This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Suffix for __float128 FP constants
- From: "H. J. Lu" <hjl at lucon dot org>
- To: Steve Ellcey <sje at cup dot hp dot com>
- Cc: ubizjak at gmail dot com, joseph at codesourcery dot com, gcc-patches at gcc dot gnu dot org
- Date: Mon, 25 Jun 2007 16:17:08 -0700
- Subject: Re: Suffix for __float128 FP constants
- References: <20070625230652.GA13392@lucon.org> <200706252313.QAA27721@hpsje.cup.hp.com>
On Mon, Jun 25, 2007 at 04:13:41PM -0700, Steve Ellcey wrote:
> > I said I preferred "Q" before. But if HP-UX compiler already uses
> > "L"/'l' for __float128, I think we should try to follow the precedent
> > when there is no standard.
> >
> > H.J.
>
> I think "L"/'l' is required to be 'long double' by the C standard
> regardless of what size long double is. Since, on HP-UX __float128 is
> long double, we can use the L's for it. We also allow "Q"/'q' as a
> suffix for the quad type and quad is defined to be __float128 too, same
> as the HP-UX (but not Linux) long double type.
>
> So I think we have:
>
> W - __float80
> Q - __float128
> L - long double (may be __float80, may be __float128)
In that case, Q/q is good to me.
Thanks.
H.J.