[Freeciv-Dev] Re: Agents and network

[Christian Knoke <chrisk@xxxxxxxx>]

On Tue, Nov 26, 2002 at 08:40:49PM +0100, Raimar Falke wrote:
> On Tue, Nov 26, 2002 at 07:20:46PM +0000, Gregory Berkolaiko wrote:
> > On Tue, 26 Nov 2002, Raimar Falke wrote:
> > 
> > > The client already knows how a city is recalculated. Note that this
> > > doesn't help since only the server state is important. So at some
> > > point the client has to "update" the server state.

Yes, your CMA tries to do things very "correct", maybe too much? We remember
this well from the many asserts the CMA has thrown in the past. Here's my
idea:

Just let the CMA calculate *once*. This will be, in the most cases, perfect,
because, there are at most a few seconds between setup and result. *If* the
result shows conflicts, just solve them locally, that is, correct the one
city with the misused tile, don't recurse through all cities. This maybe
not perfect in some cases, but saves you a lot.

Another possibility is to limit the CMA activity from the top: since the
workers deployment only affects at the end of turn, it should be enough
to do *one* recursion per turn.

I don't know whether this fits into your concept.

> > Which means you already moved a chunk of the server into the client and 
> > compatibility issues wil arise whenever somebody decides to change the 
> > rules (of, say, specialists / happiness relationship).
> 
> Yes. However I don't know a way to have a client-side AI which is
> still fast (don't ask the server a lot of question like "uhh what if I
> place a worker here. Do you think this is good? How much food would
> this give me?").
> 
> 
>       Raimar

Christian

-- 
Christian Knoke     * * *      http://www.enter.de/~c.knoke/
* * * * * * * * *  Ceterum censeo Microsoft esse dividendum.