Re: A future for the SSH File Transfer Protocol?

[Simon Tatham <anakin%pobox.com@localhost>]

Barney Wolff  <barney%databus.com@localhost> wrote:
> It strikes me that the whole notion of internal uid is the wrong way
> to go and just makes life complex for both client and server.  I'd
> propose instead that all conversation be in terms of external
> user/group names.

Yep, I'd sign up for that. Only trouble is, it would need an
incompatible protocol change. But if the politics of that can be
plausibly managed, it sounds like a good simplification.

(It wouldn't _completely_ remove the purpose of my extension,
because I'd still like to be able to retrieve the location of a
user's home directory. But it would let me condense the entire
extension down into one single SSH_FXP_HOMEDIR("fred") command,
which would be much nicer than the huge great proposal I've
currently got...)

> What is the point of making the client ever deal with the server's
> internal uid's?

As I understand NT, its security identifiers are quite complex, come
in several parts, and might not map exactly to a simple user name.
But then, if it came to that then the server could just produce a
textual representation of a SID and be done with it, so I think even
that isn't a good argument against your suggestion.

Cheers,
Simon
-- 
Simon Tatham         "The distinction between the enlightened and the
<anakin%pobox.com@localhost>    terminally confused is only apparent to the latter."