View Issue Details
| ID | Project | Category | Date Submitted | Last Update | |
|---|---|---|---|---|---|
| 0021396 | AI War 2 | Gameplay Issue | Jul 13, 2019 12:50 am | Jul 16, 2019 1:14 pm | |
| Reporter | BadgerBadger | Assigned To | Chris_McElligottPark | ||
| Severity | major | ||||
| Status | resolved | Resolution | fixed | ||
| Product Version | 0.875 Counterattacks You Can Taste | ||||
| Fixed in Version | 0.876 Counterattacks Ate My Waves | ||||
| Summary | 0021396: No waves are generated | ||||
| Description | The commit * When waves are created, they now have a specific planet that they use from which to launch their wave. prevents conventional waves from being spawned. In PlanWave() the assignment: int tempBudget = 0; Options.approxWaveStrength = this.approximatePotentialStrengthOfWave( faction, Context, budget, out tempBudget, numUnitTypes, ShipGroupCat, maxGuardianTypes, Options, null ); was moved below ArcenSparseLookupPair<Planet, Planet> pair = this.ChooseWaveTarget( faction, Context, budget, Options ); The AI needs to know the approximate strength of the wave when deciding which planet to target. The AI needs to know the approximate strength of the wave so it can avoid attacking particularly strong planets with weak waves, or how it makes sure reconquest waves hit a target they think they can beat. | ||||
| Tags | No tags attached. | ||||
| related to | 0021405 | resolved | Chris_McElligottPark | No waves for diff 10 |
|
|
Whoops! That's going to be challenging, because the strength of the wave is now based on where it's coming from. So we'll have to make it calculate that for all the potential targets. |
|
|
Not just the composition of the wave is determined by. the source planet, but the strength as well?! Keep in mind that it doesn't need to be the exact strength, just an approximation. |
|
|
It's implicitly the strength as well, yes -- since the costs don't directly determine the strength, if the xml diverges it will potentially be very off if balance is adjusted various ways. The budget isn't determined based on where it comes from, but if you recall the budget!= the strength now. And especially with multiple ais, it could get pretty funky since things like royal ais use way more strength for the same budget. And based on how I have it set up, if a different AI recaptures a planet that was royal, they'll use the royal ships from that planet at the moment. So... I am just trying to avoid having the code make assumptions like "the modders and developers will balance costs and strength consistently." That's demonstrably not the case, heh. And even as we get into more extreme AI types I imagine it might intentionally diverge further. Really what we should be doing, and I can work on it, is that we build more than an estimate for the potential waves. Build the wave compositions and hold onto those, and so that way when we choose one to use later on, it's definitely what was advertised. I'll have to do some refactoring, but nothing too major to my knowledge. |
|
|
I think a refactor is appropriate; once strength was decoupled from AI budget the old paradigm didn't make as much sense. |
|
|
Thanks! * Really major refactor of how waves are generated, since it needs to know what is in the wave before it can decide what to do with it... and it also needs to know where the wave would be coming from now before it knows how strong it is and what is in it... so basically a lot of things had to be reorganized in order to make it so that it can tell which end is up. ** The overall logic should be the same, but it's just more accurate now in terms of actually using real numbers for strength calculations rather than trying to use estimates or approximations. The waves can vary a lot more than we can accurately estimate now, so that was definitely needed. ** The two major downsides of this revision are firstly that this does a lot more calculations on a background core of your CPU, but frankly it still should be pretty trivial; and secondly that there's a good chance we introduced some major new bugs with waves in all this. ** We have tested this and basic waves function as well as apparently reconquest waves, but we haven't checked hacking waves, waves in response to alarm posts, etc. Knock on wood that those should work. This is at least a step up over last version, where it was sending no waves at all because the approximation code got moved around without a full refactor (oops). |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| Jul 13, 2019 12:50 am | BadgerBadger | New Issue | |
| Jul 13, 2019 11:42 am | Chris_McElligottPark | Note Added: 0052223 | |
| Jul 15, 2019 9:38 am | BadgerBadger | Note Added: 0052228 | |
| Jul 15, 2019 10:54 am | Chris_McElligottPark | Note Added: 0052230 | |
| Jul 15, 2019 10:54 am | Chris_McElligottPark | Assigned To | => Chris_McElligottPark |
| Jul 15, 2019 10:54 am | Chris_McElligottPark | Status | new => assigned |
| Jul 15, 2019 10:57 am | BadgerBadger | Note Added: 0052231 | |
| Jul 16, 2019 12:11 pm | BadgerBadger | Relationship added | related to 0021405 |
| Jul 16, 2019 1:14 pm | Chris_McElligottPark | Status | assigned => resolved |
| Jul 16, 2019 1:14 pm | Chris_McElligottPark | Resolution | open => fixed |
| Jul 16, 2019 1:14 pm | Chris_McElligottPark | Fixed in Version | => 0.876 Counterattacks Ate My Waves |
| Jul 16, 2019 1:14 pm | Chris_McElligottPark | Note Added: 0052268 |
