipelines in data processing area, a framework, all pipelines called astrotools, lots of freeware 
used in framework.  Commercial Objectivity database, not in data processing area.
Each of freeware products has homemade wrapper on top.
Largest amount of home made code in astrotools and dervish.
Always chose possibility of free when available instead of commercial.
Wide variety of freeware tools: math, astronomy, plotting (pgplot), tk, graphics (Vogl), scripting language, 
cvs, web, X for displays.  Some very successful,, others not as much
General comments:
- freeware useful if stable or easy to get updates
- must see all source code
- overacrching standards important, difficult to enforce them (so choose only a few, and try to find ones 
which tools for enforcing them
- should have done requirements analysis first
- would be nice to have a person overseeing whole process
- bad to pretend SW development is a democracy (this is worse in C++)
- Regression tests are essential, and they need to be started in concert with the coding
PASFRG requirements:
- Data Access: science database
- Data Selection: chains
- Data Analysis: pipelines
- Category X (Tcl, fits)
- Data Presentation (Pgplot, Fsao, Vogle, C2html)
- Miscellaneous: 
Got away without requirements cause a lot of the components (fsao, fits, pgplot) were already well known 
and used in astrophysics community.
FNAl responsible for 500K out of 1M LoC; only wrote 80K of thr 500K.
Was important to have the source code, which could preclude commercial products.
http://www-sdss.fnal.gov:8000/  follow links to pipelines and data processing