[Freeciv-Dev] Re: (PR#13878) redesign observers

["Jason Short" <jdorje@xxxxxxxxxxxxxxxxxxxxx>]

<URL: http://bugs.freeciv.org/Ticket/Display.html?id=13878 >

Per I. Mathisen wrote:

>>However there is still a significant problem: the server doesn't send
>>the right notifications when changing state.
> 
> What kind of state are you talking about here?

It's kind of a long story and I don't understand the whole system.  But 
here goes:

When the client changes "state" I mean that it moves from being detached 
to taking/watching a player, or from one player to another, or from an 
observer to detach.  (Maybe there's a better name for this here.)

When this happens the current method is for the server to send some sort 
of "reset" to the client that triggers it to free and reallocate all 
game data (map, units, cities, etc.).  In detach_command() this takes 
the form of

   if (server_state == RUN_GAME_STATE) {
     send_game_state(pconn->self, CLIENT_PRE_GAME_STATE);
     send_rulesets(pconn->self);
     send_server_settings(pconn->self);
     send_game_info(pconn->self);
     send_player_info_c(NULL, pconn->self);
     send_conn_info(game.est_connections, pconn->self);
   }

Now, this is very inefficient and if we can find a way to avoid it 
(simply a client_reset_game_data packet would do it), that would be 
great.  But that's not my point here.

The bug in the patch is that when going from observer mode to taking a 
player this reset is not sent.  This is rather odd since even before the 
patch this state change was possible and I don't know why the reset 
wouldn't have been sent.

Another potential bug I find is that in unattach_connection_from_player 
there is this code:

   pconn->player->is_connected = FALSE;

but this function is called even when detaching an *observer* from the 
player.  So this seems like a bug - however I don't know what 
is_connected is used for or what effects this could have.

-jason