Such a translator would indeed make sense, though the double // is
abit tricky. Instead, it would make more sense to provide a general translator, lets call it transmux
which would allow you do do things like:

cd /transmux/ext2fs/some/place/image

and it would `mount' /some/place/image, without the need to do any
fancy running of commands like settrans. So for example with your example
of ftp, one would do:

cd /transmux/ftpfs/anonymous@foo:ftp.gnu.org/gnu/

to get access to ftp.gnu.org. This would be far more flexible, and it
wouldn't depend on any fancy URI schemes, which are hard to express in a good way due to
double and tripple slashes, which (I haven't checked POSIX, so I might be wrong)
have to be removed unless they are in the first part of the file name (//foo is
I think undefined in POSIX, while /foo/bar has to be the same as /foo/bar)

For future reference, help-hurd@gnu.org and bug-hurd@gnu.org are the
places to ask these questions,most of the developers of the Hurd
don't follow kerneltrap so questions like yours might get ignored, which
would be a sad thing.

Alfred M. Szmidt -- Hurd and GNU hacker.

First off, the paper is historic, so it will not get changed. Your question is interesting nevertheless, of course.

> not sure where to post this.

Probably best to post this on the bug-hurd mailing list, which is read by the Hurd designers and also the person who wrote the paper you cite from. This does not guarantee that somebody with insight answers, but it improves the chances :)

Mail bug-hurd@gnu.org

Michael

Now the only thing I dislike about Hurd is the lack of an easy forum. Unfortunately mailing lists require a working email address. ;(