Software packages for the reduction and analysis of astronomical data are mostly developed at institutes related to space or ground based observational facilities. Most well known reduction packages, like Midas, Iraf, Xanadu, etc. were, at least initially, developed by scientists, from a scientific point of view, and not by software engineers. This is, until today, for most packages reflected in their architecture. Which means that they each have their own, very specific requirements about how and where they should be installed on a computer, or a cluster of computers. Often the reasons behind a particular set up are quite valid, and they can certainly be understood from a historical point of view, but they do not always coincide with what is practicable at a random institute, where these packages have to be installed. Crucial points in the development of many packages were the migration from VAX VMS to Unix based platforms, and from single station, stand-alone operation to operation in network environments with clusters of workstations.
Installation of these software packages at any given astronomical institute is often a laborious and pain-staking operation. Trial and error, and re-compilation of parts of the system, will eventually lead to a working local implementation of the package. But this practice is likely to compromise the integrity of the package, jeopardizing the reliability of the scientific results.
Many of the installation problems might be overcome if the astronomical software packages were made to comply to the standards as set forth by the GNU project. The GNU (Gnu's Not Unix) project was initiated by Richard Stallman, and it is aimed at developing and distributing free, sharable software. The GNU standards, or rather the principles underlying the standards, make sure that software packages complying to them are highly portable and easy to install. Where most astronomical software packages generally allow for only one, very rigid installation configuration, GNU software allows for almost every configuration one can think of: single user, multi user, special account, etc. Moreover, the installation of the software is automated. When the automatic installation procedure completed itself successfully, the chances of errors in the local implementation are minimal.
Many institutes are left to themselves when it comes to the installation and maintenance of astronomical software packages. In the United Kingdom a unique project was initiated in 1980. It is called STARLINK, and it looks after the installation and maintenance of many astronomical software packages for the astronomical community in the UK. Institutes that join STARLINK automatically receive updates of software overnight, without local system managers having to worry about each individual package or update. It hardly needs pointing out that this allows for a much more efficient allocation of human resources.
Another problem is the availability of software packages for different platforms. Generally, a software package is developed on a particular piece of hardware, and depending on the demand, ports to other platforms are made. It is not always possible to persuade a vendor to make a particular piece of hardware available (at little or no cost!) for porting a particular software package to that platform. However, the quality of todays networks allows for `pooling' of hardware among astronomical institutes, thus allowing developers to remotely create a port on a hardware platform at a site which they may never visit in person.
The World Wide Web allows documentation of software packages to be made available on-line. At institutes where several scientists are using the same software package, the whereabouts of the paper manuals are always a problem. System managers at several institutes have made efforts to bring documentation on-line. Since these again are duplicate efforts, institutes and organizations that provide the astronomical software packages should be urged to also provide their documentation in an on-line version, when they are not already doing so.