NT file-permissions

[Jeremy Allison]

Luke Kenneth Casson Leighton wrote:
> 
> > > The latter, however, is probably a good idea.  Note that, as there is no
> > > apparent standard for ACL implementations under Unix, a separate
> > > implementation is needed for each OS.  I've been playing with some ideas for
> > > an HP-UX implementation, myself.  Probably what's needed first is a set of
> > > generic utility functions to get/set NT ACLs on Unix files, with the actual
> > > implementation varying depending on the host OS.
> >
> > Yep, that's exactly what I'm planning for a future release.
> >
> > I intend to abstract the ACL interfaces (which are different
> > on almost *every* UNIX, as the POSIX committee failed to
> > agree - arghhhhh !) in the same way as the lib/system.c
> > calls.
> 
> i would recommend that the API have exactly the same arguments as the
> Windows NT MSDN API, and be redirected / mapped from there to the various
> UNIX ACL APIs.

Luke,

	Why do you keep wanting to re-create Win32 on
UNIX (isn't that what Wine is doing :-) :-) ? It isn't
even a *stable* API (the security api calls change every
WinNT release as they're *still* too complicated for any
application to actually use), let alone a good one :-).

There is a *draft* POSIX ACL API document, that I think
some of the UNIX vendors tried to use (I have it somewhere
in my docs directory, but it's a bit big). Using that as
a 'least-worst' effort would probably be the best idea (IMHO).

Jeremy.

-- 
--------------------------------------------------------
Buying an operating system without source is like buying
a self-assembly Space Shuttle with no instructions.
--------------------------------------------------------