Gnudl motivation

Gnudl should provide some of the capability of high level languages for data analysis and plotting (such as Yorick, IDL, Matlab, Octave).

The most powerful of these, IDL, is a tool used by many astrophysicists and other scientists to analyze their data. It is a very powerful kitchen-sink type of language ("a la" perl), with many features and much contributed code.

As an example, some features of IDL which make it quite useful are:

• The immediate availability of most of the graphing operations you might want to perform on your data. The defaults for these graphing operations are good, and they can be further customized.
• The facility with which operations can be applied to arrays: any scalar operation can be applied to an entire array, and that operation is optimized.
• The fact that the graphs in IDL are publication-quality.
• IDL is available on many platforms.

Most of these products have some limitations. A couple of them are:

• They invent special-purpose languages, and those languages are then the only language that can be used with that system. It would be good to have a widely accepted language like scheme, and a variety of other languages that can be run on top of scheme, so that the user has a choice.
• There is a nice modern way of writing documentation (TeXinfo) so that it is available in source form, can be transformed into embedded hypertext docs for dumb terminals and fancy displays, can be made into a web page, or printed as a high quality hardcopy manual.
• Command history editing is important, and the best way of doing that is probably the GNU readline library. That way a user has the full emacs or vi history searching/editing capability.
• [probably a quote from Abelson and Sussman; I don't quite remember]

  Programming languages should be designed not by piling feature on top of feature, but by removing the weaknesses and restrictions that make additional features appear necessary.

Another motivation for Gnudl is that the GNU scripting and extension language Guile is now available in an experimental release. We are convinced that Guile is an ideal language for Gnudl, since it is based on scheme, offers several (fast?) interpreters running on top of scheme, and incorporates the Tk toolkit.