### 21 Custom KAPPA

21.2 Parameters
21.3 Commands

Kappa applications can be modified to suit your particular requirements. Since this document is not a programmer’s guide, instructions are not given here. Programmers should contact the author for details until a new Programmer’s Guide appears to replace the old SUN/101, which was a good summary of Starlink infrastructure libraries and programming.

All the source files can be found in /star/kappa/*.tar on Starlink machines. The /star path may be different outside of Starlink, so check with your computer manager. There is a separate tar file for each Kappa subroutine library (with a _sub suffix) and the interface files, with obvious names. The remaining files: the monolith routines, link scripts, include files, the help source, shell scripts, ICL procedures, and test data are in kappa_source.tar. There is also a Starlink standard makefile and mk script.

Many of the general-purpose subroutines which previously formed part of Kappa have now been moved into a separate software item called KAPLIBS (see SUN/238). Kappa itself now links against the libraries in KAPLIBS .

Here is a worked example. Suppose that you have _REAL-type datasets for which you want to compute statistics including the skewness and kurtosis. One way is to modify STATS. First to save typing define environment variables, say STAR and KAPPA and KAPLIBS to point to where the Starlink software, Kappa and KAPLIBS source is stored. Next we extract the source files to change.

% setenv STAR /star
% setenv KAPPA /star/sources/kappa
% setenv KAPLIBS /star/sources/kaplibs
% tar xf $KAPPA/kappa_sub.tar stats. % tar xf$KAPPA/kappa_ifls.tar stats.ifl
% tar xf $KAPLIBS/kapgen_sub.tar kpg1_statr.f % tar xf$KAPLIBS/kapgen_sub.tar kpg1_stdsr.f
% tar xf $KAPPA/kappa_source.tar kappa_link_adam We modify kpg1_statr.f to compute the additional statistics; kpg1_stdsr.f to list the statistics; stats.f to update the documentation, to use the revised argument lists of the subroutines, and to output the new statistics to parameters; and stats.ifl to add the output parameters. kappa_link_adam need not be modified, but it is needed during linking. Next some soft links to include files need to be made. % star_dev % ndf_dev % prm_dev % par_dev % kaplibs_dev For some other application and subroutines, you can find what is needed by trying to compile them and see which include files the compiler cannot locate. You then invoke the appropriate package definitions: pkg_dev, where pkg is the three-letter package abbreviation. Now compile the modified code. This is for OSF/1: % f77 -O -c -nowarn stats.f kpg1_statr.f kpg1_stdsr.f and this is for Solaris: % f77 -O -PIC -c -w stats.f kpg1_statr.f kpg1_stdsr.f The -nowarn and -w prevent warning messages appearing. And this is for Linux: % g77 -fno-second-underscore -O -c stats.f kpg1_statr.f kpg1_stdsr.f Now link the task to produce a new stats executable. % alink stats.o -o stats kpg1_statr.o kpg1_stdsr.o \ -L$STAR/lib ‘./kappa_link_adam‘

If you want to use Kappa subroutines for your own application here are words of warning: the code may undergo alterations of subroutine name or argument lists, and those without a pkg_ prefix will either be replaced or renamed. Therefore, you should copy the modules you need.

#### 21.2 Parameters

If you don’t like Kappa’s parameter defaults, or its choice of which parameters get prompted for and which get defaulted, then you can change them. Extract the interface file from
/star/kappa/kappa_ifls.tar to your work directory and make the required modifications, and then recompile it. See SUN/115 on the meanings and possible values of the fieldnames, and how to recompile the interface file. If you use ICL, you’ll need to modify a monolith interface file: $KAPPA_DIR/kappa_mon.ifl, kapview_mon.ifl or ndfpack_mon.ifl. Finally, you will need to specify a search path that includes the directory containing your modified interface file. % setenv ADAM_IFL /home/scratch/dro/ifls:/usr/local/kappa This asks Starlink programmes to look in /home/scratch/dro/ifls to find the interface file, and if there isn’t one to look in /usr/local/kappa. If the interface file search is unsuccessful, the directory containing the executable is assumed. Thus if you’ve not created your own interface file for a task, you’ll get the released version. Of course, once you have done this, the documentation in the application specifications will no longer be correct. #### 21.3 Commands There is an easier method of tailoring Kappa to your requirements. If you frequently use certain commands, especially those with a long list of keywords and fixed values, you can define some C-shell aliases or ICL symbols for the commands. Like the shell’s$HOME/.login, ICL has a login file. (See “ICL for Unix” Appendix in SUN/144, or SSN/64 for details.) If you add symbols to this file, each time you activate ICL these abbreviations will be available to you without further typing. What you should do is to create a login.icl in a convenient directory, and assign the environment variable ICL_LOGIN to that directory in your \$HOME/.login file.

It is possible to have several ICL login files, each for different work in different directories. Now to abbreviate a command you put a DEFSTRING entry into the ICL login file. For example,

DEFSTRING MYC{ON} CONTOUR CLEAR=F PENROT MODE=AU NCONT=7

defines MYC or MYCO or MYCON to run CONTOUR without clearing the screen, with pen rotation and seven equally spaced contour heights. The symbols are not limited to Kappa. Indeed they can include shorthands for shell commands. For example,

DEFSTRING DA ls -al

would make DA produce a directory listing of all files with sizes and modification dates.

You can check what the current login files are as follows.