Basically you can do it in one day, I suspect. Its straight forward.
Since it doesn’t have cut (!) and clause/2 you could do it in one day.
For example using the numbervars/3 trick.
The ZIP paper with the 7 instructions, has a subset 4 instructions
called “data” instructions, these are already good to unify with the
head of a clause and to instantiate a clause.
The rest of the instructions, 3 more, of the ZIP make the ZIP
interesting, because the ZIP shows how to realize last call
optimization. But you can also run code without them.
Compiling the 4 “data” instructions can be done with
numbervars/3 and utilizing DCG, taking the first solution
of data/3. Your data should not contain $VAR(n) already:
data('$VAR'(K)) --> {integer(K)}, [var, K].
data(C) --> {compound(C), functor(C, F, N)},
[functor, F/N],
{C=..[_|L]},
data_list(L),
[pop].
data(X) --> [const, X].
data_list([]) --> [].
data_list([X|L]) --> data(X), data_list(L).
?- T=f(X,g(a,X),b), numbervars(T, 1, _), data(T, L, []).
L = [functor, f/3, var, 1, functor, g/2, const, a,
var, 1, pop, const, b, pop]
The paper doesn’t show these things, in A Portable Prolog Compiler,
Clocksin et al. from 1983 we merely find only a description
of the intermediate language (IL) and its execution and less
how to compile it. But one can sense the vibe of the paper,
and the facination of having created something inbetween
interpreter and compiler. Whats also missing is the cut (!),
but the code is reversible. One can easily implement clause/2,
and turn the “data” back into a clause by “executing” it
in a special way so as to obtain the body instead running it.
BTW: Historical bit, the intermediate language was designed
for Prolog-X, right? But at one moment of time Prolog-X was
ditched, so we have only the intermediate language surviving.
Not quite, the software preservation group has more resources:
Prolog-X
“Prolog-X was designed just before I left Edinburgh in 1980.
[..] The first version of the Prolog-X system was written in
Pascal under VMS for the DEC VAX in 1982.”
https://softwarepreservation.computerhistory.org/prolog/#Edinburgh
You find in the above, that the intermediate language had even a
prototype adding a microcode for a hardware emulator ORION.
This was all around the time of the Japan Fifth Generation Project.