I wonder if there is a certain bus contention problem at power-up, when at first moment port lines are undefined. You know, mcu needs a certain time for start-up oscillator, and as long as reset mode isn't invoked port lines are undefined. As consequence two or more devices could enable their outputs at the same time! Did you ever see bus contention at this moment?

An interesting point and, I'm sure, valid for the old non-static derivatives. With a short thought of how to do a '51 in a CPLD (I have absolutely no intention of doing that) I would connect the FF reset to the reset pin for the ports.

I have never designed to avoid contention before all port pins are a 1 and never had a problem.

I'll forward this post to a Philips guy supposed to know such things and hope he chimes in

Erik