Bug in Asksign, also in Askexp?



Raymond Toy <toy@rtp.ericsson.se> writes:

(C2) ?limitp:true;
(D2)                                 TRUE

>     Wolfgang> (C4) asksign(log(x));
>     Wolfgang> Is  x  positive or negative?
> 
>     Wolfgang> neg;
>     Wolfgang> (D4)                                  PNZ
> 
> Shouldn't maxima say something else since the sign of a complex number
> isn't well defined?

I should think that PNZ is just the sign of complete agnosticism while
any other result implies of course that the expression in question is
real (cf. the description of `sign' in "Definitions for Operators").

So the point is that PNZ is always a correct result ;-) The
disadvantage of this is that an opportunity for actually determining a
sign might be missed: Try for example `asksign(exp(x+%i*y))' with both
settings of `?limitp'.

Wolfgang
-- 
wjenkner@inode.at