DragonFly-1.0 RELEASED!
Matthew Dillon
dillon at apollo.backplane.com
Mon Jul 12 13:52:33 PDT 2004
:> We have managed to retain 4.x's vaunted stability throughout the development
:> process, despite ripping out and replacing major subsystems...
:
:I think I'm more impressed with this than anything else -- I still can't
:believe how few problems I've had over the last year -- and the few problems
:I've reported have been fixed with amazing speed.
I'm amazed myself, actually. What I really love about the codebase is
that most of the bugs have been incredibly easy to track down. The
FreeBSD guys thought it would be more difficult to debug a threaded
system then a mutexed system, but in fact I believe we are proving that
it is the other way around! It is far easier to debug our threaded
system. The FreeBSD guys depend way too much on WITNESS to find
problems for them, rather then focusing on proper coding abstractions.
Our focus on making the coding methodologies straightforward and obvious,
and inherently MP safe without requiring massively complex coding
rules, is resulting in much better-designed, less bugful code going into
the tree.
Plus, it's actually *fun* to code for DragonFly. Coding MP clean
mechanisms in FreeBSD-5 is decidedly *not* fun.
-Matt
More information about the Kernel
mailing list