Re: EXTERNAL_LIST_LOOP_2 interface and semantics changed - propose reversion

Ben> martin, what are you smoking? as i read through your diatribe Ben> here, the idea that _NO_DECLARE should indeed not declare any of Ben> its variables is the only thing i agreed with. but i could have Ben> sworn that i did indeed do this -- and i did! Ben> there is no EXTERNAL_LIST_LOOP_2_NO_DECLARE. Hmmm. You're right. Since there's a EXTERNAL_LIST_LOOP_4_NO_DECLARE, shouldn't there be an analogous EXTERNAL_LIST_LOOP_2_NO_DECLARE for consistency? (Of course, assuming we keep the API change to the EXTERNAL_LIST_LOOP macros, which I think is a bad idea). Ben> the docs for EXTERNAL_LIST_LOOP_4_NO_DECLARE read Ben> EXTERNAL_LIST_LOOP_4_NO_DECLARE is like EXTERNAL_LIST_LOOP_4 but none Ben> of the variables are automatically declared, and so you need to declare Ben> them yourself. (ELT and TAIL are Lisp_Objects, and LEN is an EMACS_INT.) Ben> go look at the source. See how the first thing the EXTERNAL_LIST_LOOP_4_NO_DECLARE macro below does is *declare* a variable? Do you see how this could confuse people? To get *no* declarations (and therefore no need for extra braces), you use PRIVATE_EXTERNAL_LIST_LOOP_6. #define EXTERNAL_LIST_LOOP_4_NO_DECLARE(elt, list, tail, len) \ Lisp_Object tortoise_##elt; \ PRIVATE_EXTERNAL_LIST_LOOP_6 (elt, list, len, tail, \ tortoise_##elt, CIRCULAR_LIST_SUSPICION_LENGTH) By the way, I don't understand why PRIVATE_EXTERNAL_LIST_LOOP_6 has been renamed to be PRIVATE_ . I intended this function to be part of the public API. There certainly are cases where a user would want to use this full version, for example to use a better value for suspicion_length. I still think we should revert to the prior *_EXTERNAL_LIST_LOOP_* API. Martin