"quadb.pl+transf dict"

- just prolog form of list prolog with fn'ism, large db of preds x 1-level deep preds x fnism
- run all finds preds for which data, not name is known *1
- findall, maplist-done?
- enter ?([1,1],[2]) for ?=+,
a([1,1],[b]),?([b,1],3).
- to find which sub-fn is missing

rec'n
- enter preds:
- +(a([],b,b),
a(a,b,c),":-", (
	a=[d|a1], %% x head, tail
	append(b,[d],e),
	a(a1,e,c))).

x just convert p->lp, *1, run, add new part to to db, add alg by itself to file
- only new part is run all
- different algs which parts of are taken from
- alg guesses next line of spec set by decomp *2

	- finds with each time *2

- fn'ism means var name->fn name x just with findall,maplist
	- with a number of levels, it could take too long x limit fn'al levels
- needs [],"",_

- x use existing caw*, specs for algs and supercomputer