Date: Wed, 26 Mar 2003 02:32:56 -0800 From: Terry Lambert <tlambert2@mindspring.com> To: Donn Miller <dmmiller@cvzoom.net> Cc: David Schultz <dschultz@uclink.Berkeley.EDU>, fs@freebsd.org Subject: Re: Why is there no JFS? Message-ID: <3E8181D8.46C77E08@mindspring.com> References: <b2ejfe$1sl7$1@FreeBSD.csie.NCTU.edu.tw> <3E8169A3.7090309@cvzoom.net>
next in thread | previous in thread | raw e-mail | index | archive | help
Donn Miller wrote:
> I personally think LFS would be the best choice as an alternative to
> UFS, but I haven't really looked into the technical aspects of LFS yet.
The unified VM and buffer cache changes which were made to FFS,
but not to LFS, by John Dyson, and the lack of a working "cleaner",
were the primary reasons LFS "died" to the Attic.
NetBSD has since written a working "cleaner", and the VM system
interface changes aren't really *that* hard to understand. LFS
would be moderately easy to revive to a working condition.
You would get a lot of crap from certain corners of the peanut
gallery about "why bother?", though. Some of this is justifiable
as a criticism, because the real reason there's interest in a new
FS design is the ability to boot fast after a crash, and LFS does
not give you a heck of a lot of this, because the data it protects
is only the metadata. You would need to add an NTFS-style "flip-flop"
log on top of that for data logs, and then delay wakeups until they
were known to be committed to one of the flip-flops.
Also, though LFS is loosely based on the FFS code (which is why a
forward port would not be difficult), it doesn't contain soft
updates technology, which would annoy some people.
> And what about ReiserFS?
It's s trivial port, but it's GPL, unless you pay for it. It also
has (IMO) some patent encumberances for Novell patents. I'm not
sure they are *exclusively* licensed to USL/SCO, but they *are*
licensed to them, so it may be an SCO vs. IBM thing pretty quickly
(the specific patents are US Patent 5666532 and 5642501). It may
also infringe 5218695, but Epoch Systems is less litigous.
PS: The only way you "prove" infringement is by being sued over
infringement, and losing; hence the (IMO).
-- Terry
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-fs" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3E8181D8.46C77E08>