There's still a large class of problems that can be addressed with a solution of this type. The problem, for example, of needing a programmer to put the code inside the MCU. While those of us fortunate enough to live in western Europe or in the U.S. or in OZ, can probably afford to buy a programer for each and every variant we settle on for our application(s), there are lots of folks, particularly newbies and students in the developing world, who need to spend their resources on other things, like food and rent, and books and tuition.

If they build a circuit, it's painful because the first have to get the materials. Consequently, they don't buy materials separately for each potential application, but, rather, they try to do things in a way that will help solve a number of problems. Their instructors try to assign construction problems that will meet the needs of the majority of the remaining assignments in their courses, rather than doing as those in the U.S. often do, assuming all their students have an infinite reserve of cash. They assign homework based on the application circuit in the textbook, which uses an old i8032 and a 2764 and a 6264 along with a little "family" logic, since they have an old Data I/O model 29B in the lab, and that can be used to program the 2764's.

Now, is this optimal? Of course not, but it CAN work, and if it does, a good bit of learning about MCU application can be learned from it. Would you want someone to follow the same approach in a commercial setting? Certainly not! Nonetheless, they have to start with something.

So, is it worth it? Well, for educational purposes, I'd say absolutely, "YES!"

BTW, I'd point out, also, that many of the SBC's listed on this site do, in fact, include an 8255 or two, suggesting that they're not totally useless yet.

RE