Re: AC_DEFUN_ONCE semantics

[Paolo Bonzini]

>> Before the patch, foo was expanded inside the conditional; after the patch, 
>> foo 
>> is hoisted and expanded before the conditional.  Yes, hopefully this is a 
>> small 
>> incompatibility; but hopefully it only results in longer execution time in 
>> the 
>> case where the shell condition used to skip the processing of foo.
> 
> On the other hand, the manual already said that inside the body of an
> AC_DEFUN'd macro (such as AS_IF), you should use AC_REQUIRE([foo]), never
> direct foo.  So if you were already obeying the manual, you should never
> run into this particular change in semantics.

I never knew about this, and I actually disagree with it.  (It is
redundant with the new AC_DEFUN_ONCE).

Anyway, this possible incompatibility simply means we have to be careful
about what is defined as oneshot.  In practice (given my analysis in the
other email) we won't define so many things that way.

Paolo