Re: Info: prepare for release 4.0.13

On Tuesday 27 September 2005 08:53 am, Tomasz Kłoczko wrote:
> On Mon, 26 Sep 2005, Mike Frysinger wrote:
 > how about the proposed patch ?  this just fixes the selinux case but can
> > be applied to the other withvals as well ...
>
> - shadow ac suit handles more than one --with{,out}-<feature> and the
>    same bugs are in in other AC_ARG_WITH(),

i pointed out that this patch was a PoC against selinux and that the other 
ones had similar bugs ... no point in fixing all of them if they were all 
going to be rejected ;)

> - put with_<feature>=no after AC_MSG_ERROR() does not heve sense
>    because AC_MSG_ERROR() stops configure script with error exit code.

the with_<feature>=no happens in the other branch of the if statement, not 
after AC_MSG_ERROR ...

> I'm just commit fix quoting all comments inside AC_HELP_STRING() as in
> your patch and I'm just finish prepare patch for reported bugs (in
> attachment) which seems is correct (can you try test this ?):

the issue i addressed in my patch and which has been cut out from the new one 
is the autodetection.  the configure script claims that selinux support will 
only be enabled if it's found, but the code actually tries to enable it all 
the time thus forcing the user to run `./configure --without-selinux`.  run 
`./configure` on a machine w/out selinux support and it will error.  the 
patch i posted behaves correctly ...

if the user runs `./configure`, with_selinux is set to "maybe" which means 
configure will check to see if selinux supports, but *not* error if it isnt 
found.  if the user runs `./configure`, then with_selinux is set to "yes" 
which will cause configure to abort if selinux is not found.  if the user 
runs `./configure --without-selinux`, then selinux is never checked for.
-mike