Re: the dangers of DB_LOG_AUTOREMOVE

On Jul 16, 2004, at 9:14 AM, kfogel@collab.net wrote:

> Greg Hudson <ghudson@MIT.EDU> writes:
>> In the long run, I think the answer is to make FSFS the default in 1.2
>> if it seems to work well, and to kill the BDB back end in 2.0 if we've
>> managed to write an SQL back end by that time.
>
> +(1 / sqrt(2)) on making FSFS the default in 1.2.
>
> Don't have any thoughts right now on lifespan of BDB. A lot of
> scalability testing has been done on BDB that hasn't been done on
> FSFS.
>

I'm really in favor of an SQL back end. In particular, I would like to
investigate the viability of running Subversion on a replicated DB
setup. FrontBase is my commercial RDBMS of choice (almost complete
SQL92 compliance, blazing speed, and awesome tech support).

I may be able to help with such an undertaking, depending on how sales
do of my next product (scheduled to ship in less than two weeks). I'm
all for killing off the BDB back end.

If we make an SQL back end, SQLite would also be an option for people
who didn't want to run a DB server.

Regards,

Eric Ocean

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sat Jul 17 02:42:43 2004