As I've repeatedly said, I don't even like the 82C55. It's too weak to do really useful work (though no weaker than they typical 805x), and it's too slow for use with some devices, though my current 805x of choice is not one of them.

There are a few, and I'd emphasize FEW, places where it's optimal, in the context of 805x applications, mainly because of its density, and those would be cases such as the one I mentioned before, where one has to have many dozens of outputs and inputs, all working within the confines of a single task, and all processed as a group, but with relatively precise timing (remember those control valves in the nuclear reactor?). If you have, say, 64 remote valves you have to open or close, and two local indicators for each, it's pretty easy to see how you might benefit from using 8 82C55's to do that.

I've never said that 82C55 was better than something else, though I have said that it's easier to use the 8255 that's already on the board than to add something else.

Now, parallel I/O is where ouputs change and inputs are sampled BYTEwise. Serial I/O is where outputs change and inputs are sampled BITwise. I think there's room for debate about whether a serial device to which the MCU writes or from which it reads in serial format but whose inputs and ouputs are sampled or driven bytewise are serial or parallel. I'm not suggesting that we have this debate now, but I do want to point it out, since remote parallel I/O has long been accomplished with serial means. It can also be argued that bit manipulation, particularly as done in the 805x, where port bits can be addressed directly, is not BYTEwise.

My only purpose in beating on Erik every time he says that using an 82C55 is wrong, categorically, is to remind him that, in this regard, HE is wrong.

I don't think anyone can infer from my comments over the years that I'm unwilling to compromise. In fact, I've made no statement from which I should retreat at all, as I've merely pointed out that Erik's position is too narrow and interferes with progress in many learners' instances every time he asserts it.

As for the 8255 being old, well, yes, it's old. So am I. But nobody seems to object to the use of the 1N914A diode. They've been around much longer than the 8255. I'd point out that both continue to do what they've always done, and quite well.

RE