The use of options in Riemann

A Gauss procedure has a bare necessity of inputs, whereas more detailed information often is provided by global variables. In Riemann we have replaced SOME of these, namely those that has a yes-no character, with options. This means that there is one global string matrix, __options, which is read by all procs that accept options with the same result. For example, instead of having one global, _logged, which is 1 means that data is logged, and if it zero that data is not logged, we can let __options contain the string "LOG" which has the same effect.

Options are added and removed from __options by the procs

• ADDOPTION
• REMOVEOPTION
• CLEAROPTION

Many procs only looks for key parts of elements of __options. This means that you can often define an option in a rather lucid way, like "Logistic Regression", wheras the proc only looks for "LOGI" in upper cases. Therefor you can specify the analysis by the alternative name "Logit analysis" also if you want to.

The overall purpose with using options this way is to bring down the number of global variables needed. The back side of the coin is that you must keep control over what options are in effect, and which options affect a particular procedure.

Comments and suggestions, please mail: Anders Källén
Last modified: 98-09-12 9:00