How much of a risk is 5.x ... ?

[Robert Watson]

On Fri, 11 Jun 2004, Marc G. Fournier wrote:

> So, I think the question more or less comes down to whether or not 5.x
> is to the point where I could be as confident with it as I am with
> 4-STABLE?  Like, I believe that most of the fixes that David and Tor put
> into 4.x for the unionfs stuff were migrated up to 5.x (or vice versa),
> but I also know that there are several things I can't do with it under
> 4.x, nor expect to be able to under 5.x ... but should I expect 5.x's
> unionfs to be in about the same stable as 4.x?  Or in a worse state? 

I can't speak specifically to unionfs, but I can speak generally to 5.x. 
It's not yet "-STABLE" -- many people are using it in production, but
they're using it (hopefully) with the knowledge that we're still actively
working in areas that produce intermittent instability and disruption.
We're also actively working on performance: in some areas, the performance
of 5.x is still substantially more poor that 4.x (in others, it's a lot
faster).  For at least the next month or two, I would expect to users for
5.x (as opposed to 5.2.1) to be aware that they will want to read
freebsd-current carefully to avoid disruption, and expect nits (that we'll
work to sort out as quickly as we can).  Not everyone will have problems,
and if you're careful to pick update dates carefully, you might well find
it runs very well for you.  However, it is -CURRENT still!

Robert N M Watson             FreeBSD Core Team, TrustedBSD Projects
robert at fledge.watson.org      Senior Research Scientist, McAfee Research