Note that, technically (at least by J's dictionary definition), the result of
hook =: 2 : '([: u v) : (u v)' + hook +: ([: + +:) :(+ +:) is not actually a hook. The dyadic definition is a hook, but the monadic definition is something else (it's a fork with an empty left tine). To avoid confusion you might want to use a different name for this? Still, mkerror seems like it might be handy in some circumstances. Thanks, -- Raul On Tue, Apr 29, 2014 at 8:04 PM, 'Pascal Jasmin' via Programming < [email protected]> wrote: > That is an interesting difference, that shows problem well: > > +: 2 : ('([: u v)';':';'(u v)') +: > +: +: > > but, the "right answer": > > hook =: 2 : '([: u v) : (u v)' > > +: hook +: > ([: +: +:) :(+: +:) > > does not pre-assume dyad: > > 2 + (hook+:) 2 3 > 6 8 > +: (hook+:) 2 3 > 8 12 > 2 +: (+hook) 2 3 > 6 8 > > Here is a cool haskell like error handling system: > > mkerr > =: (0&boxeach@:)( ::((13!:11 ; 13!:12)@:(''"_))) > chkerrA =: ((hook;)(hook}.))(^:(0 = >@{.@:])) > > +: mkerr 4 5 6 > ┌─┬───────┐ > │0│8 10 12│ > └─┴───────┘ > 2 +: mkerr 4 5 6 > ┌─┬─────────────────────────────────────┐ > │3│|domain error | 2 +:mkerr 4 5 6 │ > └─┴─────────────────────────────────────┘ > > mkerr behaves much like sdcheck (sockets). Either returns 0 and result, > or err number ; err text > > +: chkerrA +: mkerr 4 5 6 > 16 20 24 > +: chkerrA 2 +: mkerr 4 5 6 > ┌─┬───────────────────────────────────────────────┐ > │3│|domain error | +:chkerrA 2 +:mkerr 4 5 6 │ > └─┴───────────────────────────────────────────────┘ > > chkerrA either runs the u verb with the argument, or if argument is error, > returns the error. > > 2 + mkerr chkerrA +: mkerr 4 5 6 > ┌─┬────────┐ > │0│10 12 14│ > └─┴────────┘ > > > > ----- Original Message ----- > From: Aai <[email protected]> > To: [email protected] > Cc: > Sent: Tuesday, April 29, 2014 2:13:19 PM > Subject: Re: [Jprogramming] prob with redefining hook > > Follow me: > > > +: (hook +:)(hook +:) > (+: +:) +: > > > ((+: +:) +:) y > y (+: +:) +: y > 2 3 (+: +:) +: 2 3 > 2 3 (+: +:) 4 6 > > 2 3 +: 8 12 > |domain error > | 2 3 +:8 12 > > ah! dyadic +: needs binary arguments > > 0 1 +: 1 0 > 0 0 > > ... and the dyadic case works because it ends up with: > > 4 + 8 12 > 12 16 > > > > > > > On 29-04-14 15:09, 'Pascal Jasmin' via Programming wrote: > > hook =: 2 : ('([: u v)';':';'(u v)') > > > > no problems with dyad: > > > > 4 + (hook +:)(hook +:) 2 3 > > 12 16 > > > > but why does this not work? > > +: (hook +:)(hook +:) 2 3 > > |domain error > > > > ---------------------------------------------------------------------- > > For information about J forums see http://www.jsoftware.com/forums.htm > > -- > Met vriendelijke groet, > @@i = Arie Groeneveld > > > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm > > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm > ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
