Current cvs bits and shadow.h ?

[Cliff Green]

--On Thursday, December 03, 1998, 1:44 PM +1100 Ken Stone <ken at sdd.hp.com>
wrote:

> 
> Current cvs bits bomb on HPUX 10.20 due to configure not detecting the
> presence of shadow.h (but detecting the lib call that needs it :-)
Yeah - we noticed this too.  It's in both CVS branches, but not in the
2.0beta tarballs, so it's been introduced pretty recently.

After much rummaging about in the three sets of
configure/configure.in/config.h/Makefile/config.cache files (from the beta2
tar, the SAMBA_2_0 branch, and the HEAD branch), my vote goes to the tests
in configure for the checks for putprpwnam, set_auth_parameters, getspnam,
bigcrypt, and getprpwnam.  Checking the cvs log for aclocal.m4, I read:


> 
> 1.3.2.1 Wed Nov 25 20:14:58 :Australia/NSW 1998 by jra 
> Branch: SAMBA_2_0
> Diffs to 1.3 
> 
> Makefile.in: Added 'maintainer mode' fixes.
> aclocal.m4: Added new function - AC_LIBTESTFUNC.
> configure.in: Fixes for the messy -lsec, -lsecurity code.
> include/includes.h: Added a default printcap define.
> Jeremy.

Now I'm no initiate into the ways of autoconfig, but when reading the
definition of AC_LIBTESTFUNC, I couldn't find a corresponding definition of
AC_CHECK_FUNC, which is only called there.  Could this be part of the
problem?

> Danged if I can figure out what configure is doing to miss it.  Its right
> in /usr/include/shadow.h and the test prog that configure uses seems to
> work right ?
A diff of config.cache for beta2 and the current cvs branches shows that
there's a few defines that have been reversed from the earlier (makeable)
set :

diff of config.cache for SAMBA_2_0 and beta2.tar.gz, in that order:

4c4
< ac_cv_func_set_auth_parameters=${ac_cv_func_set_auth_parameters=yes}
---
> ac_cv_func_set_auth_parameters=${ac_cv_func_set_auth_parameters=no}
6c6
< ac_cv_func_getspnam=${ac_cv_func_getspnam=yes}
---
> ac_cv_func_getspnam=${ac_cv_func_getspnam=no}
8c8
< ac_cv_func_bigcrypt=${ac_cv_func_bigcrypt=yes}
---
> ac_cv_func_bigcrypt=${ac_cv_func_bigcrypt=no}
10c10
< ac_cv_func_getprpwnam=${ac_cv_func_getprpwnam=yes}
---
> ac_cv_func_getprpwnam=${ac_cv_func_getprpwnam=no}


> Any thoughts ?
I wish I knew more about setting up autoconfig.

c
--
Cliff Green
green at umdnj.edu
Academic Computing Service, UMDNJ