730 lines of code

OK, the key change now is controlling TE syntax on either side of the general select/Insert. Any data coming back to the machine while running an sql will be encoded as an event mask, and matched against the prepared event masks to extract the micro actions. Prior to launching the statement, the launch has prepared opcodes, with event masks and micro actions at the ready. So, the general development in adding robustness to the match process, is to write a snippet of code for starting, and a snippet for acting. Writing code snippets, when the entire chain is running perfect, that is where we want to be. Then adding specialized overlays is a snap. This path is the longest, mainly because it uses sqlite inefficiently by unbinding variable from the select * which it uses in default. Who cares for now. I dumped the current undebugged code for the new gfun.

The approach again: 1) Set up events actions mapped to the match functionality. 2) Have code to insert these actions into the event list 3) Have code to implement micro actions when the even t happens. Since this process controls the syntax of TE, the event list variable can be bound to intelligent sql prior to launch. Hence, the code will lose about 20-30 lines from bind, as al the necessary ingredients are in the ready structure. The pinball architecture is everything from the event exec down, the even exec should typically execute five to ten lines of code per event.