Least PC-Compatible MS-DOS machines?

The minor annoyuance is that you can't individually control the dirction of the port lines. You can set port A, port B, and each half of port C to be input or output (at least in the simplest no-handshake mode). The major problem is that any write ot the mode cotnrol register clears all output lines to zero. This is ridiculous. Fristly it prevents you cases. But worst than that it ven makes the intial setup of the system hard. At the time it was made, the most likely thing to drive from a noutptu port line was a TTL input. And as is well-known, TTL input float high, anf ae much easier to pul lhig htan to pull low. After a reset, all ports o nthe 8255 wwre set to be inputs, so were effectively flating. They could be trivially puleld up with a resistor if necessary. So after a reset, the TTL device driven by the 825 sees a l=high elve on its input. As soon as you write ot the mode register, that becomes a low. You can then set the lien how you like, but that low state has to extist, maybe only for a brief period. But your hardware hs to be able to handle it. On a more resonaly parallel prot IC, one where writign to a mode register does nto clear the outptu data registers, you can write 1's to output data registers (even when the lines are set to be inputs), then set the lines to be outputs. The external device will ways see ahigh level (firstly from the floating 'input port' and/or a pullup resistor, then the extrnal hardware cna be designed to have active low inputs, and there will be no glitches to worry it as the parallel port IC is set up. Btu alas that doesn't work with the 8255. The 8255 is proablay OK fror reading switches/keyboards and driving LEDs/displays, but a right pain to use for anything remotely complicated. What I want to know is why clearing the data registers upon writing to the mode register was regarded as a desirable thign to do. I cna think of no useful applciaiton for this -tony