Overview

The process of developing code for doing data analysis or control can be broken down into a number of well-defined steps. The aim of this document is to describe these steps and to propose a standard way of doing them. Standardising procedures will make it easier for all programmers to share and exchange software.

The basic processes which can be identified in developing software are :

1.

editing - entering and modifying source code in human readable format in files,

2.

source code organisation - how to organise source code into subdirectories

3.

compiling - converting source code from human to machine readable format,

4.

make - a tool used mainly for managing the compiling process,

5.

debugging - that age old art of removing the creepy-crawlys from your software,

6.

source code management - keeping track of multiple versions of source code files,

7.

distributing binaries - the altruistic action of letting others benefit from your work,

8.

distributing source code - the even more altruistic action of letting others continue from where you left off,

9.

documentation - the hardest part of all (so they say) of putting into a human language what you have already put into machine language,

10.

backups - saving all that hard work.

This document will now treat each of these topics individually and make proposals of how they should be handled at the Hartebeesthoek Radio Astronomy Observatory (HartRAO) in the context of the NCCS and Linux. The proposals are valid for programs written for doing scientific data analysis, programs written for controlling the telescope, utility programs, test programs and programs written elsewhere and which are ported to HartRAO. Where differences exist between the different environments this will be noted.