* C++ lighting engine
Runs better using my new system for storing and iterating through lights.
I removed the lighting ctx profiler because after I've realized its very inaccurate and was probably programmed incorrectly from the start. Although I remember it working fine before and not constantly showing 20 ms or higher when I'm in an extremely simple level with extremely simple lights and the FPS is way too high for it to be taking 20 ms.
* Whoops
* Make some optimizations
* Fix small bug with LE inside graphics vertex function
I also brought back the lighting profiler because it's better than nothing.
* Some optimizations
* Cache active lights and 4 lights per vertex limit
Less branching and less iterating with the limit
* Peachy changes
## `void geo_skip_interpolation(struct GraphNode *node, struct GraphNodeObject *obj)`
Skips graph node interpolation for a frame
feed it a graph node and its corresponding object
## `void obj_skip_interpolation(struct Object *o)`
Skips object interpolation for a frame
takes care of only the object transform, otherwise you need to use
## `void obj_anim_skip_interpolation(struct Object *o)`
Skips animation interpolation for a frame
useful for choppy animation
------------
## `u32 get_time_stop_flags(void)`
Gets the current time stop state
* Fix wall double bonk glitch
[03/02/2026 12:10] colbyrayz.z64
i might have found an insane consistent way getting this somehow
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468202050981593244/2026-02-03_03-08-24.mp4?ex=6983d1f6&is=69828076&hm=6d9e564d8b422cdb7e53097f910ff122bff97c12d661222b929085b9b6b206cf&
[03/02/2026 12:12] blocky.cmd
what.
[03/02/2026 12:14] colbyrayz.z64
its semi consistent
[03/02/2026 12:14] colbyrayz.z64
it happens everytime
[03/02/2026 12:14] colbyrayz.z64
has multiple causes it seems but this is one of them
[03/02/2026 19:08] peachypeachsm64
thanks
(unrelated, but you need to exclude `MARIO_ANIM_PART_` when scanning for animation names in your debug display)
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468307238165676305/image.png?ex=69838b2d&is=698239ad&hm=084431682e34741dbc1a1b919a38e6cd37c9592bdeec9b1424410e6d9c122334&
{Reactions}
👀👍
[03/02/2026 19:15] altiami
I'm noticing the angles are perfectly aligned with the axis. It looks LuigiGamer's and my theory about reflection being buggy and about problematic angles could be relevant
[03/02/2026 19:40] wbmarioo
When did double wall jumps start happening
[03/02/2026 19:40] wbmarioo
I’ve only just seen people talking about it since yesterday
[03/02/2026 19:47] theincredibleholc
Many ages ago. Perhaps Millennia... Eons, even.
[03/02/2026 20:34] blocky.cmd
ever since coop was made
[03/02/2026 20:34] blocky.cmd
i'm pretty sure
[03/02/2026 20:49] theincredibleholc
Makes me wonder if it's in Sm64EX
[03/02/2026 20:50] theincredibleholc
-# (not EX-Coop)
[03/02/2026 20:51] maniscat2
makes me wonder if its in sm64
[03/02/2026 21:05] birdekek
makes me wonder if its in smb
[03/02/2026 21:20] altiami
soup
[03/02/2026 21:22] theincredibleholc
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468340870515261441/hXy2f1U.gif?ex=6983aa7f&is=698258ff&hm=a0e0625d6b27ea42ea2b4db9eb970bb93871862506752582ff16c48c316107b3&
[03/02/2026 21:24] emeraldlockdown
Oh, I did like a bunch of research into this bug and completely forgot to send it here. Seeing how this has evolved, I'll give a summarized copy that isn't me ranting about other things for 50 paragraphs
Double bonks, if I recall correctly, have been in the game ever since romhacks were added. I don't recall them existing beforehand, but perhaps I'm wrong. A commit that lines up with this memory, and changes many things related to wallkicks, is [this commit](3e46cc1161). The tree for that commit so you can look around other commits is [here](https://github.com/coop-deluxe/sm64coopdx/commits/main/?before=6092488d1c4fc741b16a0789ef9c08ec0279333f+2651). This commit changes how wall normals work, so perhaps it's worth taking a look at. While I would love to look at it, I'm not really that good at math, so I couldn't help all that much, although I'll try to skim through it to find anything interesting.
