Strange problem with solver

[Marco Polo]

I am facing a strange problem with solver (Plus42 1.2.6).
I was playing with solver using the equation 'x^3+2*x^2-5*x-5', which has three real roots: 1.931, -0.837, -3.094

After finding the first solution, for example 1.931 using 2 as first estimate, the solver seems to ignore any new estimate (i.e. figures next to other solutions)
and remain stuck on the first solution found.
The only way is to get the solver working is to delete variable X.

The solver is in "Numeric" mode. Same behaviour if set "Direct".
Equation is in STD mode
4-level stack enabled.

Any suggestion?
Thank you

[Vincent Weber]

That's normal. The actual 27S has the same behaviour, I have just confirmed it. 

You need to enter 2 estimates in a row, not just one. If you enter only one, the root you've just found is still considered an estimate, and of course a very good one, so it is returned... 

[Marco Polo]

Ok, i got the point, thank you

I was misled by the different behaviour of HP48 solver, which does not rely on previous result.

Also 15c has 48's behaviour, at least with this equation.

35s needs to store one estimate in X variable.

200lx solver behaves like p

Plus42's

[Thomas Okken]

Note that the two starting guesses do not have to be different, so if you want to provide just one guess, while making sure you're overwriting the previously found root, you can simply enter the guess twice. So, in the original example, if you want to look for the second root with a starting guess of -1, you can do -1 X ENTER X X.