Mercurial > octave-libgccjit
diff libinterp/corefcn/data.cc @ 18869:23681c9ea7ba stable
better guess if rem or mod could be zero (bug #42627)
* lo-mappers.h (xmod, xrem): Don't treat abs(x)<=1 specially.
Return 0 if x/y is assumed an integer.
* data.cc: New tests for rem and mod.
author | Olaf Till <i7tiol@t-online.de> |
---|---|
date | Fri, 27 Jun 2014 11:49:22 +0200 |
parents | 9d185537e5d1 |
children | 23e511f3395d |
line wrap: on
line diff
--- a/libinterp/corefcn/data.cc Fri Jun 20 18:55:38 2014 -0400 +++ b/libinterp/corefcn/data.cc Fri Jun 27 11:49:22 2014 +0200 @@ -722,6 +722,8 @@ %!error rem ([1, 2], [3, 4, 5]) %!error rem (i, 1) +# bug 42627 +%!assert (rem (0.94, 0.01), 0.0); */ DEFALIAS (fmod, rem) @@ -873,6 +875,9 @@ ## non-integer real numbers %!assert (mod (2.1, 0.1), 0) %!assert (mod (2.1, 0.2), 0.1, eps) + +# bug 42627 +%!assert (mod (0.94, 0.01), 0.0); */ // FIXME: Need to convert reduction functions of this file for single precision