Forum Navigation
math.IEEE.feqrel and float
Moderators:
kris
Posted: 05/25/08 12:44:53 Modified: 05/25/08 12:47:25It seems that feqrel only supports mantissa sizes of 106, 113, 64 and 53. At one stage it must have supported floats (mantissa==24) because of the assert:
static assert(is(X==real) || is(X==double) || is(X==float), "Only float, double, and real are supported by feqrel");Yet forcing feqrel to accept a float triggers another assert:
assert(0, "Unsupported: " ~ std.string.toString(X.mant_dig));which prints "Unsupported: 24"
What was the reason for removing native float support? Can this be added in easily? If not, then the first assert should be updated.
Thanks in advance
Microbox
Author Message
Posted: 05/25/08 13:57:14Hi microbox,
could you please create a ticket on this topic? I agree that it sounds like float should be supported here.