Partial solution for bug 1452341: solve(x^(5/2)+1,x)

>>>>> "Richard" == Richard Fateman <fateman at cs.berkeley.edu> writes:

    Richard> I'm pretty sure that you will not be able to confirm all the roots
    Richard> that are produced by solve, by using ratsimp.
    Richard> (Sorry).

Yes, I understand that.

This particular bug only applies to equations of the form a*f(x)^n+b,
so it's a very specific equation.  It may not be possible to confirm x
is a root, but if maxima can, that would be useful.  The issue is what
to do with the roots it can't confirm.  If these roots are placed on
*fail, it just shows up in the output of solve anyway, except you
can't tell maxima couldn't solve it.  

It would be nice if solve returned 2 lists:  One list for the roots that
maxima is pretty sure are roots, and another for roots that maxima is
not sure about, or cannot solve.  Currently maxima returns sqrt(x)=-1
as one of the roots.

Or maybe just ignore this bug completely.

Ray