On May 8th, 2020, Herb J. wrote:

This reply is long, because I'm arguing in opposition, and that means I have to make a case about, and explain about. But I'll save some people, some time. I"m going to fuss about these microcontrollers becoming obsolete. If you don't care about that, save time and stop reading here.

[Herb's cogent argument elided here]

And in closing ....

The reason I'm familiar with this kind of "these go obsolete" arguments, is because I've seen this show before. Over the last few decades, various modern means to read ancient floppy diskettes have been produced. They have become in-favor and discussed and distributed. Then the developers lose interest; then the users lose interest; and they go out of production and use. Over geologic time (several years), these things simply come and go.

So the vintage computers remain, but these modern devices do not. That's about as simple as I can put the argument.

I can't materially dispute any of this; in fact, it neatly lays out something that has bothered me for years without my having been able to work out why. The obvious next question, then, is -- to borrow the words of Charles Schultz -- "Now that we know that, what do we do?" It looks to me as though the common thread, amongst all of these unintendedly ephemeral solutions, has been their use of "easily available" hardware which failed to stay that way. Obviously, then, solutions of this type must be implemented on hardware which is widely available from multiple independent sources, _and_ will be straightforward to substitute for when that eventually does change. We use this stuff in support of systems which remain usable, if not always all that use_ful_, for decades. We already know what stays around, or at least remains well-documented and -supported. By sheer weight of installed base, it would be very hard to lose all support for x86 or 6502 software. Equally, despite their clumsiness, I find it very difficult to believe that we will suddenly become unable to substitute newer devices for EPROMs; hell, such substitutes are already made by at least two unrelated sources just amongst the retrocomputing community, let alone industry at large. The only real irritant, then, becomes working out how to cheaply do whatever it is these other projects are using FPGAs for. Those are the parts that get updated whenever someone works out a better technique, but... as a general rule, the interface to whatever vintage hardware is more or less invariant, correct? That would mean the alterable functions can be carried out in software rather than in (redefinable) hardware. Flash updates are still flash updates, regardless of whether they contain computer programs or FPGA configuration files. Can anyone poke holes in this? Gordon S.