You've just focused on the wrong aspect of my suggestion.

I've "hammered" the notion of ONE HOST INTERFACE for all field-programmable devices. That way, Jethro, with his big screwdriver and 150 watt soldering gun, can still build the adapter in his garage without straining his limited supply of gray cells.

Application designers who want the end-user to be able to apply patches, field-fixes, etc, would have to ensure that their product provides the necessary features to allow this to be done through the "standard" host interface and via the vendor-provided software.

If a little 6-pin part requires a quadrillion-gate FPGA to support the ISP task, well, that has to be figured into the cost of using that device in that way, but the host interface remains the same, whether it's talking to a complex programming adapter or a simple one, or, for that matter, directly to the IC. New devices, of course, would have to have the required features built into their hardware so that the host interface could talk directly to their internals.

BTW, a decade or two ago, when Motorola was introducing their small-package MC68HC705 variants, the way in which they were programmed was by applying a supervoltage pulse to a pin that was normally used for something else, rather than having pins dedicated specifically for programming. Clearly there's a way to do all this without depriving the designer of useable pins.

RE