The perennial i386 floating-point rounding issue...

[Stephen L Moshier]

> Or, maybe we are just required to recommend the user set the 64 bit
> rounding mode thang?

Or maybe we should offer a way to set it ourselves.

Some years ago one of the Next people implemented a way for gcc to
keep track of the fpu precision setting.  I don't know how well it worked
then or how practical it would be now, but I think many users would
be happier if double meant double and long double meant long double.

The vast majority of the user problems like this that I have analyzed
involve an innocent-looking program that does a compare operation
to an extra-precise register value.  In other words, the most common
problem is not that we truncate values but that we fail to truncate
values.