On Tue, 08 Apr 2008 21:45:13 +1000, Christoph Hellwig <hch@xxxxxxxxxxxxx>
btw, a little comment on how to proceed. After your xfs_name patch it
would be good to get the nameops and leafn cleanup in as almost
independent cleanups and then the ascii only CI but already using the
complex and correct lookup method. I think unicocde will take a bit
longer to sort out and getting the simpler bit in first would be
Yep, that is how I was proceeding when working on the return name patch
with the xfs_vn_ci_lookup code.
So, it will be:
2 leafn refactor
4 ascii ci
5 unicode ci
With NLS, it would be very nice if the Linux VFS or some layer above
always pushed UTF-8 strings down rather than handling it in the
filesystems that require it (there are about half a dozen at least
that I'm aware of).