Re: VFS integration with kernel mounts
- From: Alex Jones <alex weej com>
- To: Alexander Larsson <alexl redhat com>
- Cc: "gnome-vfs-list gnome org" <gnome-vfs-list gnome org>, "gtk-devel-list gnome org" <gtk-devel-list gnome org>, David Zeuthen <david fubar dk>
- Subject: Re: VFS integration with kernel mounts
- Date: Wed, 02 May 2007 17:47:58 +0100
Hi Alexander
On Wed, 2007-05-02 at 14:58 +0200, Alexander Larsson wrote:
> On Wed, 2007-05-02 at 00:19 -0400, David Zeuthen wrote:
>
> > This should be a piece of cake to do (assuming HAL) and if people think
> > it's a good idea (I'm not entirely convinced it is) I'd be happy to take
> > a stab at it when GVFS is ready for this. Alex?
>
> I'm not really convinced it is a good idea. Seems a bit fragile and
> bound to case trouble in various edge cases. Aliasing filenames is
> always problematic, so is using filenames in a non-canonical way that
> not all apps support.
One thing to consider is that apps that remember URIs of files on
portable media will be mislead if we continue to use /media/usbdisk-n
convention.
For example, I might have a file /Document on one UMS device, open it in
AbiWord, close it all down and swap out the disk for another, which also
has the file /Document. These would both have exactly the same URI.
I think we do have mechanism in place to identify the physical device
some file belongs to, but the whole idea of a URI (not a
URI-Reference[1]) is that it is Universal. Therefore, I strongly believe
there should be some distinction between the two files' URIs, either by
changing the naming policy on the mount point to /media/some-volume-uuid
or by simply devising a new addressing scheme, URI or otherwise.
Personally I'm of the opinion that cramming everything into the POSIX
model doesn't work, as it's not generic enough for DAV/VFAT/FTP/SMB/etc
to work without lots of dodgy policy with regard to permissions, file
ownership and naming. Either something massive could go into the kernel
or GVFS could deal with this all in userspace, but either way this is a
massive shock to the system and I don't expect this to ever happen, even
though it would make things nicer.
Consider being able to access files in a tar-bzipped FAT filesystem
image that is accessed over DAV with a single address. (The addressing
scheme here is made up on the spot. XRI might be a better scheme but I
don't know that much about it.)
* "dav://myserver/mystuff.bz2" addresses the bzip image on the DAV
share
* "bzip2:(dav://myserver/stuff.bz2)" addresses the uncompressed
file
* "tar:(bzip2:(dav://myserver/mystuff.bz2))" addresses tar
contents
* "tar:(bzip2:(dav://myserver/mystuff.bz2))/myVFAT" addresses a
file with name myVFAT inside the tar archive
* "vfat(tar:(bzip2:(dav://myserver/mystuff.bz2))/myVFAT)/some/path/to/document addresses a file inside the file system
Imagine the possibilities!
[1] http://gbiv.com/protocols/uri/rfc/rfc3986.html
>
> =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
> Alexander Larsson Red Hat, Inc
> alexl redhat com alla lysator liu se
> He's a globe-trotting amnesiac cowboy whom everyone believes is mad. She's a
> radical green-skinned queen of the dead looking for love in all the wrong
> places. They fight crime!
>
> _______________________________________________
> gnome-vfs-list mailing list
> gnome-vfs-list gnome org
> http://mail.gnome.org/mailman/listinfo/gnome-vfs-list
--
Alex Jones
http://alex.weej.com/
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]