Re: [java] RFA: builtins vs. -findirect-dispatch

[Andrew Haley <aph at redhat dot com>]

Andrew Pinski writes:
 > 
 > On Jul 13, 2006, at 8:33 PM, Andrew Haley wrote:
 > 
 > > Andrew Haley writes:
 > >>
 > >> It would be better for calls such as Math.sin() to map directly to
 > >> sin(); the indirection does no good at all.  But it's no big deal.
 > >
 > > On second thoughts, forget that: sin() is defined in libgcj.
 > 
 > On third thought, the sin in libgcj should really be only used for
 > StrictMath which currently does not use builtins anyways :).

I don't think so.  Even Math has strict implementation specifications
that aren't guaranteed by simply calling some random libc.

We need to drop the floating-point functions from libgcj on systems
(such as x86_64) that are known to have high-quality native
floating-point functions in their libc.  However, that is something of
a digression.

Andrew.