From: Herbert P÷tzl (herbert_at_13thfloor.at)
Date: Tue 19 Aug 2003 - 15:04:26 BST
On Tue, Aug 19, 2003 at 08:24:26PM +1000, GuruJ wrote:
> Sam Vilain wrote:
> > - UID24/GID24 (works on all 32bit U/GID FSs)
> > the format uses the upper quarter of user and
> > group id to store the context information, again
> > transparently. you'll end up with 16 million
> > user and group ids, which should suffice for the
> > majority of applications ...
> >Quite a few systems create one group per user by default, so that they
> >can leave users wandering around with a 002 umask without worrying
> >about security. So if the external method doesn't work out, my vote
> >would go for this option.
> I prefer the UID24/GID24 option. I can't see any reason why having
> 'only' 16 million users and groups could be limiting within a virtual
> context situation, and it would mean that we were just making more
> efficient use of UID/GID identifiers. Even within different contexts,
> they are still just users and groups, right?
> Also, wouldn't there also be a smaller chance of future conflicts if
> using UID24/GID24? That 'unused' part of the inode may get used at some
> point, but UID/GID values are unlikely to be re-mapped in the future.
to make this clear, this was not an opinion poll *smile*
all three formats should be available from now on,
it's a config option, so you simply decide which to
use ... if you like UID24/GID24 go for it!
> -- GuruJ.