To release Samba 4.0 'as is'

Andrew Bartlett abartlet at
Tue Nov 22 14:39:25 MST 2011

On Tue, 2011-11-22 at 09:43 -0800, Richard Sharpe wrote:
> On Tue, Nov 22, 2011 at 9:33 AM, Jeremy Allison <jra at> wrote:
> > On Tue, Nov 22, 2011 at 05:13:16PM +1100, Andrew Bartlett wrote:
> Andrew Bartlett's proposal trimmed because you can infer it from
> Jeremy's response.
> >> What do others think?
> >
> > -1. We can't do this. It isn't an integrated product yet,
> > it's just a grab bag of non-integrated features.
> >
> > Without proper design and code to implement the source3
> > fileserver, winbind and nmbd integration we simply don't
> > have a finished product. The integrated fileserver is a
> > MUST HAVE for a 4.0 release.
> >
> > Rightly or wrongly Samba4 is seen as the replacement
> > for Samba3. If you ship what you have now you put the
> > Team members who support OEM fileservers in an incredibly
> > difficult position w.r.t. marketing and communications
> > with our customers, not to mention the Linux distros.
> >
> > Don't just ship whatever you have out the door, you need
> > to produce a plan to finish the missing features, and
> > create bug reports to track them.
> >
> > If you want to split the codebase and Team efforts completely,
> > releasing an unfinished Samba4 is an excellent way to do it.
> I concur with Jeremy's view. I get lots of people who are involved
> with using Samba as a CIFS server asking the difference between Samba
> 3 and Samba 4, and I think you will find it very hard to position
> Samba 4 if it does not have the sort of functionality Jeremy talks
> about and you risk turning many people off of Samba altogether.

Jeremy and Richard,

I am surprised by your comments, and would like to understand more about
how you feel this proposal would hinder those working on OEM
fileservers.  Indeed, my proposal is to address the specific discussion
about Samba 3 and Samba 4 file servers by allowing you to simply point
to smbd and say 'this is the Samba 4.0 file server'. 

It is also a great opportunity to remove the overbearing significance of
the 4.0 version number, and complete our internal integration in our own
time.  We can stop people asking about Samba 4.0 by simply providing it!

My proposal is that we release the whole of Samba, as currently built in
master - this includes the smbd file server that OEMs currently ship, as
well as all the other parts.  No components are being removed or
deprecated, and nothing need change for these OEMs, nor the team members
they employ.  Indeed, this may give a good opportunity to enable SMB2 in
smbd by default, if the past few months have given us sufficient
confidence in that code.

Just as the WHATSNEW has described for a a number of months now, we can
continue to describe the file server included in the AD component is a
specialised file server built to support the AD DC functionality.   OEMs
that have no interest in the AD server would simply not need to
reference or include the functionality - just as they do not expose the
ability to be a DC at the moment. 

This would be our first combined release, and just as we made Samba 3.6
with our first production SMB2 server, we would explain that this is our
first AD server.  It will also be our first release with the new waf
build system (producing smaller binaries), and hopefully a first release
with TDB2. 

We are actually in a pretty good state with regard to Samba's AD server
at the moment.  Our users generally report success with their AD
installs, and while there remain a number of issues we need to address
(some of which are raised in this thread), I certainly do not regard it
as 'unfinished'.  That said, the work of building Samba will always
continue, and new features are always being added and bugs are being

Why not mark a milestone here, a stable release for our users to build
on, while we continue towards our eventual goal?

Andrew Bartlett

Andrew Bartlett                      
Authentication Developer, Samba Team 

More information about the samba-technical mailing list