{Reactions}
👀 (2)
[03/02/2026 21:46] altiami
something I'm noticing is that in `perform_air_quarter_step`, there's an iterator over walls that uses the wall collision data instead of the wall assigned to the mario state, but only if the fix collision bugs flag is set, which isn't by default
[03/02/2026 21:46] altiami
otherwise it uses the last wall, which should be the same as the wall assigned to the mario state in the new `mario_update_wall` function
[03/02/2026 21:49] altiami
the only way that flag can be set is through mods or with a special djui menu that's only there if the game is compiled with `DEVELOPMENT` defined
[03/02/2026 21:58] altiami
wait, I just realized in this video that Mario immediately turns around when hitting the wall, before the wall kick happens (shown by angle). That's not supposed to happen right?
[03/02/2026 22:19] theincredibleholc
Not sure. I did another video at 240 fps and used frame advance to look frame-by-frame.
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468355148718932241/2026-02-03_15-17-04.mp4?ex=6983b7cc&is=6982664c&hm=1b2ff76684239890abe996631612eb05a8bc5649875b7784bb13ce692a7f8aa6&
[03/02/2026 22:21] altiami
are you able to add more info to that display on the side?
[03/02/2026 22:21] theincredibleholc
Sure, what do you want?
[03/02/2026 22:22] altiami
let me make sure I'm gonna say the right thing
[03/02/2026 22:22] mysterd
Whoa I have no memory writing that lol. That may have been done before fixCollision stuff was a toggle, if so all of that should probably only happen if fixCollisions is enabled and vanilla behavior should be preserved otherwise.
[03/02/2026 22:23] altiami
I can see fixCollisionBugs in the level values struct
[03/02/2026 22:23] mysterd
oh yeah it is there
[03/02/2026 22:23] altiami
and there are some if conditions for it
[03/02/2026 22:23] mysterd
hmm
[03/02/2026 22:23] altiami
either way
[03/02/2026 22:23] altiami
holc, your frogger level doesn't use that value right?
[03/02/2026 22:23] altiami
and surely the sm64 levels don't
[03/02/2026 22:23] theincredibleholc
I use fixColBugs in Frogger.
[03/02/2026 22:24] theincredibleholc
But this recording is vanilla Sm64 with just a readout. No other mods.
[03/02/2026 22:24] altiami
okay, two things I cna say to add
[03/02/2026 22:24] altiami
first, add the flag for fix collision bugs
[03/02/2026 22:24] altiami
it'll be a sanity check
[03/02/2026 22:25] altiami
second, from the `MarioState`, add `wall`, and its `normal` values
[03/02/2026 22:25] blocky.cmd
you do NOT need this many frames 🙏 240 is insane
[03/02/2026 22:25] altiami
and also add, from the `MarioState`, `wallNormal`, which is a vec3f
[03/02/2026 22:26] altiami
the idea here is to see if there are any desyncs (redundancy desyncs, not network)
[03/02/2026 22:27] altiami
I think that's everything I can see being possibly relevant in that commit
[03/02/2026 22:28] altiami
also, that clip shows an angle that isn't on a 45 degree multiple, so my theory is out
[03/02/2026 22:29] blocky.cmd
i want to mention because it might be important
[03/02/2026 22:29] blocky.cmd
but if you get hit by an enemy (i.e a goomba) the bug gets fixed
[03/02/2026 22:29] blocky.cmd
and its a common fix to ask a player to punch you to fix the double bonk
[03/02/2026 22:30] altiami
tf
[03/02/2026 22:30] altiami
that feels like it'd be entirely unrelated. I can't tell if I hope that's a placebo effect or not
[03/02/2026 22:31] blocky.cmd
it would be an awful coincidence
[03/02/2026 22:31] altiami
on the one hand it'd be absurd if it was. On the other, it'd provide extra context that'd help in tracking it down
[03/02/2026 22:32] altiami
in my experience though it tends to just happen without any consistency
[03/02/2026 22:34] theincredibleholc
I think it has something to do with warping/resetting
{Reactions}
✅
[03/02/2026 22:34] blocky.cmd
i wouldnt be surprised if any of the knockback actions did soemthing
[03/02/2026 22:35] theincredibleholc
So, as @ColbyRayz! pointed it you can get it to happen consistently by doing exactly what he did. Enter BOB and reset the server, then you can double bonk in CASTLE_GROUNDS. What I noticed is that after re-entering the castle, then going back to castle grounds, I can't double bonk.
