[Freeciv-Dev] Re: client/server authentication (PR#1767)

["Per I. Mathisen" <per@xxxxxxxxxxx>]

On Tue, 25 Mar 2003, Mike Kaufman wrote:
> hmm, I think that this may be making the whole procedure too complicated.
> In any case, I think that this is a metaserver issue. A player in this
> circumstance is never going to able to join the game automatically, he'll
> have to connect and take the player whether the player's login is"joebob"
> or "Unassigned". The question is simply what the metaserver will say. It
> can easily be done such that if a person is /cut with prejudice, the cut
> command can rename the player (login) so that that user can't log back
> into the game directly, he has to go back through the staging area.
>
> I'm not saying that this isn't possible to do, it just that we don't want
> to bloat the command set uselessly. How often do you see this getting used?
> It seems like putting a message up on the metaserver: "Slots open" should
> be good enough.

Well, the big question is how this fits into the strategy for player
authentication. You're the one to answer this, so I'll await your next
patch for the answer.

> currently, the take command is level INFO so that anybody can take a
> player for himself. the global take is HACK. We could make it CTRL, but
> that's a really powerful command... eventually we'll have an game.allowtake
> that can prevent connections from taking dead players or any players at
> all.

I don't like this game.allowtake idea. Let's not propagate the hacks of
yesterday into this new auth code. With good code and good policies this
should not be necessary.

For all players, only connections/clients that were connected at game
start should be allowed to reconnect, all other take over attempts or
addition of another controlling connection to a player should be permitted
by unanimous vote.

  - Per