[PATCHES] Rationalizing and unifing Schannel

[simo]

On Tue, 2010-02-23 at 20:41 +1100, Andrew Bartlett wrote:
> On Mon, 2010-02-22 at 10:53 -0500, simo wrote:
> > I have been working on a patchset to enhance the schannel interface and
> > also to make it possible to share this code between S3 and S4 again.
> > 
> > The first 6 patches shouldn't be controversial.
> > The last 2 patches remove the use of ldb and goes back to use a tdb for
> > samba4, the comment on patch 7 explains the rationale.
> > 
> > Unless there are objections I will push this patches during the week.
> 
> Except for the comments on ldb performance, which I think don't really
> apply here (unlike sam.ldb, schannel.ldb does not load modules, and does
> not have a very high connect rate anyway), this seems like a very
> reasonable approach.  In particular, it's important to unify subsystems
> like this, and the temporary nature of this DB lends itself to NDR'ed
> structures and a single key lookup.  It's nice to be able to see the
> details of a running server with a simple ldbsearch, but it does not
> seem to be needed often. 
> 
> Anyway, I presume the TDB code is fully transaction safe etc?  

If you look at patch 5 you will see that the tdb code is protected with
the same transaction logic we had for the ldb one.

> My only comment is that I would not want the ldb -> tdb conversion in
> this very particular case to establish a precedent, or to come about
> only because it's too hard to talk about ldb.  We do need a mature
> debate about the use of ldb in Samba as a whole.

It's not about removing ldb or anything, I found out it simply seemed
the right thing to do to keep things simple, given the usage pattern is
very basic.

Simo.

-- 
Simo Sorce
Samba Team GPL Compliance Officer <simo at samba.org>
Principal Software Engineer at Red Hat, Inc. <simo at redhat.com>