On Thu, May 26, 2011 at 11:02 AM, Andrew Nikitin <[email protected]> wrote: > Since we are on ? #. and x: subject, here is a trick question > > require 'plot' > t=:?10^100x > plot x;((100) 6!:2 ' #.inv t' ,~ ":)"> x=.10000+i:20 > > Why there is a performance hit at 10000 and not 10001, as one would expect if > extended integers were represented base 10000?
I am not sure why I should expect that. Ok, hypothetically speaking, if extended integers were represented in base 10000, the implementation could optimize for the 10000 #.inv case when dealing with extended integers. But it is not clear to me that that kind of optimization would result in people writing better programs in J. -- Raul ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
