[NTLUG:Discuss] Re: Strange Request, Maybe?

[cbbrowne@hex.net]

On Thu, 10 May 2001 16:53:20 CDT, the world broke into rejoicing as
Steve Baker <sjbaker1 at airmail.net>  said:
> cbbrowne at hex.net wrote:
>> They're paying Stephen Tweedy to do filesystem work, and "his
>> filesystem" is ext3, which is essentially ext2 plus a journal file
>> alongside.  They have been pretty careful to not say anything too
>> negative about ReiserFS, but it sure looks like it suffers from
>> "Not Invented Here" syndrome.

> You can kinda understand that - you spend all that time and effort
> to write a fancy filing system only to find everyone rushing off to
> adopt a different one...not a good feeling.

More importantly, you spend money paying for someone to work on such,
only to find ... all that other stuff...

>> XFS may be pretty neat, but I wouldn't expect people to start using
>> it en masse until, _at the earliest_, when kernel 2.6 is released.
>> And 2.5 work hasn't started yet, so that is RATHER afar off.

> But at least SGI have been running XFS under IRIX for a couple of
> years now.  I'd expect it to have the worst of the kinks worked out
> after all that practical service.

That assumes that significant revisions aren't needed to connect it to
the Linux VFS.  Connecting up a filesystem _isn't_ as simple as
saying, "Oh, we'll just have to tie in some inodes."

The fighting that took place surrounding trying to get ReiserFS into
the "official" kernel stream was enough to demonstrate that throwing
big subsystems into Linux is _not_ a trivial matter.  It didn't happen
easily with ReiserFS; it didn't happen easily with GGI; it will not
happen easily with XFS or JFS or Tux2.

The only other journaling FS that _could_ get in quickly is ext3, and
that's because it's a natural extension of ext2.

I'd be quite prepared to bet money that XFS isn't in stable kernels
before 2.6, _if then_.  It would be a pretty safe bet.  [Actually,
since the odds are pretty poor, the payoff would be poorer, so it's a
bad bet anyways...]
--
(reverse (concatenate 'string "gro.mca@" "enworbbc"))
http://vip.hyperusa.com/~cbbrowne/resume.html
"Now, if someone proposed using people who spam comp.sys.* groups with
political  screeds  in  place  of  lab  rats  for  drug  testing,  I'd
wholeheartedly concur".  -- John C. Randolph