Reading the Dictionary page, I quibble with the phrase 'may not produce 
correct or consistent results'.  If the results are inconsistent, so be 
it; but if they are not correct, you should fix them.  It would be 
better to omit 'correct or', since your position is that anomalies 
involving _. are not incorrect.

Henry Rich

Roger Hui wrote:
> Almost nothing involving _. can be considered a bug.
> The only exceptions are 128!:5 and certain primitives
> noted in http://www.jsoftware.com/help/dictionary/d031.htm .
> 
> 
> 
> ----- Original Message -----
> From: Mark Niemiec <[email protected]>
> Date: Monday, May 3, 2010 17:29
> Subject: [Jprogramming] Bug in Self-Classify
> To: J Programming Forum <[email protected]>
> 
>> I encountered the following behavior, which surprised me
>> (even though it follows from the strict definition of Nub in the 
>> dictionary):   ~. 1 _. 2
>> 1 2
>> So it appears that Nub not only removes duplicates, it removes all
>> indefinites as well.
>> From this, it is reasonable that:
>>    $= , _.
>> 0 1
>> However, if one uses an atom (which should produce the same result),
>> one sees this:
>>    $= _.
>> 1 1
>>    = _.
>> 1
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
> 
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to