Or, hmm, maybe since it's APL, you don't call f at all, you just replace missing keys in the output with ⍬ (zilde)?
-Dan -----Original Message----- From: Dan Bron [mailto:[email protected]] Sent: Wednesday, April 24, 2013 4:35 PM To: '[email protected]' Subject: RE: [Jprogramming] key Occurrence in x. It's not even a question. Anything else is heresy. If in this "other context" you have an analog to !. (maybe []FIT ?) then it would be valuable to admit something like f/.!.universe_of_all_possible_keys such that x f/.!.universe y returns a list of whose length, elements, and order are the same as (universe). If there are some keys in (universe) that are not in (x) then the corresponding f-bucket in x is fed an empty array with the same prototype as y (or an item of y). -Dan -----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Roger Hui Sent: Wednesday, April 24, 2013 3:05 PM To: Programming forum Subject: Re: [Jprogramming] key Thank you to everyone who responded. A further question: do you prefer that the cells be ordered by their occurrence in x, or do you prefer that they be sorted (by x)? On Mon, Apr 15, 2013 at 5:43 PM, Roger Hui <[email protected]>wrote: > I have a question concerning your use of *key*. Do you find yourself > wishing that in x f/.y, the f would be invoked with a left argument > which is key value corresponding to the items in the right argument? > This does not mean that the J definition or implementation will be > changed, but to find out that if I have an opportunity to > define/implement key in another context ;-) whether I should consider > changing it. > > ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
