Lessons Learned from GIPSY

This page summarizes the experience with GIPSY including astronomical usage, design, maintenance etc.
-- HansTerlouw - 07 Jul 2006

Good Points

• Data structure . Supports multi-dimensional images for which headers are possible for every substructure (..., cube, plane, line, pixel). The user can address these substructures with an elegant syntax. Table support. The structure transparently supports multiple endian-nesses and data formats. (e.g. a data set created on a Sun can be used on a PC)

• User interface . Both interactive and batch work is possible. Multiple tasks can run concurrently.The user has full control over the task's execution and parameters. The keyword-based parameters can be specified and modified at any time. A task may prompt the user when a parameter is needed but not yet specified. Parameters can also be specified by any task at any time. Numerical parameters can be entered as expressions in which functions can be used, including functions which obtain data from the data structure. The parameters are integrated with a general event mechanism which in turn is integrated with GIPSY's gui elements.
  Batch jobs can be prepared using one of the scripting languages COLA and Python. The latter was recently added and is also suitable for writing general tasks.

• Support for units.

• Programmability. Tasks communicate with the environment (OS, user, data, etc.) through standard subroutines and functions which make it easy to write new tasks. Programming languages are C, Fortran and Python.

• Maintenance. When properly configured, an installation can automatically obtain updates without re-installing the whole package.

Bad Points

• Installation. There is no simple configure/make procedure.
• Data structure. The image size is currently limited to approx. 2 Gpixels.

to top