An econometrician/statistician who began programming using Manchester autocode in the 50's.
Code in APL with quad-IO 1 could produce almost anything with quad-IO 0, so I used to have a lamp at the begining of many pieces of code. I did use both origins. I shifted to primarily using J early after its release. At that time I was teaching and students did find it 'different' to have indices starting with 0 but seemed to quickly learn to use it. However the common ways of labelling data whether from time series, experiments or surveys use sequences beginning with 1 (days, weeks, months, years, plots, individuals or firms ). It is not an accident that R uses 1 as the index for a first position. It may be very convenient for users - though perhaps less so for programmers. However I find J fine as it is with a fixed origin 0. For a language primarily oriented towards handling data as arrays it seems natural to refer to cells using the representation of the cell location in an arbitrary mixed number base, and for that you need origin 0. Fraser ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
