SPEC : Write Support For NTFS

[Florent Mertens]

Hi,

Le mercredi 18 octobre 2006 à 23:46 +0100, Mark Shuttleworth a écrit :
> This is a longstanding feature request. If you think this is getting
> ready for prime time, it would be good if someone from your team could
> come to UDS Mountain View (Nov 5-11) to present it and make the case.
> 
> Mark

There is unfortunately no way that me or one of my co-writer can come to
Mountain view in November (if your 'team' refer to the linux-ntfs team,
i have to precise that i'm not really a member of this team, more an
obscure contributor ;) ) I can't really afford me such trip (even if i
'd like to :) ).
That's why i wanted this spec to be the more complete, and if you see
some outstanding issue, tells me them.
But in fact, i don't think my proposition require long discussion. As
you said, this is a 'longstanding feature request', but even if there
were lot's of progress since few months, there is still some issue that
prevent him to get a more official place that it can :

  * ubuntu is a mainstream linux distribution. Telling people that they
will use a driver based on reverse engineering as default read/write
driver for NTFS could scare many users. It will need some time to be
accepted (it's already headed in the right direction).

  * Currently, linux lacks an equivalent of chkdsk. If a user don't
cleanly unmount his device, this may leave the NTFS journal file
unclean, and the only way they have to repair it is to boot to windows.
This is IMHO unacceptable for a default installation.

  * Executing manually chkdsk can revel some inconsistency due to the
fact that the driver is not yet optimize, and so, the allocation size of
sparse files isn't set always correctly. Those are none fatal error, and
don't corrupt in any way the device, but can scary a lot of users.

What i think is that those issues needs to be fix before we can make it
default. What i propose is first a small step in the good direction in
the prevision of the big step that could happens in the future.

And this step require :
- some more official support of ntfs-3g -> bring it in main.
- some packaging effort, but that are quite easy.
- some changes in pmount, but that are already file
as wishlist bug, and martin pitt is aware of them.

So yes, those needs some discussion, but i'm sure others than us can
present the case, and my plan is to be on IRC (or other communication
channel) at this time.

What do you think ?

Regards,

Flo