From Stephen Jaffe (ExxonMobil Research)

Zark Newsletter Extracts, introduced by Jon Sandles

Transcribed from the printed issue; not yet reviewed. Read it in the PDF of the issue, page 25

Here is another Zark extract. Happily I can report that we have seen some interest in some of the challenges set in this regular feature and we have received excellent entries from Carlo Alberto and Stephen Jaffe. I have not had a chance to investigate how these compare to the Zark entries, but they look pretty good to me. If anyone fancies doing a more formal analysis, then please send it in.

From Stephen Jaffe (ExxonMobil Research)¶

In use at Mobil since 1974, modified several times as we migrated from APLSV to APLSF on DEC then to VMSAPL on VAX to APL*PLUS and finally to Dyalog.

     ∇ Z←X CMIOTA Y;I;B;C;M;N;L
[1]   ⍝ ⍳ For char matrices  (⎕IO is an implicit argument)
[2]    M←(¯2↑1 1,⍴X)⍴X ⋄ N←(¯2↑1 1,⍴Y)⍴Y   ⍝ Let arguments to be matrices
[3]    C←¯1↑(⍴M)⌈⍴N
[4]    N←(⌽C,1↑⍴N)↑N ⋄ M←(⌽C,1↑⍴M)↑M       ⍝ ... with the same no of cols
[5]    M←M⍪' '                             ⍝ Add row for "not found" case
[6]    Z←⍋I←⍋N⍪M                           ⍝ Sort (and grades) the together
[7]   ⍝ Z←⍋I←⎕AV⍋N⍪M      ⍝ Use this if APL version requires a collating seq.
[8]   ⍝ Z←⍋I←⍋⎕AV⍳N⍪M     ⍝ ... or this if it doesn't allow char grade (VSAPL)
[9]    B←I>1↑⍴N                            ⍝ Select the items ∊ M
[10]   L←(1↑⍴M)--⎕IO                       ⍝ Result if not found
[11]   Z←((B/I-1↑⍴N),L)[⎕IO+(+B)[(1↑⍴N)↑Z]]   ⍝ See each item of N after
                                             which item of M was gone
[12]   Z←L⌊Z+L×∨/M[Z;]≠N                   ⍝ If they are equal this is
                                           the result, if not give L
[13]  ⍝∘ (c)APLIT A FUB spi 13/09/1996 18:12             10363 D95
     ∇