Hi,

On 4/19/20 10:02 PM, Kacper Gutowski wrote:
On Sun, Apr 19, 2020 at 08:48:27PM +0200, Dr. Jürgen Sauermann wrote:
Not sure what;s wrong with your qt (character ?) since I don't have it.

This behavior is indicative of a NaN:

      ⊢NaN←12⎕CR17⎕CR'AAAgIAAAACAAAAABAgAAAAAAAAAAAPj/AAAAAAAAAAA='
0.000000000
      0×NaN
DOMAIN ERROR
      0×NaN
      ^ ^
      (⌈/⍳0) > NaN
1
      10⎕CR'NaN'
 NaN←0.0000000000000000
      2⎕TF'NaN'
NaN←0.0000000000000000


Which suggests there might be some bug in one of primitives used to compute that value as it shouldn't be possible to create NaN in APL by normal means and you should get a domain error somewhere earlier.

Correct. To fix this I would need to know how exactly qt was produced. Or, even better,
an APL 1-liner that produces the NaN (not using ⎕CR of course).
-k


Reply via email to