[03/02/2026 22:35] altiami
https://tenor.com/bvZwe.gif
[03/02/2026 22:36] theincredibleholc
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468359516793082072/2026-02-03_15-36-01.mp4?ex=6983bbdd&is=69826a5d&hm=e15aeb9d18419cf29c67d5034ebf9deb56ed820ce5b99fe5dd5fa222718e2bd3&
[03/02/2026 22:37] altiami
and fix collision bugs is indeed 0
[03/02/2026 22:37] theincredibleholc
I turned it on for some tests too, and it made no difference.
[03/02/2026 22:38] theincredibleholc
proof
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468360133435461806/2026-02-03_15-38-29.mp4?ex=6983bc70&is=69826af0&hm=3df29070d9707fafd82d4a620a08ac88227be7e7168b33eb9f4e4fcedd1a768a&
[03/02/2026 22:40] altiami
I'm especially interested in the `MarioState.wallNormal` field now, because that's used in `mario_bonk_reflection`
{Reactions}
👀
[03/02/2026 22:40] altiami
and that's a field that was newly added in that commit
[03/02/2026 22:42] altiami
so the fields of interest are
`MarioState.wall` (pointer)
`MarioState.wall->normal.x`, `.y`, and `.z`
`MarioState.wallNormal[0]` to `[2]`
[03/02/2026 22:43] altiami
actually, maybe not the wall pointer yet
[03/02/2026 22:43] altiami
nothing to compare it to
[03/02/2026 22:43] altiami
unless we want to manually do a wall collision detection so we get that data lol
[03/02/2026 22:43] theincredibleholc
I'm not going to pretend I know enough to contribute more to this conversation, but you need me to test anything let me know!
[03/02/2026 22:43] theincredibleholc
https://tenor.com/view/ralph-wiggum-simpsons-eating-glue-gif-20921387
[03/02/2026 22:45] theincredibleholc
I'm wondering what specifically would change when Mario enters BOB (or maybe his falling level-entry Action) that would "store" the bug.
[03/02/2026 22:45] altiami
if you can show those two different normals in that sidebar with this test, that'd be great, because if they're not the same, then we have a big problem
[03/02/2026 22:45] theincredibleholc
**m.wall.normal** and **m.wallNormal** ?
[03/02/2026 22:45] blocky.cmd
.y
[03/02/2026 22:46] altiami
yeah, and keep in mind that `m.wall.normal` uses .x, .y, and .z, while `wallNormal` is an array
[03/02/2026 22:46] altiami
woah wtf
[03/02/2026 22:46] altiami
hold on
[03/02/2026 22:46] altiami
okay
[03/02/2026 22:46] blocky.cmd
i thought arrays automatically were converted
[03/02/2026 22:46] altiami
are they?
