When performing a control function, isn't the most general way of describing it simply,
"1. examine inputs
2. take appropriate action" ?

Now, consider that you have seven inputs, hypothetically, of course, and they're left-justifed on P1. You have a table, called JTBL, containing 128 addresses, each corresponding to a value that can occur on P1. Many of these input "states" might require the same action. Nonetheless, a specific response is provided in this table, for each possible input state.

STP1: Load DPTR with JTBL.
Now load A with P1,
then LJMP @A+DPTR

Of course, it's really using a dispatch table, isn't it? It's a really simple state machine, too, though.

If each routine to which the table contents point ends in LJMP STP1, then you have a simple state machine using a table lookup.

What could be simpler, or more efficient ... or more powerful. It's not "rocket science" but it works well and is trivial to debug.

RE