Re: m4_defun_init

[Paolo Bonzini]

Eric Blake wrote:
> Paolo Bonzini <bonzini <at> gnu.org> writes:
> 
>> I'm not sure tracking the location is not useful.  As a first step, you
>> can use m4_define if the common version is empty and m4_defun otherwise.
> 
> I don't see why the common would ever usefully be empty (just use 
> m4_defun_once 
> or m4_expand_once instead, if you have initialization but no body).

m4_defun_once warns if the macro is called more than once.

> Ah.  m4_fatal uses m4_expansion_stack_dump, which does indeed have less 
> information if location is not being tracked.  Perhaps we could change the 
> common definition from its current:
> 
> $2
> 
> to the more complex:
> 
> m4_expansion_stack_push(_m4_defn(
>  [m4_location($1)])[: $1 is expanded from...])$2[]m4_expansion_stack_pop()
> 
> while still trimming out the diversion manipulation in _m4_defun_{epi,pro}, 
> so 
> that we are still lighter weight than a full blown m4_defun on the common 
> code.

Yes, pretty much.

Paolo