Reiser4

Per Bjornsson perbj at stanford.edu
Sat Aug 28 18:59:11 UTC 2004


On Fri, 2004-08-27 at 19:36, Alexandre Strube wrote:

> What about reiser3? It's been part of red hat's distros for years. Why
> don't let it as an option on default installations? (eg wihtout typing
> linux reiserfs)

Hmm, let me guess:
1) Red Hat has no particular interest in ReiserFS since they use ext3 as
their supported file system for RHEL.
2) Nobody else has stepped up to make sure that ReiserFS actually works
well in Fedora.

Thus the option is there as-is, but it's not available by default since
it's less tested, not well-known by Red Hat people, and thus not a good
choice for inexperienced users of Fedora. (It may be a good choice for
inexperienced users of SUSE since they have people working on it and
making sure that it's doing OK.)

Just wondering here: Why do you care about ReiserFS anyways? Reiser4 has
some very interesting features, with its treatment of metadata which
esssentially turns files into directories, but it sure takes some real
consideration of the consequences before moving to something like that.
ReiserFS 3 is, as far as I can see, just another filesystem, and one
that its originators have pretty much abandoned for Reiser4 (although
e.g. SUSE people are effectively maintaining it in the kernel). It may
be fast under certain circumstances, but not always: recently ReiserFS
has shown up as one of the major causes of latency problems in the
kernel (which have been dug up thanks to the investigative and fixing
work of Ingo Molnar and others). Latency problems have bothered e.g. the
high-end audio community - they have very stringent latency requirements
as audio processing is in many cases essentially a soft real-time task,
but in general latency is also a problem for e.g. desktop responsiveness
and interactivity. Given that ReiserFS 3 is by no means a lear-cut
slam-dunk it seems completely pointless to care about raising the
profile of it at this point. 

/Per

-- 
Per Bjornsson
Ph.D. Candidate, Dept. of Applied Physics, Stanford University





More information about the test mailing list