Comment added that we do full scan of the parent entity when all its childs
authorshort <>
Sun, 1 Jul 2001 20:11:42 +0000 (20:11 +0000)
committershort <>
Sun, 1 Jul 2001 20:11:42 +0000 (20:11 +0000)
were 'consensing' to the S_and/S_or operator.

cfgeval.c

index fcd3a97..5435709 100644 (file)
--- a/cfgeval.c
+++ b/cfgeval.c
@@ -1326,6 +1326,13 @@ top_down:
                    report(LOG_DEBUG, "expr_eval_immediate: bottom_up and_or: full scan: " PF_EXPR,
                            PA_EXPR(expr_single));
 
+               /* It would be nice to pretend that all 'veto' decisions already made in the child
+                * had to set our 'result' to the correct value. But 'consent' decisions don't set
+                * us and the behaviour of auto-set from the child in 'veto' case may get changed
+                * in the future versions.
+                * So we rather don't depend on it.
+                * This overhead doesn't change altgorithmic complexity anyway.
+                */
                result_parent = consent;
                for (expr_child = expr_parent->u.and_or.child_first; expr_child; expr_child = expr_child->next)
                {