Re: error compiling shadow 4.1.5.1

[Isaac Dunham <idunham@lavabit.com>]

On Tue, 21 May 2013 14:36:26 +0200
John Spencer <maillist-musl@barfooze.de> wrote:

> On 05/21/2013 06:49 AM, Isaac Dunham wrote:
> > Adding<sys/socket.h>  above<netdb.h>  helps get it building.  
> [...]
> > undefining this makes it build.  

In other words, use this command line:
ac_cv_member_struct_utmp_ut_addr_v6=no \
ac_cv_member_struct_utmpx_ut_addr_v6=no \
/configure --without-nscd --without-libpam

But "it" is "one particular file", not the whole package.
login still FTBFS.

ut_addr_v6 being detected breaks things for a completely unrelated
reason, though:
sizeof(sa->sin_addr) and sizeof(sa->sin6_addr)

are failing.
But what's a good replacement? Or shall we make it transparent?

> do you have a patch handy ?

See this part:
> > Finally, there's an unconditional use of ruserok; of course they
> > check whether it returns 0 or 1 on failure, but assume it's present!

> maybe we also should report the bugs upstream.

The status at present:
-missing <sys/socket.h> in libmisc/utmp.c
-need to make ruserok conditional.
-(musl or upstream bug?) shadow assumes that members sin(6)_addr of
struct sockaddr_in (type struct in(6)_addr) are completely defined.

Do we want to implement ruserok?

Here's a (public domain) ruserok stub conforming to the manual
description:

int ruserok(const char *rhost, int s, const char *ruser, const
char *luser)
{
	return -1;
}

Thanks,
Isaac Dunham