In that case will a fixed shunt regulator at 6.8V allow rapid-enough charging to maintain the battery at full charge? I think something more elaborate is called for. Perhaps a 7V reg which shuts off, or reduces the regulated volts, once the charge current has dropped down to 20-30mA, then turns on/reverts to 7V again when the battery voltage has dropped down to 5.5V or so?
It's really hard to say, but here is my thinking... The panel can only put out 3 watts at high noon on a sunny day so it won't be able to charge a dead battery in one day no matter what we do. By the time you add something like a 317 and a diode to keep the battery from flowing backwards into the panel almost 30% of the power would be lost to heat not to mention the dropout voltage that would have the effect of shortening the charge time. I guess we could build a MPP charge controller and do a little better or maybe a buck boost with some battery management, but is it worth the complexity and expense?
The simple circuit would get the battery back to about 80% about as fast as anything and for less money