that's been edited now ... it should read "... for NOT using an 8255 ..."

As for justifications, and I'm really NOT advocating that one should use the 8255, only that nobody's really presented a good reason not to do so, I'd say, first of all, that it works very well with the standard 8x51/52 at 12 MHz or slower and has port characteristics well matched to the 8x51/52. It's been used successfully for 30+ years, (I had two on my very first i8080 board back then.) and still does everything that they did back then, including generating timely interrupts and automatic handshakes from parallel I/O. They not only extend the parallel I/O capability of the MCU but add the automatic handshake that the MCU would otherwise have to perform in firmware, probably as an interrupt service.

The 8155 and 8156 are better suited, in that they have the address latch built in and therefore can do without any external decoding logic if one uses their internal RAM and uses address lines to decode the address space.

For loosely coupled I/O, e.g. something where a timely response isn't required, I suppose IIC is adequate, but this thing generates a closely coupled parallel port set that can do things you can't do as well with serialized I/O.

Just to put this matter to rest, I'd appreciate someone, who's more interested in this problem than I, promoting a solution that outperforms the 8255, uses less code to do so, and exhibits at least the same set of capabilities. True, one seldom needs them all, but it's certain that once you have something that can't do them all, the missing ones are the ones you need.

RE