From: Qi Xia (qx@shire.math.columbia.edu)
Date: 01/25/93


From: qx@shire.math.columbia.edu (Qi Xia)
Subject: ANNOUNCE xfs-0.6.1 test release
Date: Mon, 25 Jan 1993 23:06:59 GMT

ANNOUNCE xfs-0.6.1 test release.

l99.4-xfs6.1.note (this file)
l99.4-xfs6.1.tar.Z

has been uploaded to tsx-11.mit.edu, sunsite.unc.edu incoming directory.

This is a test release.

xfs-0.6.1 sacrifices two direct block pointer in each inode for ctime
and atime. xfs-0.6.1 use three seperate time record.

This means the layout of xfs-0.6.1 is not compatible with xfs-0.6.
Now xfs has all the fields a "real UNIX" inode has as per Bach's book.
With the target (at least for me) that the Linux will be used as a
personal workstation, light multi-user system or a server on a LAN
with similar box, as NeXT does, I believe that this is the first and
also the last time of changing layout of xfs.

xfs-0.6.1 has run on my machine since last Saturday. I think I should
make it available for test before too may people jump into xfs.
I will not write an in-place converter, because in case there is a bug,
it will be defficult to figure out if the bug is in the converter or
xfs itself. I apologize for the extra work of whose who have tested
xfs-0.6.

I have added a short tutorial to reduce the confusion of how to install
xfs-0.6.1.

The primary reports of xfs-0.6 testing are quite encourage. Except
one tester failed to make it work after heavy hand-editing the patch
(He has many other stuff to add into the kernel), all other testers
(3-4) made their system with xfs work fine so far. There
are no bug found in the kernel part. Three silly bugs found in the
xfsck. They are fixed in 0.6.1.

I only have IDE disks. I can not test if xfs happy with disks have
bad blocks. If you are a xfs tester and has a non-IDE disk, please
let me know if it works.

Warning!!! This is still an ALPHA test release. You may experience
data lost. Keep this in mind.