[03/02/2026 22:46] blocky.cmd
i think so yeah
[03/02/2026 22:46] altiami
mmk
[03/02/2026 22:47] blocky.cmd
i may be high af rn tho
[03/02/2026 22:47] altiami
can we get much higher
[03/02/2026 22:47] blocky.cmd
i wish
[03/02/2026 22:59] theincredibleholc
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468365251509489807/2026-02-03_15-58-34.mp4?ex=6983c134&is=69826fb4&hm=4385e7d104f6de68962aaeb866c3f7eef2e4caf7201b5bbcabe065cdf34e503c&
[03/02/2026 23:00] altiami
ok, and then `wallNormal`
[03/02/2026 23:01] altiami
interestingly though we can see that the normal does not change after hitting the wall, despite being turned around
[03/02/2026 23:02] altiami
so you're still technically wall kicking off of that wall
[03/02/2026 23:08] theincredibleholc
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468367655273500753/image.png?ex=6983c371&is=698271f1&hm=3092c87bf5044791c8da9d6959f28f3acd6cab7fd35f25f70d3161becfda781e&
[03/02/2026 23:08] theincredibleholc
lol
[03/02/2026 23:08] altiami
that would mean only the bonk would be relevant, and not the kick, I think
[03/02/2026 23:09] altiami
hrm
[03/02/2026 23:09] xluigigamerx
wall.normal
[03/02/2026 23:09] xluigigamerx
.x/y/z
[03/02/2026 23:09] altiami
no, there should be a `wallNormal` property too
[03/02/2026 23:09] altiami
unless it was removed
[03/02/2026 23:09] xluigigamerx
There isn't, there never was
[03/02/2026 23:09] theincredibleholc
It autofills for me, but that might be for something else.
[03/02/2026 23:09] xluigigamerx
Normals are vec3fs
[03/02/2026 23:10] altiami
3e46cc1161/include/types.h (L390)
[03/02/2026 23:10] altiami
and yeah it's still there
<https://github.com/coop-deluxe/sm64coopdx/blob/main/include/types.h#L571>
[03/02/2026 23:11] xluigigamerx
Well is it exposed
[03/02/2026 23:11] altiami
that's a good question
[03/02/2026 23:11] altiami
though I'm also seeing that error says "Surface"
[03/02/2026 23:11] xluigigamerx
It's a vec3f
[03/02/2026 23:11] peachypeachsm64
yes it is
m.wallNormal
[03/02/2026 23:11] altiami
you're trying to get the `wallNormal` of the `MarioState` right
[03/02/2026 23:12] xluigigamerx
Are you indexing something other than .x/y/z
[03/02/2026 23:12] xluigigamerx
Vec3fs are converted to .x/y/z in lua altiami
[03/02/2026 23:12] altiami
blocky mentioned that already
[03/02/2026 23:12] theincredibleholc
guys
[03/02/2026 23:12] theincredibleholc
I did m.wall.wallNormal
[03/02/2026 23:12] theincredibleholc
:>
[03/02/2026 23:12] altiami
OOPS
[03/02/2026 23:12] xluigigamerx
Wallnormal is a vec3f
[03/02/2026 23:12] theincredibleholc
Thanks peachy
[03/02/2026 23:13] xluigigamerx
GET OU-
[03/02/2026 23:13] theincredibleholc
https://tenor.com/view/wojak-loading-dumb-soyjak-brainlet-gif-5905960021355918533
[03/02/2026 23:13] blocky.cmd
add
y
[03/02/2026 23:13] blocky.cmd
add .y
[03/02/2026 23:13] altiami
yes luigi I know, I'm looking right at the source code. I don't need it repeated to me three times
[03/02/2026 23:14] xluigigamerx
Ok, but may I interest you in... vec3fs
[03/02/2026 23:14] xluigigamerx
They are very peak, you can store position and get it without confusion
[03/02/2026 23:14] blocky.cmd
ok im home i might try myself at this too
[03/02/2026 23:14] altiami
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468369124223881297/reimuAAAAAAAAAAAAAAAAA.webp?ex=6983c4d0&is=69827350&hm=33117149457125ac19973c8b284af34edc719f27fbf9ab54cd3abd55ec5097ee&
[03/02/2026 23:15] altiami
anyway yes, blocky's right, since it's a vec3f it'll have the .x .y and .z properties
[03/02/2026 23:15] altiami
and luigi
[03/02/2026 23:15] xluigigamerx
Hello world!
