Perhaps Bron could make available some test data. Miller's fsm is leaner but equivalent to mine, and I'm equally curious to know which results were not according specs.
R.E. Boss -----Oorspronkelijk bericht----- Van: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Namens Miller, Raul D Verzonden: dinsdag 6 februari 2007 7:09 Aan: Programming forum Onderwerp: RE: [Jprogramming] Efficiently converting fixed-width to delimited Dan Bron wrote; > If your code can produce the same results as the following in all cases, > it is correct: > > load 'strings' > fw2dl =: [: ; [: (dltb@:}: , {:)&.> e.&(10 9{a.) <;.2 ] NB. Assume y ends with LF I belive trim, as defined below, is equivalent to fw2dl for the cases where y ends with LF. If not, I'd be very interested in the case where this does not hold. NB. other SPACE DELIM s=:0 10#:10*".;._2]0 :0 2.1 0.0 1.1 NB. 0: skipping initial whitespace 2.2 0.3 1.2 NB. 1: found delimiter 2.0 3.2 1.2 NB. 2: found non-whitespace 2.2 3.0 1.1 NB. 3: potentially trailing whitespace ) trim=: (1;s;(a.=' ')+2*a.e.LF,TAB)&;: P.S. this post is meant to be read in a fixed-width font. My broken email client is not rendering it that way, but past experience leads me to believe that it will probably be distributed in a fashion which lends itself towards that end. If not, perhaps pasting the code into J will be adequate... -- Raul ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
