Re: New testing game

[Lincoln Peters <peters2000 at mindspring dot com>]

Having performed tests using both Space Civilization and the new testing
game, I have not once seen the AI build a single facility, not even when
common sense would dictate that a player build a particular type of
facility (e.g. an aqueduct).  As a result, it seems that no matter what
condition the facility_worth code is in, if the condition of
PLAN_IMPROVING or PLAN_BUILD goes unchanged, the AI performance in these
games will remain at its current, horrendously dismal level.

In the facility test game, I seem to have discovered that the
country-separation-min and country-separation-max variables are not
always respected.  Although I've gotten almost everything else to work.


On Tue, 2002-08-06 at 03:40, Hans Ronne wrote:
> Then I am at a loss to explain this. Citites do grow beyond size 9 in other
> games (e.g. advances). Could this be some kind of numeric overflow problem
> due to the large number of facilities (20 farms and 20 granaries) that you
> added in this specific case? What happens with just one facility of each
> type?

I tried building 20 refineries one time (which should have increased
fuel production in a size 4 city from 4 to 4,194,304 [4*2^20]), and
found that fuel production dropped to zero.  Looks to me like it is
indeed an overflow (I hadn't previously thought to do the math and see
what it should have been producing).  Perhaps the variables need more
bits to work with

> 
> >If I disabled the "supply" code, would that prevent cities from sharing
> >resources (i.e. fuel) with occupants?
> 
> Not at all. Sharing of resources with occupants is handled by run_economy
> (which can also be disabled if you wish). The supply code is some kind of
> pipeline machinery for distributing stuff over larger distances. There are
> problems with it, which is why I ususally disable it.

That seems to work in the test game, although I imagine that disabling
the supply code in Space Civilization would cripple it.

One thing I am seeing in Space Civilization is that an advanced unit
that can share something with nearby units oftentimes will not share as
much as it should.  This is most apparent with shipyards, because they
depend on an outside supply of ores and solar to be built, and the
nearest civilization will not supply as much as it could, even if it's
over it's own capacity!

> 
> >I set the 1000-unit limit on fuel because if there was no limit (or the
> >limit was absurdly high), cities would probably hoard fuel and never
> >supply it to the units that needed it.
> 
> This is not a problem. I have tested it in another game, and sharing of
> fuel with occupants is not affected by how much the advanced unit can
> store. The occupants are filled up with whatever fuel is available.

Are you sure about that?  When there are facilities that affect
production, fuel doesn't always seem to go to the occupants that need
it.  Although that's probably not a problem with storage limits.

> 
> No. The occupant (unit2) comes after the city (unit). Your tables should be:
> 
> (table occupant-multiply-maxcells (city arcology 200))
> (table occupant-add-maxcells (city collector 1))
> 
> Hopefully, things will work then. BTW, this is one example of the unit
> confusion that we were just discussing on this list. It would be much
> easier if units were named "transport" and "occupant" instead of "unit" and
> "unit2".

It seems to be working now.  If a size 9 city has an arcology and a
collector, it can cover 19 (2*9+1) cells