Grr! <rant>

Adam McDougall mcdouga9@egr.msu.edu
Thu, 22 Feb 2001 11:43:37 -0500 (EST)


On Thu, 22 Feb 2001, Paul Melson wrote:

> > On Mon, 19 Feb 2001, Edward Glowacki wrote:
> >      This might be a point of contention.  Have you checked out
> > softupdates? From what I can tell on the *BSD lists I'm on, softupdates
> > are fairly stable on FreeBSD.  I know they've been around for awhile and
> > I've heard of some people running them on production machines.  If you
> > haven't heard of softupdates its basically another way to accomplish what
> > journaling does.  I guess in some ways its better and the author has some
> > pointers to some in depth info about softupdates on his web page
> > (mckusick.com), but I think you need to be a usenix member to read them.
>
>
>      What about JFFS?  Any plans to incorporate that into *BSD?
>      It'd be pretty neat to see an open (source) standard fstype
>      that does journaling.  I just recently added a JFFS partition
>      to my Linux system under 2.4.1.  I haven't tested its ability
>      to recover from hard downs (since I still have ext2 partitions
>      that I value), but it performs nicely with large files and
>      heavy r/w usage.

For FreeBSD, the only plans I have heard about are souping up ffs (ufs) to
be fsck-less and be more safe during sudden power loss, as well as
supporting snapshots and maybe some other neat stuff.  I'm pretty syre
NetBSD has a semi-functional LFS (Logging File System) but I haven't
looked to see if its production quality yet.  I think we probably wont see
reiserfs or XFS or stuff like that ported unless one or more people show
up who are incredibly skilled at de-porting them from linux, porting
them to bsd, have copius free time, have the will, and aren't convinced
that it would be easier to take some FS existing in BSD and souping it
up.  I believe someone named Terry Lambert has ported JFS to FreeBSD
and/or FreeBSD to something that uses JFS (cant remember which) but he is
disenchanted by people's reluctance to accept code officially into
products just based on their technical accomplishments.  Unfortunately
some code gets rejected due to bad coding style and sometimes bad author
attitude instead of cleaned up :/