> At any rate, I'm interested in documenting these macros, as well as adding > a counterpart macro for m4_set iteration (the recent > _AS_DETECT_BETTER_SHELL change can be sped up). It turns out that m4 is > more efficient when you pass the list element as an argument to a macro > invocation than it is with m4_foreach semantics of doing m4_define of the > list element to a temporary name, then doing m4_defn of that temporary > name to get back to the list element. I originally went with the name > m4_transform because m4_map was taken, but maybe m4_map_each and > m4_map_pair might make more sense?
... or m4_map_args and m4_map_args_pair? Paolo
