Replacing smbclient and nmblookup binaries with the source3 equivalents for the Samba4 release.

[Andrew Bartlett]

On Thu, 2012-03-01 at 14:12 -0800, Jeremy Allison wrote:
> On Fri, Mar 02, 2012 at 08:37:52AM +1100, Andrew Bartlett wrote:
> > > > I'll post the patch for review once I'm done.
> > 
> > I'll just warn that this is a bit harder than it looks at first.  My
> > suggestion is to start with Kai's patch, but the issue is that the
> > Samba4 tests use and expect these binaries in a few places. 
> > 
> > What we need to do first (as a separate patch) is extend the binary
> > mapping in selftest/wscript to include smbclient4 and nmblookup4.  Then
> > these need to be passed down into the test scripts as a parameter.
> > 
> > Then, once the mapping system is in place, the rename is easier, as you
> > only adjust wscript files (including the mapping line in
> > selftest/wscript) and do not change test semantics. 
> 
> Yeah, I've done lots of work on keeping the tests working.

Thanks.

> I didn't know Kai had a patch for this, I'll ping him to
> get a copy.
> 
> > Finally, we should deal with the many other duplicate testing binaries.
> > I would like us to have just one smbtorture, outputting subunit, and
> > avoid having multiple sources of masktest, locktest et al. 
> 
> I don't care about these really, although I think we need
> to keep a separate smbtorture3, as it's diverged significanly
> from smbtorture4 - they really do different and complementary
> things.

I think you misunderstand.  We already call into 'samba3' code from
smbtorture to run the libnet and libsmbclient tests, I just want to
extend that to the code in all the smbtorture3 tests.  

It isn't a high priority, but as we are talking about binary
rationalisation, I want flag a future with just one smbtorture that runs
all of our tests.  This would mean we can share more of the testing
infrastructure that has been built up in smbtorture, with things like
the fine-grained knownfail handling that subunit gives us. 

> > I'm currently working on this.  Once ntlm_auth3 uses gensec on the
> > client and server, the key functionality will be pretty much identical
> > anyway.  In short I expect to essentially merge these. 
> 
> Great ! I'd love to have just one ntlm_auth for Samba4,
> I don't care which one so long as the scripts written
> around the existing one still work.

Hence why I've been writing and extending test scripts :-)

> Thanks for your help on the merge.

No worries.  Thanks for helping us move towards Samba 4.0.

Andrew Bartlett

-- 
Andrew Bartlett                                http://samba.org/~abartlet/
Authentication Developer, Samba Team           http://samba.org