Re: Android port

[Bruno Haible]

Eli Zaretskii wrote:
> According to my reading of vasprintf.m4, if ac_cv_func_vasprintf=yes,
> the rest of the tests, including adding asprintf to LIBOBJ, should not
> have happened:
> 
>   AC_DEFUN([gl_FUNC_VASPRINTF],
>   [
>     AC_CHECK_FUNCS([vasprintf])
>     if test $ac_cv_func_vasprintf = no; then  <<<<<<<<<<<<<<<<<<
>       gl_REPLACE_VASPRINTF
>     fi
>   ])

There are invocations of gl_REPLACE_VASPRINTF from other places as well.
This is why we couldn't turn the AC_LIBOBJ invocation into a lib_SOURCES
augmentation in the module description. (It would yield duplicate symbols
at link time, in some circumstances.)

Bruno