[WIP][PATCH] 'fix' smbspool_krb5_wrapper, libexecdir and ctdb helper binary locations for 4.4

[Andreas Schneider]

On Sunday, March 20, 2016 08:04:56 AM Andrew Bartlett wrote:
> On Sat, 2016-03-19 at 12:41 +0100, Andreas Schneider wrote:
> > On Saturday 19 March 2016 14:42:32 Andrew Bartlett wrote:
> > > On Wed, 2016-03-16 at 20:33 +0000, Miguel Medalha wrote:
> > > > > When building Samba 4.4.0rc and then doing an install to
> > > > > /usr/local/samba, a directory "libexec" is being created which
> > > > > then
> > > > > contains a "samba" directory with a single
> > > > > "smbspool_krb5_wrapper"
> > > > > file inside (/usr/local/samba/libexec/samba/
> > > > > smbspool_krb5_wrapper). This is not very clean, is it?
> > > > > 
> > > > > Since all the files are already inside /usr/local/samba, why
> > > > > repeat
> > > > > "samba" again inside the tree? Is it possible to clean this
> > > > > before
> > > > > the final release of Samba 4.4.0?
> > > > > 
> > > > > This doesn't happen with version 4.3.6. The same file is placed
> > > > > in
> > > > > "/usr/local/samba/bin/smbspool_krb5_wrapper".
> > > > 
> > > > Some additional information: if I configure the build with
> > > > "--libexecdir=/usr/local/samba/bin" the file goes into
> > > > "usr/local/samba/bin/samba/smbspool_krb5_wrapper" which also
> > > > contains
> > > > an
> > > > unnecessary repetition. That additional "samba" directory seems
> > > > to be
> > > > hard coded somewhere.
> > > 
> > > Try this patch.
> > > 
> > > Andreas:  Can you comment on this and let me know if this is OK for
> > > what you were looking for?
> > 
> > It looks fine for me. Martin should look at the ctdb changes. If the
> > binaries
> > move, some scripts need to be updated too iirc.
> > 
> > So Martin needs to look into this!
> 
> Agreed.  We can always create a different dynconfig variable that
> always expands to libexecdir/ctdb if that is preferred.
> 
> > > For some reason smbspool_krb5_wrapper doesn't build on my system,
> > > so I
> > > cna't fully test it, so consider this a WIP.
> > 
> > What error do you get?
> 
> It just didn't choose to build, probably some missing headers (and so
> missing HAVE_CUPS).  I didn't want to dive into that on the weekend,
> but wanted to get the patch back the the OP given I had written it.

Strange, it uses enabled=bld.CONFIG_SET('HAVE_CUPS') so it is only built if 
cups has been found ...