Re: [Freeciv-Dev] Fantasy Modpack

[David Pfitzner <dwp@xxxxxxxxxxxxxx>]

Jules Bean wrote:

> Hmm.. interesting point.  But, if the trees were made so that there are
> multiple requirements for everything, then the effect would lessened. 

True.

> Alternatively, we could have a new option 'strict_techs'.  I can think
> of three possible implementations of strict_techs:

> 1) You can steal technologies, but they are useless (in that you can't
> build any of the things they permit, units or improvements) until you
> have researched (or stolen) all the technologies which normally lead up
> to them.  They are 'orphaned' knowledge - you have the blueprints, and
> you keep them safe, but you don't understand them until you have
> researched everything they depend upon.

Another possibility is that you can _use_ the tech, but cannot
research new techs based on it until you know all the pre-reqs
for the stolen tech.  (Presumably because you don't have enough 
in-depth knowledge, or tech reqs are implicitly recursive.)

I think people have suggested this even for normal games,
where gaining techs other than by research can sometimes lead 
to some pretty silly-seeming results.

> This has to be the way to go, IMO. Keep modpacks server-side, and send
> them to the client, as far as possible.  Give the modpacks unique IDs,
> and dates, so the client can cache them:
> 
> (Client joins game)
> S-->C "This game is using mod_pack 'COOL_FANTASY', do you have it?"
> C-->S "I have version 26th May 1999"
> S-->C "Now sending you version 16th June 1999"

Currently, the server always sends the client _all_ the current
ruleset data, _every_ time a client connects/starts, including
for the "default" ruleset.  No worries then about whether the 
client has the correct data :-)

Provided the amount of data is reasonable, I prefer the "send
everything always" approach, rather than the potential mess of 
(over)writing files on the client.  But the current approach 
wouldn't be appropriate for graphics...

-- David