What is the portable 128-bit floating point type?

[Diane Bruce]

On Sun, May 26, 2019 at 12:02:34PM +0200, Dimitry Andric wrote:
> On 26 May 2019, at 02:20, Yuri <yuri at rawbw.com> wrote:
> > 
> > On 2019-05-25 14:30, Dimitry Andric wrote:
> >> But from clang's OSTargets.h file, it looks like 128 bit float support
> >> is never set to enabled for FreeBSD.
> > 
> > 
> > If clang supports __float128, and clang is FreeBSD's main compiler, shouldn't it be enabled on FreeBSD then?
> 
> It probably should, but those kinds of things don't just magically appear
> out of thin air: Somebody(TM) has to do the actual work to implement the
> support, test it, maintain it, and so on.  Apparently there has not yet
> been any great need for this floating point precision.
> 
> Are you volunteering? :)

I've been slowly looking at it already in my copious spare time.
It's really something we should be supporting. The libgcc_s wrong version
problem is being dealt with by others by linking gfortran with a static
version of libgcc_s (!). (As discussed with Cy at BSDCan 2019)
See https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=208120#c24
specifically comment 35

> 
> -Dimitry
> 
> P.S.: I believe one of the first requirements is a <quadmath.h> header.
> 

- Diane
-- 
- db at FreeBSD.org db at db.net http://www.db.net/~db