Dumping long double

Eric S. Raymond esr at thyrsus.com
Sun Nov 5 04:02:29 UTC 2017

Hal Murray <hmurray at megapathdsl.net>:
> >> We ran for years without long doubles.  There were no destabilizing
> >> problems.
> > No, but changing the code *back* before 1.0 was a thing I didn't want to do.
> I was surprised you put that change in that close to 1.0.  Taking it back 
> should have been a simple revert.

The difference is that I thought the original move was a win...and I
wasn't wrong, exactly, it immunized us against overflows for at least
some hardware. It just wasn't the fully *cross-platform* win I
originally thought.

> > I rememenber now thinking that long double was a net gain, if imperfectly,
> > because on systems with full long double support it would address the
> > overflow issue, while leaving us no worse off than before elsewhere. 
> How much hardware has support for long doubles?

x86-64, for starters.

> What is the performance impact of no hardware support?

That I do not know.  But I'm not worried about it either - we're not doing
calculations in them at any kind of serious volume.
		<a href="http://www.catb.org/~esr/">Eric S. Raymond</a>

My work is funded by the Internet Civil Engineering Institute: https://icei.org
Please visit their site and donate: the civilization you save might be your own.

More information about the devel mailing list