[PATCH] Proposed merge of some NTLMSSP crypto

[Andrew Bartlett]

On Fri, 2009-12-11 at 11:27 +0100, Kai Blin wrote:
> On Friday 11 December 2009 10:25:14 Stefan (metze) Metzmacher wrote:
> Hi Metze,
> 
> > I think this patches need some more work to reduce the risk.
> 
> I think this is a good general guideline for the merging process.

I'm not so sure.

> > 1. Both trees need patches which bring the code step by step
> >    in to the same code.
> [...]
> > 2. All this steps need to be small easy to review patches,
> >    which build and pass make test.
> >
> > 3. Then we need to move the file(s) from one tree to the topdir
> >    and make sure everything still works fine.
> [...]
> > 4. Let the other tree use the different file(s)
> >
> > 5. Remove the unused files(s)
> 
> Do we want this in some document anywhere?

I understand the motivation, and I'm not normally involved in merge work
- so I've not really got work pattern for it going - but it seems over
the top.

I like quality mechanisms, and perhaps to require anything less is to
invite nothing less than total disaster, but it also seems to be a very
large amount of 'make work'.  

I do realise that for a lot of this code the risk of doing nothing is
almost zero, and so the risks created by merge work is therefore almost
infinite in comparison.  But if we make it too difficult to do this
work, then we will continue to see almost no work done in this area, and
fail in our agreed goal to merge the two Samba codebases, particularly
in the libraries. 

For the patches I was doing, in the area of my expertise, I was planning
to:
 - Prepare a common shared lib
 - Read git log/blame looking for particular points to pay attention to
 - Adjust Samba3 and Samba4 to use the shared lib
 - Test in both trees
 - Fix failures from 'make test' in both trees
 - Amend code till 'make test' succeeds (this is where I'm up to)
 - Perform manual tests with Windows clients/servers on both branches.
 - Propose patch series for possible merge
 - Rebase patches (with merge of patch parts to avoid disjoint API
changes)
 - Ask for someone to sign off the patch queue.

Anyway, I'll try and fix up the NTLMSSP work next week, so we can have a
more concrete debate about what to do with this particular patch
series. 

Andrew Bartlett

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

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
URL: <http://lists.samba.org/pipermail/samba-technical/attachments/20091211/e538c070/attachment.pgp>