Re: [Ext2-devel] [RFC 0/13] extents and 48bit ext3

[Jes Sorensen <jes@sgi.com>]

>>>>> "Ted" == Theodore Tso <tytso@mit.edu> writes:

Ted> On Fri, Jun 09, 2006 at 12:55:09PM -0400, Jeff Garzik wrote:
>> 1) clone a new tree 2) cp -a fs/ext3 fs/ext4 3) apply extent and
>> 48bit patches 4) apply related e2fsprogs patches
>> 
>> Then update ext4 step-by-step, using the normal Linux development
>> process.

Ted> We don't do this with the SCSI layer where we make a complete
Ted> clone of the driver layer so that there is a
Ted> /usr/src/linux/driver/scsi and /usr/src/linux/driver/scsi2, do
Ted> we?  And we didn't do that with the networking layer either, as
Ted> we added ipsec, ipv6, softnet, and a whole host of other changes
Ted> and improvements.

Maybe it's just me, but I am reading oranges vs apples there. The SCSI
comparison is like suggesting we go from the VFS to a VFS2 or fs/ ->
fs2/ for this. On the other hand going from ext3 -> ext4 to get
something incompatible (like enabling extends or 48 bit) is similar to
going net/ipv4 -> net/ipv6, which we did do indeed.

Fact of the matter is that 2.4 is dead or at least frozen solid by
now. The userland of most distros today wouldn't be able to boot with
a 2.4 kernel anyway.

Granted I am not a filesystem expert, but personally I would feel more
comfortable deciding to put my data on an ext4 file system knowing
that it was just that, rather than a hybrid.

Jes