OdysseyDecomp
OdysseyDecomp copied to clipboard
MonsterDruide1
Enemy: Implement Bubble
137 functions, 2782 new lines, 238KB, 7 mismatches. This PR fully implements bubble although there are some serious mismatches remaining that need to be addressed. I spent nearly two weeks trying to fix these with no avail.
~~updateLavaWave https://decomp.me/scratch/ffVkSI~~ Fixed by monster ~~tryBoundMoveWall https://decomp.me/scratch/VidWI~~ Fixed by monster ~~updateHackOnGround https://decomp.me/scratch/d5cE6~~ Fixed by monster ~~makeDisplayQuatInHackJump https://decomp.me/scratch/EE9He~~ Fixed by monster accelStick https://decomp.me/scratch/ACufd exeHackJump https://decomp.me/scratch/P9h4T ~~exeHackResetPos https://decomp.me/scratch/7bYZR~~ Fixed
This change is 
Report for 1.0 (b83d2ce - be5d520)
📈 Matched code: 9.90% (+0.25%, +31576 bytes)
✅ 135 new matches
| Unit | Function | Bytes | Before | After |
|---|---|---|---|---|
Enemy/Bubble |
Bubble::receiveMsg(al::SensorMsg const*, al::HitSensor*, al::HitSensor*) |
+2592 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::control() |
+1816 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::constrainLavaDomain() |
+1652 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::init(al::ActorInitInfo const&) |
+1640 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::updateHackOnGround() |
+1348 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::exeHackResetPos() |
+1060 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::attackSensor(al::HitSensor*, al::HitSensor*) |
+1048 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::makeDisplayQuatInHackJump(sead::Quat<float> const&, sead::Quat<float> const&, sead::Quat<float> const&, bool) |
+916 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::exeHackMove() |
+772 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::exeHackInLauncher() |
+676 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::tryShiftLand() |
+660 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::exeUp() |
+600 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::tryBoundMoveWall() |
+600 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::exeDown() |
+584 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::endHackCommon() |
+572 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::updateLavaWave() |
+564 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::faceToMoveVec() |
+536 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::Bubble(char const*) |
+528 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::Bubble(char const*) |
+516 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::recalcClippingInfo() |
+508 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::exeHackLand() |
+484 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::exeReady() |
+472 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::exeWait() |
+436 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::revertTargetQuatInHackJump(sead::Quat<float>*, sead::Quat<float>*) |
+408 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::checkEffectPos() |
+396 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::headUp() |
+368 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::updateCollider() |
+364 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::tryShiftContinuousJump() |
+360 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::updateCollisionPartsMove() |
+352 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::shiftSink() |
+332 | 0.00% | 100.00% |
...and 105 more new matches
📈 2 improvements in unmatched functions
| Unit | Function | Bytes | Before | After |
|---|---|---|---|---|
Enemy/Bubble |
Bubble::exeHackJump() |
+680 | 0.00% | 25.00% |
Enemy/Bubble |
Bubble::accelStick() |
+353 | 0.00% | 25.00% |
Report for 1.0 (0db3e9a - b9b235a)
📈 Matched code: 9.61% (+0.24%, +30516 bytes)
✅ 134 new matches
| Unit | Function | Bytes | Before | After |
|---|---|---|---|---|
Enemy/Bubble |
Bubble::receiveMsg(al::SensorMsg const*, al::HitSensor*, al::HitSensor*) |
+2592 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::control() |
+1816 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::constrainLavaDomain() |
+1652 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::init(al::ActorInitInfo const&) |
+1640 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::updateHackOnGround() |
+1348 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::attackSensor(al::HitSensor*, al::HitSensor*) |
+1048 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::makeDisplayQuatInHackJump(sead::Quat<float> const&, sead::Quat<float> const&, sead::Quat<float> const&, bool) |
+916 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::exeHackMove() |
+772 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::exeHackInLauncher() |
+676 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::tryShiftLand() |
+660 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::exeUp() |
+600 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::tryBoundMoveWall() |
+600 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::exeDown() |
+584 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::endHackCommon() |
+572 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::updateLavaWave() |
+564 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::faceToMoveVec() |
+536 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::Bubble(char const*) |
+528 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::Bubble(char const*) |
+516 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::recalcClippingInfo() |
+508 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::exeHackLand() |
+484 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::exeReady() |
+472 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::exeWait() |
+436 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::revertTargetQuatInHackJump(sead::Quat<float>*, sead::Quat<float>*) |
+408 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::checkEffectPos() |
+396 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::headUp() |
+368 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::updateCollider() |
+364 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::tryShiftContinuousJump() |
+360 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::updateCollisionPartsMove() |
+352 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::shiftSink() |
+332 | 0.00% | 100.00% |
Enemy/Bubble |
Bubble::exeWaitHackFall() |
+332 | 0.00% | 100.00% |
...and 104 more new matches
📈 3 improvements in unmatched functions
| Unit | Function | Bytes | Before | After |
|---|---|---|---|---|
Enemy/Bubble |
Bubble::exeHackJump() |
+680 | 0.00% | 25.00% |
Enemy/Bubble |
Bubble::accelStick() |
+353 | 0.00% | 25.00% |
Enemy/Bubble |
Bubble::exeHackResetPos() |
+265 | 0.00% | 25.00% |
![decomp-dev[bot] avatar](https://avatars.githubusercontent.com/in/970722?v=4 "Published by a pub.dev verified publisher"){kind=small}
The first one matches with an asm("") hack in multiple variants:
https://decomp.me/scratch/t7ZYY
It is a different function than what you wrote in the description though 🤔
For the next one, it's somehow the memory of quat2 being reused for quat3, but I can't figure it out today.
