DDraceNetwork

Broso56

specifically linear has a weird bug where it seems to merge 2 different teams/players into the same rank kek

(edited)

tested on sqlite local server with same db and works fine (edited)

idk how to test ddnet mysql on windows (edited)

i have "abuse 174" in fng

Tater

but I also don't think it should be applied to the regular move box functionality without elasticity because it might break those "automatic" map sections that rely on the specific floating point precision at those parts of the map. (edited)

That's why I suggested maybe counting up a mapversion (or similar) with a change like this. I don't think a map saves a ddnet version (or does it?) so you can apply fixed physics to newer maps and return to the old ones with the existing ones

07:31

But this is up to discussion, there is a good reason your "automatic" map sections will still work even with that change in place, do you have a list of this in order to test this? If only 1 map out of all ddnet maps is affected, I don't see a reason to make this backwards compatible

idk it just seems like such a fundemental change that I don't even know where you would begin testing it.

does ddnet have a demo validation/run validation? If you change physics and then re-validate some demos/run could be invalid

no it doesnt

07:38

but it changes a lot more than just auto parts actually

How does the counting in a list even help with the problem I mentioned? Where exactly would this list be filled? Maybe I skipped that

Parts that look like this will get harder after you start checking freeze interaction with move box more accurately (edited)

07:43

obviously the amount you move in a tick is exaggerated in that illustration but the effect still applies to smaller movements

07:46

actually your hitbox center doesn't go that high so it looks more like this

Jupstar ✪

How does the counting in a list even help with the problem I mentioned? Where exactly would this list be filled? Maybe I skipped that

the list would be filled by CCharacter::DDRacePostCoreTick()

    // handle Anti-Skip tiles
    std::vector<int> vIndices = Collision()->GetMapIndices(m_PrevPos, m_Pos);
    if(!vIndices.empty())
        for(int Index : vIndices)
            HandleTiles(Index);
    else
    {
        HandleTiles(CurrentIndex);
    }

(edited)

07:48

instead of checking between m_PrevPos and m_Pos. you check between each 2 positions in the list (edited)

Tater

the list would be filled by CCharacter::DDRacePostCoreTick()

    // handle Anti-Skip tiles
    std::vector<int> vIndices = Collision()->GetMapIndices(m_PrevPos, m_Pos);
    if(!vIndices.empty())
        for(int Index : vIndices)
            HandleTiles(Index);
    else
    {
        HandleTiles(CurrentIndex);
    }

(edited)

Oh that confuses me. Is the bounce thing a tile?

no it happens in MoveBox()

07:53

so you create the list inside MoveBox and pass it to CCharacter::DDRacePostCoreTick() somehow

Ah yeah but the important part is where it's filled. Because move box happens before player collision. So this bounce thing could trigger even is you land on a dummy with enough speed or not. But tbh maybe that isn't even really possible, since without bounce that would mean you skip into the dummy. I dunno if I ever seen it

hmm yeah

07:58

I guess there's other things that happen after move box

07:58

basically everything

Well maybe it still could. But that is why we some kind of test tool xd

Tater

actually your hitbox center doesn't go that high so it looks more like this

I'm pretty sure the instances like this would break a lot of demo replays

07:59

oh also edging becomes harder too

07:59

I forgot about that

Tater

I'm pretty sure the instances like this would break a lot of demo replays

I mean you could only count when you really are about to bounce and else return an empty list. But would make physics less intuitive

Jupstar ✪

I mean you could only count when you really are about to bounce and else return an empty list. But would make physics less intuitive

well if you only apply it to bouncing there's not any issues because it won't break old physics