In case the issue is in the code for singletons, you should try both
^ _744.440
and
^ 2 # _744.440
There is a compiler switch to allow/disallow denorms, isn't there?
Henry Rich
On 9/1/2016 1:23 AM, bill lam wrote:
My test of win64 built with msvc failed
g200 ran under wine, (win32 failed more tests)
0 < ^ _744.440 NB. _1074*^.2
0
|assertion failure: RUN1
|[-21] Y:\jbld\test\g200.ijs
0 = ^ _744.440
1
0 = ^ _744.440
1
0 < ^ _1074*^.2
0
0 = ^ _1074*^.2
1
This g200 test ran fine in other platforms.
^ _744.44
4.94066e_324
This looks like a demormal number since range of
exponent should be e_308. I don't know whether
this is related to msvc or wine. Can anyone
confirm the result of ^ _744.44
on natural win64?
* https://en.wikipedia.org/wiki/Denormal_number
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
