On Sun, Sep 09, 2007 at 09:19:05PM -0500, Joshua Isom wrote:
> On Sep 9, 2007, at 6:40 PM, Doug McNutt wrote:
>
> >At 21:16 +0100 9/9/07, Nicholas Clark wrote:
> >>On Sun, Sep 09, 2007 at 10:56:20AM -0700, Jrg Plate wrote:
> >>># <URL: http://rt.perl.org/rt3/Ticket/Display.html?id=45309 >
> >>>This patch implements the sign function for I, N, BigInt and Complex
> >>>numbers.
>
> I'm curious as to why these ops should be added. These things can
> easily be done with PIR, and any compiler can auto generate the code
> for that.
Not only that, but the C<cmp> op already does exactly what we want
(perhaps with the exception of negative infinity or NANs -- I'd have
to think about those).
cmp I0, N0, 0.0 # $I0 is -1, 0, or 1
At any rate, I'd be against adding a separate set of "sign" opcodes,
in favor of just using "cmp" or writing special-purpose subs to
handle it.
Pm
