Re: [Bug-gnulib] Re: getopt trouble on uClibc systems

[Paul Eggert]

Simon Josefsson <address@hidden> writes:

> How about making gl_GETOPT test for _getopt_long_only_r too, instead?

That sounds a bit extreme.  glibc doesn't export _getopt_long_only_r
(it uses linker magic to prevent it), so it would cause (say) "cp" to
use its own copy of getopt, even though it doesn't need
_getopt_long_only_r.

> doesn't it seem like _getopt_long_only_r should be renamed to
> getopt_long_only_r, if applications are supposed to be using it at
> all?

Only argp uses _getopt_long_only_r, and argp is part of glibc, so it's
not a good example for why getopt_long_only_r should be public.

With this in mind I'm becoming inclined to put in a special hack for
argp here....