smbd/source3 or ntvfs/source4 for new VFS module development?

Andrew Bartlett abartlet at
Fri Jan 29 23:42:16 UTC 2016

On Fri, 2016-01-29 at 06:48 +0100, Volker Lendecke wrote:
> On Thu, Jan 28, 2016 at 09:58:42PM +1300, Andrew Bartlett wrote:
> > I know you would love to get out the pitchforks, but I would be
> > much
> > happier if we kept with the current state where it is disabled in
> > default production builds, but is still kept building for
> > developers
> > and working with the testsuite.  
> > 
> > For one thing it is still the only part of Samba with the CIFS
> > proxy
> > just discussed. 
> This is *one* user in years. Sam Liddicot did not contribute
> anything back, so the CIFS proxy is orphaned.

BTW, part of the back-story there was our previous policy against
company-owned contributions.  Thankfully we sorted that one out

> If you look at the knownfails for samba4 in the file server
> area it becomes clear how broken the ntvfs file server is.
> By keeping it around we send the wrong message.
> We used to have it as an example of a better file server
> architecture. With SMB2 we have that now in source3 too. All
> parsing and protocol handling is pretty much asynchronous.
> And who wants a perfect architecture for SMB1 these days?
> So, Andrew, if you want to keep it, I urge you to start
> fixing the knownfails. This would give us confidence that we
> are not sitting on Roaming code :-)

I understand your point.  And as you correctly called me on regarding
Heimdal, that my feeling is one more of nostalgia than current
engineering reality, I'm willing to accept this may be the same here. 

Like Heimdal, and unlike the much-unloved internal winbind, it is
entangled in ways within our codebase (spoolss tests at least) that
will require some work.  

I do like that it helps ensure we have an unchanging target against
which to additionally measure our tests, but that may not be enough


Andrew Bartlett

Andrew Bartlett             
Authentication Developer, Samba Team
Samba Developer, Catalyst IT

More information about the samba-technical mailing list