Optimizing the Antecedent Precipitation Index

Transcribed from the printed issue; not yet reviewed. Read it in the PDF of the issue, page 7

Am I missing something, or is Bob Bernecky unaware of decode (see Vector Vol.13 No.4 pp.7-11)? This solves the above problem in six characters, viz: x⊥¨,\y and incidentally runs about four times as fast on APL2/PC as either of Bob’s api1 or api3. I described the analogous J version, namely >@(#.&.><\) , in Vector Vol.13 No.3 p. 12.

[Ed - see also Andrei Buzin’s note on page 140]