José -
Thanks for pointing out that issue in the first place.
I wasn't aware of the compiler module being overzealous sometimes ...
Are there any other known (I'm fighting for
words, might I call them) 'glitches' comparable to this one, and
where would they be documented for lookup..?
-M
At 2018-10-17 16:25, you wrote:
Martin,
In my example, using + instead of -, u was the double of a single roll and
v was the sum of two "independent" rolls.
Yours is a far better example for illustrating the issue,
( u=. 3 : ' T - (T=. ? y)' )
3 : ' T - (T=. ? y)'
u is one roll minus itself which produces zeroes; whereas
( v=. 13 : ' T - (T=. ? y)' )
? - ?
is the difference of two "independent" rolls.
Thus, for example,
u 0 0 0 0 0
0 0 0 0 0
whereas,
v 0 0 0 0 0
_0.0389238 0.321837 0.0862073 0.552116 _0.134629
Thanks! I had one of those moments after I read your posts... Why did I
not think of that? :)
On Wed, Oct 17, 2018 at 8:08 AM Martin Kreuzer <i...@airkreuzer.com> wrote:
> Addendum:
> Consider, in contrast, the behaviour of this *explicit* definition:
> ee=. 3 : 'T - T=. ?y'
> ee 10+i.7
> 0 0 0 0 0 0 0
> Same result as working *outside* a verb definition:
> ] T=. ? 10+i.7
> 6 6 0 9 12 1 5
> T - T
> 0 0 0 0 0 0 0
> -M
> At 2018-10-17 11:40, you wrote:
> >Let's modify Jose's example (by taking the *difference* of two
> >identocal terms):
> >
> >NB. this works as expected:
> >
> > d=. 13 : 'T - T=. +:y'
> > d 10+i.7
> >0 0 0 0 0 0 0
> >
> >NB. this doesn't:
> >
> > e=. 13 : 'T - T=. ?y'
> > e 10+i.7
> >_1 1 5 _5 8 _2 _4
> >
> >NB. this yields the same result (zero) for every argument:
> >
> > d
> >+--+-+--+
> >|+:|-|+:|
> >+--+-+--+
> >+: - +:
> >
> >NB. this yields arbitrary results (because minuend and subtrahend do
> >not necessaryly have the same value):
> > e
> >+-+-+-+
> >|?|-|?|
> >+-+-+-+
> >? - ?
> >
> >But ... from the underlying structure they should have.
> >
> >NB. in comparison this works as intended:
> >
> > f=. 13 : 'T - T=. ?.y'
> > f 10+i.7
> >0 0 0 0 0 0 0
> > f
> >+--+-+--+
> >|?.|-|?.|
> >+--+-+--+
> >?. - ?.
> >
> >-M
> >
> >
> >At 2018-10-17 10:22, you wrote:
> >
> >>Jose,
> >
> >>You said: You might be interested to know that 13 : '...'
> >>occasionally is too smart; some verbs, such as ?, which do not
> >>always produce the same result given the same argument, sometimes
> >>do not mix well with 13 : '...'; for example,
> >
> >>Something is not wrong when verbs with ? don't agree. I would
> >>worry is they did agree. They areproducing different results each
> >>time they are used.
> >
> >>Linda
> >
> >
> >>-----Original Message-----
> >>From: Programming <programming-boun...@forums.jsoftware.com> On
> >>Behalf Of Jose Mario Quintana
> >>Sent: Tuesday, October 16, 2018 7:37 PM
> >>To: Programming forum <programm...@jsoftware.com>
> >>Subject: Re: [Jprogramming] Using rank to avoid looping - reference?
> >
> >>Linda,
> >
> >>You might be interested to know that 13 : '...' occasionally is too
> >>smart; some verbs, such as ?, which do not always produce the same
> >>result given the same argument, sometimes do not mix well with 13 :
> >>'...'; for example,
> >
> >> ( u=. 3 : ' T + (T=. ? y)' )
> >>3 : ' T + (T=. ? y)'
> >
> >>and
> >> ( v=. 13 : ' T + (T=. ? y)' )
> >>? + ?
> >
> >>do not mean the same process.
> >
> >>I hope it helps
> >
> >----------------------------------------------------------------------
> >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
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
