Re: signbit and logical tests

[Daniel J Sebald]

On 02/09/2013 09:02 PM, Michael D. Godfrey wrote:
> On 02/09/2013 09:26 PM, John W. Eaton wrote:
> > On 02/09/2013 09:02 PM, Michael D. Godfrey wrote:
> > > I looked a bit more at the "logical" tests as described in Sections
> > > 4.6-8.
> > > If signbit() returned a logical it would fit with "isxxxx" functions and
> > > make more
> > > sense. Referencing Section 4.6 through 4.8 in the signbit()
> > > documentation
> > > would then make this all fit together and more understandable.
> > >
> > > Is this a useful idea?
> >
> > The signbit function in Octave is just a simple wrapper around the
> > system library signbit function, and that returns the value of the
> > sign bit as an integer. I don't know whether it would be a good idea
> > to change that definition for Octave. The other function that I know
> > of in Matlab that returns something different from what the C library
> > function returns is strcmp. So there is precedent for that kind of
> > thing. Or, if you want a predicate-style function, maybe we should
> > have isnegative instead?
> >
> > jwe
> This does not quite appear to be the case because this problem started
> with bug #38291 which
> showed that on at least one 32bit system (Ubuntu) the returned value for
> true is 512.
> That is why my test for == 1 failed.

Just curious how that is coming about.  Any guess?  I would think that C 
defines a logical true as 1.  If this where a sign bit test then 512 is 
the result

1000000000

Such a thing could result if the underlying assembly is using a register 
shift (albeit incorrect shift) to isolate the sign bit.

Dan