f=: -.&= will test for _. , but you should be careful
about calling it ISNULL. I believe there are other
nulls and if your data came from an external source
f may not test correctly for those.
If you want to be totally foolproof, research into
what 64-bit floats can be NaNs, and write a function
to look for those patterns in the 3!:1 or 3!:3
representations. The details are left as an exercise
for the reader.
----- Original Message -----
From: bill lam <[EMAIL PROTECTED]>
Date: Thursday, July 6, 2006 5:18 pm
Subject: Re: [Jprogramming] _. and/or
> R&S HUI wrote:
> > When an argument of a verb is _. , all bets are off.
> > ("All bets are off" is an idiom meaning: anything can
> > happen and don't blame me if it does.)
>
> What is the verb that test for _. which is guaranteed to work?
>
> isnull _. -> 1
> isnull 1 -> 0
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
