I have code that depends on it, and it is a technique I used for years. I wrote an APL function editor in APL 30 years ago. I discovered the problem while entering the system into GNU APL. I used this code in IBM APL.
On Fri, May 23, 2014 at 11:20 PM, David B. Lamkins <dlamk...@gmail.com>wrote: > This is ambibiguous w.r.t. the two spec on which GNU APL is based. > > Page 120 of the IBM Reference Manual says: > > "Valueless Expression: If R is empty or represents a defined function or > operator without explicit result, R has no value." > > Page 135 of the ISO Spec says: > > "Informal Description: Z is the result of evaluating the character > scalar or vector B as a line of APL." > > > So the IBM spec allows an expression which doesn't return a value; the > ISO spec doesn't. > > Personally, I find the IBM definition to be more accommodating of the > behaviors I'd like to see. > > > > On Fri, 2014-05-23 at 15:24 -0500, Blake McBride wrote: > > )CLEAR > > CLEAR WS > > ∇fun > > [1] 'in fun' > > [2] ∇ > > ⍎'fun' > > in fun > > VALUE ERROR > > fun > > ^ > > > > >
