library(cont) semantics regarding jump target unification

[jjtolton]

:- use_module(library(cont)).


a :-
        write(a_enter),nl,
        reset(b, a_returned, _),
        write(a_end),nl.

b :-
        write(b_enter),nl,
        reset(c, b_returned, _),
        write(b_end), nl.

c :-
        write(c_enter), nl,
        shift(a_returned), %% SHOULD this return to a?
        shift(b_returned),
        write(c_end), nl.

?- a.
%@ a_enter
%@ b_enter
%@ c_enter
%@    false.

I'm curious if we think this should be the desired behavior or not for shift/reset. It probably makes sense but it does not allow for targeting reset points by using the second argument, as @triska mentioned here. It means there is effectively a 1:1 relationship with a shift and a reset.

The semantics of unifying against an arbitrary reset might be madness, but also it would be quite powerful. (I'm not sure Djikstra would agree).