On Fri, Aug 23, 2002 at 09:56:41AM -0700, chrisl wrote: > > If we are looking to use TEA, why even bother with endian-neutral > > half-MD4? I don't think _anyone_ has ext3 code which uses that, > > and it is not compatible with the existing half-MD4 code either. > > We may as well cut out complexity while we have the option to do so. > > Agree. If somebody do use some not compatible hash, we still can > can e2fsck to regenerate the hash for the directory? Given that TEA is faster, I agree, there's no reason to even bother including new half-MD4 code into the kernel. We should just simply go to TEA. Yes, we can use e2fsck to regnerate the hashes for all the directories. It's just a matter of setting the hash type in the superblock, and then running e2fsck -D on the filesystem. - Ted
