Roy Schestowitz wrote:
> So make it harder for the user (to break). Protect the user
from shooting
> itself in the foot. Security at the expense of ease of use has
proven to be
> a trade-off that can be tackled. You can restrict users
depending on level
> of experience or introduce the notion of user privileges
(which Windows was
> too late to introduce, quite artificially and improperly so).
>
Agreed, to a point. Permissions can be tough for a user to
handle effectively. A higher level "smart system" tool that
makes descisions for users might be useful. And untrusted
programs should be run, auto-magically, at *lower* than the
user's level, because the user needs protection too, not just the
system.
And I think this could be done with fewer restrictions preceived
by the users than many people believe.
> What about Beagle for some data indexing? It's included in
SuSE. I know it
> does not address your question directly, but also see:
> http://www.linuxjournal.com/article/7771 . You can extract meta
This is pretty cool. It's not exactly what I was thinking, but
it's a neat idea in it's own right. Beagle might be a good way to
integrate new meta-data into an existing file system. It could
act as a high level abstraction layer for work done lower down in
the file system.
Thanks for pointing that out. :)
> Pitfall: It will be lost upon data movemoment from one
filesystem to another.
> I already have this problem with a storage unit that uses
NTFS. I have to
> compress many files to preserve their length, case and some
'forbidden'
> characters.
If one implements multi-stream or meta-data as multiple "file
forks," then this problem can occur. If one implements MS/MD as
encoded onto a single stream, then this problem doesn't occur.
Look at Java .jar files.
> Are you in need of one? Or are you looking to make a start one
one?
A little of both. I think the open source community could
benefit from a multi-stream or meta-data format. I'm looking at
what's been done already before I make any plans to tackle this
myself.
|
|