If you bring all the leads to the same potential, there should be no current flowing in the motor. It's unlikely, however, that you can sink the +12 supply, so you must drive (let float) all the '2803 outputs.

The '2803 is an inverting open-collector darlington array, i.e. you drive the base high with your MCU and the outputs driving the motor go to low level when the inputs are high, or, viewed another way, the darlingtons conduct when their bases are at a high level. If you are setting your port to 0xFF, then the outputs will be LOW. The result will be that all the coils have current flowing. This will hold the motor in a static state, though perhaps not the one you want. Since there's resistance in the coils, there will be heat generated as power is dissipated in the coils.

Since the '2803 is an open-collector darlington array, it will not conduct when the inputs are low. As a result, with no path to GND, there will be no current flow, hence, no power dissipated in the motor.

The idle holding torque of the motor may be of some interest, though, if you rely on the motor to hold its position under load when the motor is idle.

RE