DragonFly kernel List (threaded) for 2003-07
[
Date Prev][
Date Next]
[
Thread Prev][
Thread Next]
[
Date Index][
Thread Index]
Re: why 4.X instead of 5.X
On Thu, Jul 17, 2003 at 11:09:28PM -0700, Matthew Dillon wrote:
>
> :If nothing else, it has the older gcc so buildworlds will go
> :faster... :-) </joke>
> :
> :I like the idea of starting out with the very-stable branch
> :(both abi and api), even though there are some things about
> :5.x that I would miss. 5.x is still very much a rapidly
> :moving target. He couldn't go with 5.0 for sure (too many
> :problems), and even 5.1 was shaky. By going with the very
> :stable branch, it will be much easier to re-sync userland
> :when it's time to. It should also be easier to pull in any
> :critical security fixes, should something pop up.
> :
> :--
> :Garance Alistair Drosehn = gad@xxxxxxxxxxxxxxxxxxxx
> :Senior Systems Programmer or gad@xxxxxxxxxxx
>
> That is my feeling. Most of the initial work I did prior to the
> announcement was as much to prove to myself that I could start with
> a 4.x base. There were a couple of days where I almost didn't think
> I could do it, changing procs to threads in VFS and fixing ucred was
> a nightmare! But I buckled up and slogged through it. Getting that
> all done and the light weight kernel threading system actually working
> proved it out for me and I am now *very* comfortable with using 4.x
> as a base. I believe I have made the right choice. Even if 5.x were
> more stable the mutex model it uses is so complex that ripping it out
> (and stabilizing what was left) would have taken far longer.
Another thing, I am sure people like me will be ready to
backport some of the FreeBSD 5.x features into this OS, such as
Zero-Copy Sockets, and various other improvements that will also
apply to DragonFly; with time of course...
Cheers.
-- Hiten (hmp@xxxxxxxxxxx)
[
Date Prev][
Date Next]
[
Thread Prev][
Thread Next]
[
Date Index][
Thread Index]