#2685 radcan too ambitious

[Mikael Samsøe Sørensen]

(%i1) radcan(sqrt((5c/6-250/183)^2+(100/61-c)^2));
(%o1) (61c-100)/(6*sqrt(61))

The result could be negative, but it is a sqrt(...).

(%i2) factor(sqrt((5c/6-250/183)^2+(100/61-c)^2));
(%o2) abs(61c-100)/(6*sqrt(61))

Here we get the required abs(x) function.

The error appears regardless of the setting of radexpand

[Rupert Swarbrick]

This is actually the expected behaviour of radcan. Unfortunately, its documentation isn't all that brilliant at the moment - I think Raymond Toy has been trying to come up with a more formal specification. He explains what's going on in a similar bug report here: https://sourceforge.net/p/maxima/bugs/2479/

In this case, the point is that if c is large then 61 * c >> 100.