Tabling: another core dump

General
1

Running this example from Stack overflow:


:- table posInt_CollatzSteps/2. 
posInt_CollatzSteps(I,N) :-
   (  I == 1
   -> N = 0                                                % base case
   ;  1 is I /\ 1
   -> I0 is I*3+1, posInt_CollatzSteps(I0,N0), N is N0+1   % odd
   ;  I0 is I>>1,  posInt_CollatzSteps(I0,N0), N is N0+1   % even
   ).

i0_i_maxSteps0_maxSteps(I0,I,M0,M) :-
   (  I0 > I
   -> M0 = M
   ;  posInt_CollatzSteps(I0,N0),
      I1 is I0+1,
      M1 is max(M0,N0),
      i0_i_maxSteps0_maxSteps(I1,I,M1,M)
   ).

And this query:

2 ?- time(i0_i_maxSteps0_maxSteps(1,1 000 000,0,MaxSteps)).
[Thread 1 (main) at Wed Jun 19 22:46:21 2019] ../src/pl-trie.c:1234: find_var: Assertion
 failed: index == state->max_var_seen+1
C-stack trace labeled "assert_fail":
  [0] save_backtrace() at :? [0x7f09721abadf]
  [1] __assert_fail() at ??:? [0x7f09721e01c6]
  [2] unify_key() at :? [0x7f09721d9026]
  [3] trie_gen() at :? [0x7f09721db239]
  [4] pl_tbl_answer_update_dl2_va.lto_priv.0() at :? [0x7f09721db474]
  [5] PL_next_solution() at ??:? [0x7f09722685bc]
  [6] query_loop() at :? [0x7f097222d192]
  [7] prologToplevel() at :? [0x7f097222d26b]
  [8] PL_toplevel() at ??:? [0x7f0972279cae]
  [9] swipl(+0x1075) [0x55d3ede7b075]
  [10] __libc_start_main() at ??:? [0x7f0971f97ee3]
  [11] swipl(+0x10be) [0x55d3ede7b0be]
[1]    23511 abort (core dumped)  swipl

It works fine without tabling (albeit it takes a very long time).

3

With latest git (8.1.8-25-g583b8636c):

103 ?- time(i0_i_maxSteps0_maxSteps(1, 1 000 000, 0, R)).
% 72,659,302 inferences, 12.005 CPU in 12.061 seconds (100% CPU, 6052541 Lips)
R = 524.

104 ?- time(i0_i_maxSteps0_maxSteps(1, 1 000 000, 0, R)).
% 5,000,001 inferences, 1.165 CPU in 1.169 seconds (100% CPU, 4290451 Lips)
R = 524.

These patches also fix Tabling: Core dump

4

Great! Tremendous response time. I think you surpassed Linus on this one :slight_smile: