Which is the correct order to generate WAM code for L0 program terms?

Question

In Hassan Aït-Kaci's "Warren's Abstract Machine: A Tutorial Reconstruction" Section 2.2, the orders for compilation of L0 queries are clear enough: registers must be allocated with left-to-right breadth-first search and code must be generated with left-to-right post-order depth-first search.

In Section 2.3, the order for register allocation (of L0 programs) is clear: left-to-right breadth-first search. The order for code generation isn't. With the only example given, I can't tell if I should use BFS or DFS to generate code.

Can someone give me the WAM code for the following L0 program?

p(q(r(a)),s(b)).

Please look at [this recommendation](http://stackoverflow.com/a/14536068/772868)!. — false, Oct 29 '16 at 12:18

score 2 · Accepted Answer · edited Oct 28 '16 at 21:13

2

I put your program in sample.pl:

$ cat sample.pl
p(q(r(a)),s(b)).

Using GNU Prolog, I then did:

$ gplc -w sample.pl

The following WAM instructions are then contained in sample.wbc:

clause(p(q(r(a)),s(b)),[
    get_structure(q/1,0),
    unify_structure(r/1),
    unify_atom(a),
    get_structure(s/1,1),
    unify_atom(b),
    proceed]).

edited Oct 28 '16 at 21:13

false

10,264
13
101
209

answered Oct 28 '16 at 20:55

mat

40,498
3
51
78

What is the difference to `glpc -W`, which I rather use to get `.wam`-files? – false Oct 29 '16 at 12:23

Which is the correct order to generate WAM code for L0 program terms?

1 Answers1