"Craig A. Vanderborgh" <email@example.com> writes:
> > Cue the usual "it's not BDB's fault, it's the way svn uses it".
Heh. Just to be clear:
The reason we keep saying that is so that poor Sleepycat Software
(maker of BDB) doesn't get blamed for a problem in Subversion's code.
BDB as delivered by Sleepycat is solid, one just has to use it in the
right way. We didn't, we got burned. However, Sleepycat's reputation
shouldn't suffer... which is why we make those posts.
> I would have to second this 100%. I lurked on this list for almost a
> year before installing, aghast at the problems that so many people
> have had with BDB. When I finally installed our subversion system, I
> went for FSFS without even seriously considering BDB. We haven't had
> a problem (on Fedora Core 1).
> The Subversion developers need to remember this - BDB-based Subversion
> does not work well enough to provide a robust version control system.
> With an SCM system, data loss is simply unacceptable. It's not an
> "inconvenience", it's utterly unacceptable.
> Therefore, it does not matter that FSFS is "slower" or less "high
> tech" than BDB. Since it provides the required reliability, it's
> actually the only viable option for Subversion users. The BDB
> implementation should either be repaired, or it should be abandoned if
> it can't be. As it stands right now, it's unusable. And there are a
> lot of devoted users out there who can attest to that.
Yup. We are well aware, and we are working on it, with help from
Sleepycat (it's a tricky architectural problem). In the meantime,
making FSFS the default repository storage back end in 1.2 will
hopefully make the problem go away for 99% of users. Those who choose
not to use FSFS will be doing so consciously now.
By the way, I know of no evidence that FSFS is slower. Not sure what
less "high tech" means -- but trust me, it's never been the basis for
any technical decisions in Subversion :-).
To unsubscribe, e-mail: firstname.lastname@example.org
For additional commands, e-mail: email@example.com
Received on Fri Jul 1 21:51:53 2005