The TDB2 decision (ntdb?)

[Rusty Russell]

On Sat, 12 May 2012 12:29:04 +0200, Michael Adam <obnox at samba.org> wrote:
> Thanks for the good discussion yesterday and for the summary.

Yes, same here: it was fun!

> I have started to work on the dbwrap issue yesterady.
> I have moved the dbwrap core lib (without the still s3-specific
> parts dbwrap_ctdb and dbwrap_watch) to the top level (not pushed
> to autobuild yet).
> git://git.samba.org/obnox/samba/samba-obnox.git branch master-dbwrap

This looks pretty simple so far.

> The purpose is to be able to also turn users of tdb/tdb_wrap in
> other parts of samba than source3 to use the dbwrap interface.
> Once all parts are converted, switching between tdb and tdb2 will
> be a matter of changing dbwrap backends.
> Does this approach sound reasonable/useful?

There are some tdb-specific assumptions in dbwrap, but the translation
should be fine (eg. db_opsn's hash_size, which is irrelevant for tdb2,
and tdb_flags, which are the tdb1 flags).

I guess it doesn't *have to* be 100%, if we end up using direct tdb2 for
some temporary databases.

Thanks,
Rusty.