[03/02/2026 23:16] altiami
hello world too bland, gotta make it with "from the earth I rise"
[03/02/2026 23:16] xluigigamerx
But on which axis
[03/02/2026 23:16] xluigigamerx
Are you a y up or a z up type of person
[03/02/2026 23:16] altiami
I forget, do the axes change depending on which type of space you're using in SM64, or is that only relevant to rotation order
[03/02/2026 23:17] xluigigamerx
Speaking of Ys and Zs, may I interest you in some vec3f action
[03/02/2026 23:17] xluigigamerx
You can reach infinite distance!
[03/02/2026 23:17] altiami
hello
I am in the NaN realm
please help me
[03/02/2026 23:18] wall_e20
oh you too?
[03/02/2026 23:18] xluigigamerx
The f in vec3f stands for a float, which becomes infinite when reaching the max float depending on the compiler I think
[03/02/2026 23:18] blocky.cmd
ok i checked
[03/02/2026 23:18] altiami
actually do we have any safeguards against that for mario's position since people love their BLJs
[03/02/2026 23:18] blocky.cmd
both match
[03/02/2026 23:18] xluigigamerx
Why
[03/02/2026 23:18] xluigigamerx
PUs are cinema
[03/02/2026 23:18] altiami
at least in the original sm64, it'd crash if you got to infinity
[03/02/2026 23:19] xluigigamerx
Well, I did go to infinity, the level just turns black, I accidentally entered the void
[03/02/2026 23:19] theincredibleholc
I have to step away for a bit and grind out some work for my actual job. Ping me if I can be of assistance and I'll try to check on it.
[03/02/2026 23:19] altiami
ok if it doesn't crash then I guess it's fine
[03/02/2026 23:19] xluigigamerx
You can still exit
[03/02/2026 23:19] blocky.cmd
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468370376210776310/Super_Mario_64_Coop_Deluxe_v1.4.1_2026-02-03_23-18-28.mp4?ex=6983c5fa&is=6982747a&hm=20de6bb2a795d32522389421c20d59e3260f49c3741613d00053f8d6bcc0b315&
[03/02/2026 23:19] blocky.cmd
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468370462722490541/image.png?ex=6983c60f&is=6982748f&hm=15efd49675b065a6135c7f4617d35188bae96c6f16a2920a43a7163ff0efaa95&
[03/02/2026 23:20] altiami
what about X and Z
[03/02/2026 23:20] xluigigamerx
How inaccurate is Mario's position in the farther PUs, since floats lose accuracy as the get bigger
[03/02/2026 23:20] theincredibleholc
I already posted those
[03/02/2026 23:20] theincredibleholc
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468370615865053409/image.png?ex=6983c633&is=698274b3&hm=3162419073f9706a70903260f3891556e9ddeff1f654121853d087d7196f5ec3&
[03/02/2026 23:20] xluigigamerx
0 when wall, 1 when floor, -1 when ceiling bruh
[03/02/2026 23:21] altiami
that's just for `wall.normal`. We still need to make sure `wallNormal` matches fully
[03/02/2026 23:21] xluigigamerx
Well check the source code and see where each is being set
[03/02/2026 23:22] altiami
the suspect parts are in `mario_bonk_reflection` and `perform_air_quarter_step`
[03/02/2026 23:23] xluigigamerx
My theory is that mario bonk reflection just runs twice for whatever reason
[03/02/2026 23:24] blocky.cmd
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468371530671984771/Super_Mario_64_Coop_Deluxe_v1.4.1_2026-02-03_23-24-02.mp4?ex=6983c70d&is=6982758d&hm=53a3e8fdd1c652e245271962fce1e101889bbe5d6736606771c3c254505f2140&
[03/02/2026 23:24] xluigigamerx
But it could be false, because I know that diving into the wall causes the bug to occur as well
[03/02/2026 23:24] blocky.cmd
okay if you want my opinion it looks like wall interactions are doubled in general
[03/02/2026 23:24] blocky.cmd
you can see when i dive into the wall i get like twice as much negative speed i should have
[03/02/2026 23:24] xluigigamerx
Yeah
[03/02/2026 23:24] blocky.cmd
i don't know if bonk actions hard set your velocity but i assume it does
[03/02/2026 23:24] peachypeachsm64
