[s4] LDB TDB indexes

[Andrew Bartlett]

On Mon, 2009-11-16 at 19:20 +0100, Matthias Dieter Wallnöfer wrote:
> Hi Andrews,
> 
> since I noticed a very strange problem after executing twice the ldap.py 
> test (a rename test does break always after the first round) I did some 
> investigations on LDB indexing. And I think I've found quite a few 
> issues and also bugs there. To start a discussion I post a correction 
> patch which for sure is only a preview for the final solution but from 
> which the commit message is interesting (it explains the problems in 
> detail).

These do look like very similar changes to those I proposed in my
ldb-index branch, which I did for similar reasons.  One concern I have
is that ldb_dn_get_casefold_dn() can be quite expensive, but otherwise
it looks like a useful improvement.

However, it touches the rewritten index code, over which I am not yet
familiar, and I hope that tridge may be able to comment more.

My main concern is that, unlike my ldb-index branch, this does not make
any attempt to upgrade the on-disk format.  Without that change, won't
we get the same kind of errors, reading 'linearized (mixed case)' DNs
from disk, and then doing strcmp() on them with casefolded DNs?

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/20091117/fc8189d9/attachment.pgp>