Re: [Bug-apl] value error when using 'Cut'

[Juergen Sauermann]

Hi Elias, great idea, except that it conflicts with the NARS APL interpreter which had defined Rank with axis already (and, of course, differently). See http://lists.gnu.org/archive/html/bug-apl/2014-07/msg00042.html /// Jürgen On 02/07/2016 01:31 PM, Elias Mårtenson wrote: How about implementing a non-conflicting extension? I'm suggesting the following form as an alternative:       Z ← f ⍤[j] B       Z ← A f ⍤[j] B Regards, Elias On 7 February 2016 at 20:27, Juergen Sauermann <juergen.sauerm...@t-online.de> wrote: Hi Alex, not sure what   a ¯3⍤< m  is supposed to mean. According to the ISO standard the syntax for ⍤ is: Z ← f ⍤ j B           (monadic, page 124) or Z ← A f ⍤ j B         (dyadic, page 125) If you compare that with your example: a ¯3 ⍤ < m then the (expected value) j is the primitive function <, which triggers the VALUE ERROR. The fact that the caret points to a is not because a is the culprit, but because a is the left end of the phrase being reduced. Unfortunately the syntax in the ISO standard is somewhat ambiguous: j is a one, two, or three element vector, and B is the rest. Therefore it is sometimes impossible to decide where j ends and where B begins, and the examples for ⍤ in the ISO standard are in conflict with the IBM APL2 binding rules. This  conflict occurs only with ⍤ which - wise decision - is not implemented at all in IBM APL2. The conflict can be avoided by always putting j and B into separate variables. If you use literals for j or B, heaven forbid, then be prepared for fairly nasty error messages at times. /// Jürgen On 02/06/2016 05:17 PM, alexwei...@alexweiner.com wrote: Hi bug-apl, Why am I getting a value error here? It seems that the variable 'a' definitely exists:       a←2 2 ⍴2       m←4 4 ⍴⍳16       a 2 2 2 2       m  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16       VALUE ERROR       a ¯3⍤<m       ^       a 2 2 2 2 SVN 693 -Alex