HP50g : Symbolic solver oddities or user errors?

[Vtile]

Another interesting "feature" of the HP50g CAS I just found out when eyeballing my old math book with 50g on the another hand. If we enter an equation of (1+j)z = 2 - j and then solve the X when the comp..calculator is on exact mode, rectangular coordinates, complex on, radians on, flag 27 set ('X+*Yi'->'X+*Yi').

When solving (SOLVE) the X from simple complex equation (1 + j)*z = 2 - j
calculator gives out: X=(1-3j)/2, this is right solution, but wrong representation as it should be 'X+*Yi' or in this case 'X/a+*(Y/a)i' as set on the calculators settings.

Now if I take a numerical solution of X (->num) it turns out a mathematically right solution in totally wrong form as it gives
(0.5,-1.5) which is not a 'X+*Yi'.

Much confusing and trust eating that the calc doesn't seems to give a damn about settings (a flags) I order it to follow.