Filesystems [was: Microsoft prohibits GPLed work via licensing of CIFS standards]
Jeroen Dekkers
jeroen at dekkers.cx
Mon Apr 8 17:09:56 UTC 2002
On Mon, Apr 08, 2002 at 06:49:56PM +0200, Jeroen Dekkers wrote:
> On Mon, Apr 08, 2002 at 05:44:32PM +0200, Tomasz Wegrzanowski wrote:
> > On Mon, Apr 08, 2002 at 05:25:39PM +0200, Jeroen Dekkers wrote:
> > > > BTW. have you looked at Reiserfs's plans ?
> > >
> > > No, I don't have the time for that. But I'm certainly interested.
> > >
> > > > It seems that it will
> > > > include something like translators soon. They plan to use it mostly
> > > > for compression-on-fly and encryption-on-fly but it is going to be
> > > > more general than that.
> > >
> > > You can never implement translators like we do it in a kernel.
> >
> > Alike or not, functionality will be similar.
>
> No, read some Hurd documentation to what translators exactly are. For
> example, a normal user can't add his own written compressing algoritms
> without permission from a privileged user.
Reading this again I see that this sentence is very vague. This is my
second try.
In the Hurd, all filesystems run in user-space. Our filesystem is also
our IPC namespace, this is very flexible. It's possible for every
normal user to attach a program to a node in the filesystem, that
program extends the filesystem, it can just provide a file or a
whole directory hierarchy. The use of this is endless.
I don't see how you can ever do this in a kernel. For example, you
can't allow an user to use his own written compression algoritm
because that would place untrusted code in kernel-space.
Jeroen Dekkers
--
Jabber supporter - http://www.jabber.org Jabber ID: jdekkers at jabber.org
Debian GNU supporter - http://www.debian.org http://www.gnu.org
IRC: jeroen at openprojects
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 232 bytes
Desc: not available
URL: <http://lists.fsfe.org/pipermail/discussion/attachments/20020408/6daf157a/attachment.sig>
More information about the Discussion
mailing list