digitalmars.D - Floating point Rounding modes and exception flags. [blah blah blah]
Don Clugston <dac nospam.com.au> writes:
The updated docs for std.math explicitly refer to exception flags and rounding modes. (eg FE_TONEAREST in rndtonl(real x) ). Does this mean there's been some thought about how this will work? Currently this is arguably the very worst part of the documentation at present: ---------------- float.html ---------------- IEEE 754 floating point arithmetic includes the ability to set 4 different rounding modes. D adds syntax to access them: [blah, blah, blah] [NOTE: this is perhaps better done with a standard library call] Exception Flags IEEE 754 floating point arithmetic can set several flags based on what happened with a computation: [blah, blah, blah]. These flags can be set/reset with the syntax: [blah, blah, blah] [NOTE: this is perhaps better done with a standard library call] --------------- Will it be a library call, or done with properties? I would also like access to the precision control. It could be used to verify that code for reals will continue to function correctly with 64-bit reals as well as 80-bits (it would also allow some direct accuracy comparisons to demonstrate the merit of extended reals).
Sep 26 2005
"Walter Bright" <newshound digitalmars.com> writes:
"Don Clugston" <dac nospam.com.au> wrote in message news:dh8c0j$pue$1 digitaldaemon.com...Will it be a library call, or done with properties?
I don't know yet, haven't really thought about it. What do you think?
Sep 26 2005