[Freeciv-Dev] Re: cma assert at end of game (PR#2071)

[Mike Kaufman <kaufman@xxxxxxxxxxxxxxxxxxxxxx>]

On Thu, Sep 19, 2002 at 08:05:35PM +0200, Raimar Falke wrote:
> On Thu, Sep 19, 2002 at 07:30:40PM +0200, Raimar Falke wrote:
> > On Thu, Sep 19, 2002 at 07:28:00PM +0200, Raimar Falke wrote:
> > > > To reproduce, login as chris and press return.
> > > 
> > > I can't reproduce this here with current CVS.
> > 
> > Reason: the new dataio intrudces a new attribute format and the old
> > ones are removed -> no CMA. I will try with an older version.
> 
> Mhh we are fall above our kludge. Problem is that the CMA happily
> continues after the game is over. And because of this
>   /*
>    * We are currently ignoring the CLIENT_GAME_OVER_STATE state
>    * because the client hasen't been changed to take care of it. So it
>    * breaks the show-whole-map-at-the-end-of-the-game. Nevertheless
>    * the server is so kind and sends the client this information. And
>    * in the future the client can/should take advantage of this
>    * information.
>    *
>    * FIXME: audit all client code to that it copes with
>    * CLIENT_GAME_OVER_STATE and implement specific
>    * CLIENT_GAME_OVER_STATE actions like history browsing. Then remove
>    * the kludge below.
>    */
>   if (newstate == CLIENT_GAME_OVER_STATE) {
>     newstate = CLIENT_GAME_RUNNING_STATE;
>   }
> 
> the CMA isn't able to test for CLIENT_GAME_OVER_STATE in
> cma_core.c. Either we fix the above or I add another kludge like
> "real_client_state".

fix the kludge, don't add another.

-mike