[Freeciv-Dev] Re: Wonder obsolesence across games (PR#1434)

[Mike Kaufman <kaufman@xxxxxxxxxxxxxxxxxxxxxx>]

On Fri, May 10, 2002 at 10:06:48PM +0200, Raimar Falke wrote:
> On Sat, May 04, 2002 at 12:53:28PM -0700, dspeyer@xxxxxxxxxxxxxxxxxxxxx wrote:
> > Full_Name: Daniel Speyer
> > Version: latest CVS
> > Distribution: Built from source
> > Client: Both (or N/A)
> > OS: Debian GNU/Linux
> > Submission from: (NULL) (129.2.159.61)
> > 
> > 
> > If you start a game, build Sun Tzu's War Academy, end the game, and start a 
> > new
> > game using the same client, you will be unable to build barracks.  Other
> > wonders/improvements are presumably similar.
> > 
> > This is because the game.players[0].improvements array is not innitialized 
> > for
> > each game.  Probably the best solution is to put
> > 
> >   for(i=0; i<B_LAST; i++){
> >     game.players[0].improvements[i]=I_NONE;
> >   }
> > 
> > into the game connection routines, probably in try_to_connect
> > (freeciv/client/clinet.c).  It's a little ugly there, but it works.  I 
> > haven't
> > found another place to put it that works cross-gui.
> > 
> > I'd make a patch, but my clinet.c is pretty modified, and I'm still not sure
> > where in the function is least aesthetically offensive.
> 
> What about this patch:
> Index: client/clinet.c
> ===================================================================
> RCS file: /home/freeciv/CVS/freeciv/client/clinet.c,v
> retrieving revision 1.69
> diff -u -r1.69 clinet.c
> --- client/clinet.c   2002/04/09 14:10:09     1.69
> +++ client/clinet.c   2002/05/10 16:49:23
> @@ -105,6 +105,7 @@
>    
>    set_client_state(CLIENT_PRE_GAME_STATE);
>    agents_disconnect();
> +  game_init();
>  }
>  
>  /**************************************************************************
> 
> It is certainly correct ;) but may introduce other problems.

this is actually a dreadful fix. 

we already have a [smaller] memory leak when diconnecting and
reconnecting to the server (or quitting the server and restarting it).
this "fix" really intensifies the memory leak---depending on the
savegame size---by about an order of magnitude.

We should probably introduce a game_free() function that frees all the
lists and players and such before calling game_init()

-mike