This one is probably an inline function: https://decomp.me/scratch/hcTEn
The other three are pretty large mismatches, can you try to minimize those further yourself or have you already tried with seemingly no result?
I can retry to implement them from scratch again, all 3 are difficult to follow. "exeHackResetPos" has unreachable code that I can't see on ghidra this code is called on the else condition of normalize.
On a side note I'm starting to notice a pattern in broken quat functions and I think is due to this. Vector3 the know good one since is used everywhere uses setMul, quat doesn't even have setMul in the header. I will try to fix sead but seems out of the reach of this PR since I will need to verify both decomp projects still match
friend Quat operator*(const Quat& a, T t)
{
auto result = a;
result *= t;
return result;
}
friend Vector3 operator*(const Mtx33& m, const Vector3& a)
{
Vector3 o;
o.setMul(m, a);
return o;
}
I fixed most of the hard math's and branching on accelStick at the cost of some logic errors but the match jumped from 80 to ~~97%~~ I will keep trying on my own https://decomp.me/scratch/ACufd
Improved up to 99.33%. There's an inline that breaks pretty much everything besides that the rest is correct.
Significant improvements in exeHackJump https://decomp.me/scratch/q4PQG from 77% to 90% but these require the fixes of https://github.com/open-ead/sead/pull/216 so I can't commit those ones yet in this PR. The rest seems to be some weird vector operation and stack alignment issues.
accelStick: note the functional difference due to the wrong jump at 2d8 / 4a0c. To fix that one, insert stickAccel.set(0.0f,0.0f,0.0f); in the early if (mIsPlayerCaptured), which however introduces new mismatches.
My best attempt: https://decomp.me/scratch/UUKsu
(note that not only the mul things are wrong, but it also swapped operands at the very last add instruction)
Another one: https://decomp.me/scratch/ASmoM
exeHackJump:
1.9198622f=sead::Mathf::deg2rad(110.0f)- fix the stack issues by removing
frontDir2and re-usingfrontDirinstead
Decomp.me is down, so I can't check my solution, but here's what I have at the moment:
al::addVelocity(this, -verticalVec);
f32 fVar14 = frontDir.dot(verticalVec);
sead::Vector3f x;
x = (verticalVec - frontDir * fVar14) * 0.98f;
f32 fVar16 = frontDir.dot(hackerMoveVec2);
fVar16=sead::Mathf::clamp(fVar16,0.0f,1.0f);
x += frontDir * sead::Mathf::max(fVar14 * 0.92f + cosntB * fVar16, fVar14);
al::addVelocity(this, x);
bVar1 = false;
What's the current state on the last one, did you manage to improve that one further?
Got the second one improved by moving around random stuff (without looking at the original, so might also be a false improvement): https://decomp.me/scratch/P9h4T
The third one calculates the .length() twice - I'm not able to reproduce it yet, can you replace the visible/existing calculation with a normalize(f32 target_length) as we already theorized somewhere else?
I forgot to update the decompme link but it does use length twice https://decomp.me/scratch/hccNN normalize2(&trans,trans.length()/180.0f);
Improvement! https://decomp.me/scratch/7bYZR Just some stack stuff remaining. What's the current state on the other mismatches?
Not much I focused lately on other long lasting issues like KaronWing and ParameterBase. I updated the PR decompme links to point to the correct decompme
AccelStick has more appropriate names although I might replace the code with logically correct instead of higher matching % exeHackJump has improved considerably. Basically waiting for sead quat fixes to be merged so I can replace all operations and attempt to fix it again.
Alright, then let's put it in the freezer for now: Blocked by https://github.com/open-ead/sead/pull/221
exeHackResetPos is finally fixed. Same change fixed some stack issues in exeHackJump but flawed math is still pretty much the issue there I haven't figured out what is doing. No updates with accelStick
The code no longer contains random inlines from sead and math util.