Re: User/host name completion: how to deal with ambiguity?

[lawrence mitchell]

Kai Grossjohann wrote:

[...] tramp filename completion

> Michael's code now tries to look for all possibilities, so it tries
> to open a connection to the host ssh using the default method and the
> default user.  This, however, is wont to fail, leaving the user
> flabbergasted.

I haven't looked at Michael's code, and hence, I'm slightly
confused as to how it works.  In particular, how does one go
about completing host names?  (is it that one can specify a list
of hosts to which it is possible to connect?).

If this is indeed the case, then I would say that the "/ssh:"
example should be assumed to be an incomplete filename, using the
ssh method.

> Suggestions?

> Ideas:

> * If it looks like a method name, assume it's not a host name.  This
>   has the nasty consequence that "/flarbl:" works today as a
>   filename, but if you then define a new method named "flarbl", then
>   things will cease to work.

I'm not sure that's such a good idea either, as I (for example)
set up methods with the same names as the hosts I connect to,
it's easier to connect to multitudinous hosts this way.
For example, if I connect to host foo, then I have a method named
foo with the hostname hardcoded in.

>   On the other hand, /flarbl:~/ is unambiguous for the home dir on
>   the host, and it's only two additional characters...

I don't even think this would break most users' mental maps, I
for one nearly always type ~/ (force of habit I suppose).

> * Before trying to connect, look if the host really exists.  But what
>   happens if the host exists and allows connections using ssh, but
>   the user doesn't have an account on that host?

panic? :)

-- 
lawrence mitchell <address@hidden>