What does veterancy add to the game?

[ZenTractor]

If you have veterancy based on proportional damage dealt, then it's a single step away from an EXP system such as you have in other games, such as Etherium. The only difference is that the experience is granted on kill, and not progressively as damage is done. Implementing that would also use less memory, since you can update veterancy status as damage is dealt, instead of storing it until the kill is actually scored.

Also, what happens with regen and repairs when you are recording damage for veterancy? Could you end up scoring double from a brick that had been in a few fights and only died later? How would overkill work? Will Percivals score 'bonus' veterancy from one-shotting T1 tanks? Similar problems might arise with overcharge.

As an extreme case, if one player had a single brick, next to a large number of hives and had a donated paragon somewhere, and their opponent had a monkeylord lasering it for ages, would the monkeylord snap up to full vet when the brick finally died?

Also, what happens if an almost-dead experimental is self-destructed moments before being killed. Does veterancy for damage even get scored?



Personally I am fond of SupCom's kill-based veterancy, because it sets it apart from other games with arcane exp system. I also like following hero units around and seeing them be amazing.

[IceDreamer]

If you have veterancy based on proportional damage dealt, then it's a single step away from an EXP system such as you have in other games, such as Etherium. The only difference is that the experience is granted on kill, and not progressively as damage is done. Implementing that would also use less memory, since you can update veterancy status as damage is dealt, instead of storing it until the kill is actually scored.

Also, what happens with regen and repairs when you are recording damage for veterancy? Could you end up scoring double from a brick that had been in a few fights and only died later? How would overkill work? Will Percivals score 'bonus' veterancy from one-shotting T1 tanks? Similar problems might arise with overcharge.

The troubles you've outlined here are prevented by my system. Unlike those games, damage itself is not used to calculate the XP. Instead, the XP given out would be proportional to mass, with damage only influencing the way it's shared around.

As an extreme case, if one player had a single brick, next to a large number of hives and had a donated paragon somewhere, and their opponent had a monkeylord lasering it for ages, would the monkeylord snap up to full vet when the brick finally died?

So as the amount of XP would be limited by the mass cost of the unit, not the total damage dealt, this wouldn't happen. For example, say we had a unit with 300 HP killed by two others. One did 200 damage, the other 100. I'd get the total damage dealt (To account for regen in the target), then do 200/total and 100/total to get 2/3 and 1/3 respectively. Then each of the attackers gets XP relative to the dead unit's mass * that multiplier.

Also, what happens if an almost-dead experimental is self-destructed moments before being killed. Does veterancy for damage even get scored?

Yes

Personally I am fond of SupCom's kill-based veterancy, because it sets it apart from other games with arcane exp system. I also like following hero units around and seeing them be amazing.

This isn't going to stop that. In fact, getting rid of the whole "1 ASF got credit for that Czar death" thing is going to make 5-star vet more common in normal units

[E8400-CV]

I don't like the engie vet idea. I don't know to what goal it would be usefull (reduce number of T1 engie needed ? to have a smoother game ?). Reduce the number of T1 engie, can be done by way more interesting ways. And i don't see why an engie that have been building asf since 10 min, would be more able to build any other unit, and why it should have more health than a new engie. doesn't make a lot of sense for me.

This would make engineers more valuable in the late game as well as make T1 engineers progressively better over time. Hence the loss of a few engineers could have a profound impact on build speed as the replacement engineers lvl up.

Now as for performance this should have any SIM impact what so ever as the engineers are only checking to see if they have leveled up on build completion.

Resin

I think you meant something different in that bold part...

Absolutely not! The whole point is that people should stop spamming 300 engies to build units. It lags like crazy. I'd rather see the slave facs reduced in cost another 30-40%.

You can see this perfectly fine when "A_Naked_Girl" (the player) plays air on Setons (or Ozone, or...) compared to where I play that spot. With same number of ASF, the game speed is significantly higher with my build, that uses around 20 slave facs, compared to his/her build that uses 400 engies cluster fucked around 4 facs...

[ZenTractor]

-Snip-

Alright, I think I'm getting a hang of it. It's always tricky to read new code, I've not done LUA before, and there's just so much to supcom that it's hard to tell which variables you're defining and which you're using.

So if I am reading things right, there's still two cases that might cause problems or unintended behaviour:

• If a striker plinks away at a mantis and gets it down to 1 hp, and then a percival comes along and overkills the mantis, the percival will get the lion's share of the exp?
• If a monkeylord strays too close to some ravagers, and then is called back and self destructed, all the ravagers will rank up?

[IceDreamer]

Good points. Yes, it will, but that can be remedied. And yes, because it did damage to the ML.

[ZenTractor]

Good points. Yes, it will, but that can be remedied. And yes, because it did damage to the ML.

It might actually be worth keeping the overkill functionality in: overcharges will give the ACU priority for veterancy, which I suspect a lot of players would want. That's kind of bordering on a balance question, though, so someone who is better at the game than I should put some thought into it.

Forgive my slightly off topic question, but how much will these changes to veterancy affect the game's running speed? It's only changing a few details, but it still introduces a few new functions and data arrays for each unit created.

[IceDreamer]

Sim impact should be negligible, slight memory hit, but nothing too bad.

[codepants]

Sim impact should be negligible, slight memory hit, but nothing too bad.

I'd be curious what the current Big O (number of calculations in terms of number of desired results) is and what the new one would be. I think it's easy to say "not too bad" but then later find out you went from 2N to N^2. Remember that you aren't just accessing the unit that did the kill and adding one to their kill count but you have find and access the unit they killed and get something from it -- so you are, at the least, doubling the number of memory accesses. Given that supcom only runs on one processor and many folks complain about cpu lag (I once had someone tell me if I wasn't buying a new $2k computer every six months I wasn't doing it right -- yes, someone in a supcom game lobby said that to me, about playing supcom) I think anything that has a memory hit should be given more serious consideration.

My two cents.

[IceDreamer]

Well, I've tested what I've got (Which I'm sure can be improved) on my old Phenom II tricore, and saw no decrease in simspeed across two near-identical scenarios (As close to identical as I could make them) using several hundred units in combat. Didn't see much memory hit either.

Think is, I'm only creating one additional table per unit, which isn't much at all they have so many, and it only checks everything and splits things up when the unit dies. The fewer units contributed to that (Such as T1 VS T1, which is where you have loads of units, but everything dies fast so only a few things did anything), the lower the hit of the spread calculation because it has fewer things to dish out to.

[da_monstr]

As far as I see it, vet is great for the commander and mediocre for normal, "spam", units. However, the whole idea of ML or SR killing a swarm of T1 engies to restore a huge amount of HP is just ludicrous. (see watch?v=thQkBrDAfgs for a perfect example)