> solve([a=b,b=3],[a]) produces "Maxima encountered a Lisp error:"
Please report what version you are running. I do not get this error
in 5.10.0/windows/gcl.
On the other hand, I do get the message "Inconsistent equations".
This is because Maxima assumes that all symbols which are not
mentioned in the second argument to 'solve' are arbitrary parameters,
not variables. The error message could perhaps be made better.
> I think solve should work also when the number of equations is different from the number of unknowns.
It does. So for example solve([x^2=y^2,y=a],[x,y]) => [[x = - a, y =
a], [x = a, y = a]].
In any case, it is *always* a bug when Maxima reports a Lisp error,
which indicates an internal error.
-s