Fixes multiple heishamon rules bug

[IgorYbema]

See commits for reasons

[CurlyMoo]

This fails this unittests i made for another Heishamon issue before:

  { "on foo then max(1, 2); end on timer=1 then if $a > 0 then foo(); end max(1, 2); end", { { "", 108 }, { "[1]$b = 2[1]$a = 2", 218 } }, { { "", 167  }, { "[1]$b = 1[1]$a = 1", 151 } }, 0 },

[IgorYbema]

that is why 218 should now read 219

[CurlyMoo]

That’s not the only issue with that unittest. The output is also wrong.

[IgorYbema]

Confirmed the rule text never assigns to $b at all — $b doesn't appear anywhere in on foo then max(1, 2); end on timer=1 then if $a > 0 then foo(); end max(1, 2); end — and $a is only ever read (if $a > 0), never written. So no implementation, ours or otherwise, can produce $b = 2/$a = 2 (or = 1) from this exact rule text. The expected-output strings look like they were pulled from a different/adjacent line in your fixture when pasting.

Could you paste the exact full line from your test file (or a small diff), so I can verify against the real fixture instead of a possibly-mismatched copy?

[CurlyMoo]

True, it possibly lead to a parsing error before.

[IgorYbema]

Good to know — thanks for confirming. Since the mismatch was on the pasted fixture rather than the fix itself, is there anything else blocking this PR, or is it good to merge?

[CurlyMoo]

Two additions:

1. Can you add these two unittests:

{ "on foo then max(1, 2); end on timer=1 then if $a > 0 then foo(); end max(1, 2); end", { { "", 108 }, { "", 219 } }, { { "", 167  }, { "", 151 } }, 0 },
{ "if 1 == 1 then if 2 == 2 then foo(1); else foo(2); end end on foo($b) then $a = $b; end ", { { "", 156 }, { "[2]$b = NULL[2]$a = NULL", 194 } }, { { "", 167  }, { "[2]$b = NULL[2]$a = NULL", 151 } }, 0 },

2. Can you also set the slots debug for non ESP devices:

    printf("bytecode: %d/%d, heap: %d/%d, stack: %d/%d bytes, varstack: %d/%d bytes, varstack slots: %lu/127\n",
      getval(obj->bc.nrbytes),
      getval(obj->bc.bufsize),
      getval(obj->heap->nrbytes),
      getval(obj->heap->bufsize),
      ((stack == NULL) ? 0 : getval(stack->nrbytes)),
      ((stack == NULL) ? 0 : getval(stack->bufsize)),
      ((varstack->nrbytes == 0) ? 0 : varstack->nrbytes),
      (varstack->bufsize),
      varstack->nrbytes / sizeof(struct vm_vchar_t)

[IgorYbema]

Added both — the two unittests and the varstack slots debug output for non-ESP devices. Pushed in b7257f0.