E. does use Boyer-Moore. However, +./@:E. or E.i.1: exits on finding the first match so can be much faster; at the very least, they do not have to construct the boolean vector first. See Appendix B <http://www.jsoftware.com/help/dictionary/special.htm> of the J dictionary.
x=: 'abra' y=: 1e6$'abracadabra ' timer=: 6!:2 10 timer '(x E. y)i.1' 0.00988289 10 timer 'x (E.i.1:) y' 4.66504e_6 A=: 'not in ' 10 timer '(A E. y)i.1' 0.00573889 10 timer 'A (E.i.1:) y' 0.00455509 On Thu, Aug 28, 2014 at 4:23 PM, Henry Rich <henryhr...@nc.rr.com> wrote: > Remember that x +./@:E. y, x i.&1@:E. y, etc use Boyer-Moore, so > potentially much faster. > > Henry Rich > ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
