"Fossies" - the Fresh Open Source Software Archive  

Source code changes of the file "TODO" between
fstransform-0.9.3-src.tar.gz and fstransform-0.9.4.tar.gz

About: fstransform is a tool to change a file-system from one format to another, for example from jfs, xfs, or reiser to ext2, ext3, or ext4, in-place and without the need for backup.

TODO  (fstransform-0.9.3-src)	:	TODO  (fstransform-0.9.4)
TODO		TODO
* fstransform, fsmove: add persistence, i.e. the ability to resume a transformat		* fstransform: preserve filesystem LABEL and UUID. To collect them,
ion		run "lsblk -r -o maj:min,label,uuid" warning: special
after CTRL-C, power failure or similar interruptions		characters in label and uuid will be escaped as \xAB
		* all: test on more file systems
		nilfs: ok as target.
		as source it gets practically full very quickly, free space fluctu
		ates
		between zero and 1GB while the GC works to avoid "disk full" error
		s.
		But it works :)
		ntfs: ntfs-3g is VERY slow both as source and target because it's usersp
		ace.
		it supports only UTF-8 filenames, not arbitrary bytes in the range
		128-255.
		fsmove must check beforehand!
		use mount.ntfs option big_writes... enough or still too slow?
		logfs: as target: "mkfs -t logfs" does NOT accept option -q,
		fsmove and umount crash with kernel bugs (see tests/)
		others: btrfs, gfs, gfs2, hfs, jffs2, ocfs2, qnx4, qnx6, ubifs...
		not yet: tux3
		* fstransform, fsmove: add fsremap-like persistence, i.e. the ability to resume
		a transformation after CTRL-C, power failure or similar interruptions
* fsmove: use a larger buffer (1MB) to speed up backward copy.		* fsmove: use a larger buffer (1MB) to speed up backward copy.
beware of low free space!		Worth it? beware of low free space...
* fsremap: add a graphical interface ? probably not so worth it...		* fsremap: add a graphical interface ? probably not so worth it...
		* all: port to FreeBSD. see this:
		Procedure to mount file system images under FreeBSD
		First you have to create virtual device, using this command:
		# mdconfig -a -t vnode -f /path/to/image.iso -u 1
		The "mdconfig" command creates a device and prints the name. Just
		append that to "/dev/" and use it for whatever you need:
		# newfs /dev/md1
		# mount /dev/md1 /mnt
		To unmount the image use command:
		# umount /mnt
		Once unmounted, you can unconfigure the device with
		# mdconfig -d -u 1
		Also, many system files, as /lib/libc.so.7 are set as
		'simmutable' (cannot open O_RDWR) and 'sunlink' (cannot unlink)
		so fsmove fails
		Furthermore, fsmove df-like stats do not work on FreeBSD:
		they seem to report usage of root file system
		Additionally, fsremap wraps calls to ff_posix_mkdir() with
		#ifdef __USE_POSIX, which is undefined by default on FreeBSD,
		causing ff_mkdir() to fail with ENOSYS.
		Finally, 'umount' wants mount-point instead of device.
		And how to 'mount -o remount,ro <mount-point>' ?
WONTDO		WONTDO
* fsmove: if file system free space is critically low, launch 'xfs_fsr'		* fsmove: if file system free space is critically low, launch 'xfs_fsr'
to try and free some space before giving up.		to try and free some space before giving up.
why not:		why not:
LIMITED USEFULNESS: only for 'xfs' file system		LIMITED USEFULNESS: only for 'xfs' file system
DANGEROUS: at least on linux with an almost-full source device		DANGEROUS: at least on linux with an almost-full source device
xfs_fsr can WORSEN the problem by triggering 'loop write error' kernel err ors,		xfs_fsr can WORSEN the problem by triggering 'loop write error' kernel err ors,
which mean the source device has not enough space to accommodate the loop file contents.		which mean the source device has not enough space to accommodate the loop file contents.
typically this CORRUPTS the file system inside target (loop) device!		typically this CORRUPTS the file system inside target (loop) device!
End of changes. 3 change blocks.
4 lines changed or deleted		57 lines changed or added

---