"Hou, Zhijie" <houzj(dot)fnst(at)cn(dot)fujitsu(dot)com> writes:
> I notice that there are no Nullif case in eval_const_expression.
> Since Nullif is similar to Opexpr and is easy to implement,
> I try add this case in eval_const_expressions_mutator.
I think this patch should be about a tenth the size. Try modeling
it on the T_SubscriptingRef-etc case, ie, use ece_generic_processing
and then ece_evaluate_expr to cover the generic cases. OpExpr is
common enough to deserve specially optimized code, but NullIf isn't,
so shorter is better.
regards, tom lane