Larry McVoy wrote: > Ditto. When Stephen described the design of ext3 to me about 3 years > ago, I loved it exactly because it tweaked the on disk format almost > not at all. Absolutely stunning in its brilliance. Sees really obvious > but most people rewrite everything when they just need to build on > top of a proven entity. I originally bashed the concept of Ext3, then realized it was the solution most production systems _exactly_needed_. ReiserFS is very revolutionary and innovative -- i.e. not something I want on a production fileserver right now. ;-P > I haven't tried XFS on Linux but I have on IRIX (in fact, I worked on it > on IRIX) and it sucks for many of the things ext2/3 don't. Like file > deletes. Yes, file deletes are mega-slow on XFS. Someone actually came up with an approach to fix it too, although they are holding off on putting it into their releases (they want more "run time" with it). But XFS excels at other things, especially large volumes and file sizes. Especially on 64-bit platforms. Out of JFS, ReiserFS and XFS, XFS has the best Linux feature support. No NFS incompatibilities, POSIX ACLs (access control lists), official quota support, etc..., in addition to "newer" features like inodes-on-the-fly, etc... But since Ext2/3 has POSIX ACLs support now, I haven't been installing XFS much. I heard a [heavily SGI-influenced] extended attributes (EA) interface and POSIX ACLs patch went into 2.5.3, so XFS is going into the 2.5.x branch very soon. XFS' adoption into the stock kernel has never been about "lack of maturity," but the requirement of various interfaces. These new patches in 2.5.x make SGI's patching effort much easier now. -- Bryan -- Bryan J. Smith, Engineer mailto:b.j.smith@ieee.org AbsoluteValue Systems, Inc. http://www.linux-wlan.org SmithConcepts, Inc. http://www.SmithConcepts.com --------------------------------------------------------- 1999 IRS Data: The top 1% of income earners pay over 36% of the taxes, but have less than 20% of the total income.
