[PATCH] s4: initial code for the topology algorithm

[Crístian Viana]

Hi Tridge and metze,

> The only TODO I see is this:
>               /* TODO: don't need to ldb_search again; search in res. */

Are you sure you have applied my last patches? That TODO item is in an older
version of the code (and it doesn't make sense now, ldb_search should really
be called again). That item doesn't exist anymore, and grep tells me that
there are 24 TODOs now. You can download the patches attached to my last
e-mails or pull them from my online repository (git://
repo.or.cz/Samba/cd1.git).

> Do you think the rest of the algorithm is clear enough for you now?

Not completely yet.

> Do we generate the connection objects for each dc in our site now? I think
each dc needs to generate only it's own intra-site connections.

Yes, the nTDSConnection objects are being created for each DC, but there's
no topology algorithm in it (everyone is connected to everyone). And only a
few attributes of the connection object are filled.

On Fri, Mar 12, 2010 at 1:20 PM, Stefan (metze) Metzmacher
<metze at samba.org>wrote:

> Hi Crístian,
>
> Do we generate the connection objects for each dc in our site now?
> At least that's my impression after reading the master which are in
> master now.
>
> I think each dc needs to generate only it's own intra-site connections.
>
> metze
>
> >  > When you sent this e-mail I was nearly at the end of that algorithm,
> so I
> >  > decided to finish it. I took a couple of days reviewing the code and
> testing
> >  > it, and only today I could send the patches.
> >
> > That's fine, I think what you have done looks good, and it doesn't
> > seem to cause any problems, so I will push it soon (I'm just waiting
> > on a full test to finish).
> >
> >  > It isn't fully functional as there are some parts I don't know how
> >  > to implement. Those parts have TODO comments explaining what should
> >  > be done that I didn't understand.
> >
> > The only TODO I see is this:
> >
> >               /* TODO: don't need to ldb_search again; search in res. */
> >               ret = ldb_search(ldb, tmp_ctx, &res_site_link, base_dn,
> >                                LDB_SCOPE_SUBTREE, attrs,
> >                                "objectClass=siteLink");
> >
> > which seems to be just an efficiency thing. Do you think the rest of
> > the algorithm is clear enough for you now?
> >
> > Cheers, Tridge
>
>
>


-- 
Crístian Deives dos Santos Viana [aka CD1]