Re: DogCows or Polymorphism in the Hurd

[Jonathan S. Shapiro]

On Wed, 2006-02-08 at 11:16 +0100, Filip Brcic wrote:
> Дана Wednesday 08 February 2006 10:44, Tom Bachmann је написао(ла):
> > Patrick Negre wrote:
> > >>Marcus Brinkmann wrote :
> > >>You have to bind a view to a
> > >
> > > If a path to a file contain a non-default branch-translator, the
> > > branching must be specified by appending to the node the translator that
> > > have to be use. Ex:     "foo.tar.gz:gzip_tr/setup.exe"
> > > This make paths globals and unambiguous.
> >
> > Someone already came up with this idea. It is actually quite nice, but
> > the problem is that POSIX only specifies one reserved character, '/',
> > and thus foo.tar.bz2:as_dir is a valid name. So such a file could exist.
> 
> Well, it could exist, but it probably will not. If it does exist, the system 
> can decide to use foo.tar.bz2:as_dir1, foo.tar.bz2:as_dir2, ... until it 
> finds the name that doesn't exist.

My main objection to this line of proposal is that you have effectively
embedded a policy, and it would be simpler and less confusing to just
let the user bind a user-selected name in the first place.

In either case, you have implicitly accepted the idea that "new
interface gets newly bound name."

shap