Could someone explain:

[jc173]

Even if all the galaxy data is stored in a database, isn't it still going to be slow? The game engine will have to load up relevant data from disk\database to main memory for the AI's so they can run whatever their various strategy algorithms are.

[Suicide Junkie]

The database will only speed up searching for data and retrieving it.

If your AI is scanning through the list of known planets, deciding where to attack, you still have to look at every one.
That's what is gonna take a long time. When the AI gets star maps from an ancient race, trouble brews.

[Taqwus]

It wouldn't necessarily take a full sequential scan for determining attack locations, since the plausible attack range is usually small.

Supplies are a concern, at least until solar collectors or the QR. A fleet *could* be launched beyond resupply range, but it would need to either rendezvous with a supply source along the way (which might be beyond the AI); bring a colonizer to create a resupply colony (which would be a nifty tactic to see it use -- but it would have to defend it well, or risk stranding the fleet); or capture a resupply colony (which takes intel, to figure out where one is, *plus* the WPs and so forth may be a problem).

So if it's limited to resupply range, that's basically a series of range Queries, which won't need to touch many disk pages that only deal with far-away systems -- at least with a decent spatial indexing method.

------------------
-- The thing that goes bump in the night

[jc173]

Wouldn't you have to create a new set of indices or pointers for each race with that method? Unless I'm misunderstanding what you mean by spatial index