Re: concerns about object-inheritance wrt files/directories

On Wed, Mar 16, 2005 at 08:45:48AM -0500, Stephen Smalley wrote:
> On Wed, 2005-03-16 at 13:50 +0000, Luke Kenneth Casson Leighton wrote:
> > for file types, i have a major concern.
> >
> > if not careful, we are going to be getting into the "multiple file
> > types" thing again, which it has already been discussed that, whilst
> > doable, has some hair-raising implications that make it too complex to
> > a) understand for implementation purposes b) understand the resultant
> > policy.
>
> No, I think you misunderstood. The objection to multiple contexts on
> files is that you can't analyze the resulting policy without pulling in
> the entire filesystem state, and a modification to the filesystem state
> at any time can change the policy.

 sorry, yes: i remember the objection but not the details.

 i remember that we disagreed over the conclusion because i  could see that what you were proposing and what i was proposing  were topologically equivalent (as best i could make out),  such that one of us _had_ to be wrong :)

 hence my warnings that kaigai's scheme could also be another  [topologically equivalent] scheme where filetypes could also  get into trouble - or not - depending on whose conclusion is  correct.

 i be watching the discussions with beady-eyed interest -  without comment unless invited - to see what happens.

 l.

--
This message was distributed to subscribers of the selinux mailing list.
If you no longer wish to subscribe, send mail to majordomo@tycho.nsa.gov with
the words "unsubscribe selinux" without quotes as the message.