funny thing happened
I put a breakpoint on mario_bonk_reflection
before entering bob, bonking into a wall breaks once, in `common_air_action_step`
after doing bob + server reset, bonking breaks twice, in `common_air_action_step` AND `common_air_knockback_step`
after entering a level again, it "fixes" itself and only breaks once
[03/02/2026 23:24] xluigigamerx
I can confirm this happens
[03/02/2026 23:25] altiami
okay, so they do match
[03/02/2026 23:25] altiami
and peachy pretty much confirmed it runs twice on top of the empirical evidence of the increased speed
[03/02/2026 23:26] altiami
I noticed that increased speed before but I wasn't sure what to make of it
[03/02/2026 23:26] altiami
so now comes the part where we answer why it runs twice
[03/02/2026 23:26] xluigigamerx
I think whatever function is setting the wall kb speed is the one at fault
[03/02/2026 23:27] altiami
it's likely up the call stack, yeah
[03/02/2026 23:27] xluigigamerx
Which one sets the velocity after air hit wall
[03/02/2026 23:27] xluigigamerx
Check instances of air hit wall steps
[03/02/2026 23:28] altiami
`mario_bonk_reflection`can set the forward velocity but only if the `negateSpeed` argument is set
[03/02/2026 23:28] altiami
...wait
[03/02/2026 23:29] altiami
what?
[03/02/2026 23:29] altiami
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468372741777395945/image.png?ex=6983c82e&is=698276ae&hm=deca9bc76b84ac422764fd8bdd19feaddfcd983a5002a8672d480578c43bbb00&
[03/02/2026 23:29] altiami
is there ever a case where mario bonks and *does* turn around?
[03/02/2026 23:29] blocky.cmd
@Altiami
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468372840167374848/Super_Mario_64_Coop_Deluxe_v1.4.1_2026-02-03_23-28-42_online-video-cutter.com.mp4?ex=6983c846&is=698276c6&hm=8c2212526b95bc4142864cc00a7ded77896bdb65768bb5fa0adc8a7989cbbd1a&
[03/02/2026 23:29] blocky.cmd
i wasn't crazy
[03/02/2026 23:29] emeraldlockdown
This was also apart of my investigation. Since we're getting into act air hit wall, there's a quirk with this function. I logically thought through it, and I don't think it should be an issue, but it's something to note
`act_air_hit_wall` has an interesting quirk, it doesn’t actually return a value. I’ll let this code comment explain:
```c
//! Missing return statement. The returned value is the result of the call
// to set_character_animation (set_mario_animation in original decomp).
// In practice, this value is nonzero.
// This results in this action "cancelling" into itself. It is supposed to
// execute on two frames, but instead it executes twice on the same frame.
// This results in firsties only being possible for a single frame, instead
// of two.
```
[03/02/2026 23:30] blocky.cmd
i think it has to do with how mario is initialized at the beggining of a level
[03/02/2026 23:30] blocky.cmd
one of the knockback actions puts him back in a good state it seems
[03/02/2026 23:30] peachypeachsm64
.
[03/02/2026 23:31] altiami
dive bonk turns mario around?
[03/02/2026 23:31] altiami
period
[03/02/2026 23:31] blocky.cmd
💅
[03/02/2026 23:32] peachypeachsm64
well no
actually
he does turn around
but there is another += 0x8000 after that
so it does nothing lol
[03/02/2026 23:32] altiami
tf
[03/02/2026 23:32] altiami
mario 64 glue
[03/02/2026 23:32] peachypeachsm64
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468373577928540325/image.png?ex=6983c8f6&is=69827776&hm=507f3aec5f79c19b9ede42a8da95d2acdb7970c79ee8c21b8e86341bc816bcd3&
[03/02/2026 23:32] emeraldlockdown
It's so hard to analyze this code because it's literally running off of hopes and dreams
[03/02/2026 23:32] altiami
https://tenor.com/bukEv.gif
[03/02/2026 23:32] blocky.cmd
..hopes and dreams you say?
[03/02/2026 23:32] altiami
okay so
[03/02/2026 23:33] cooliokid956
hi
{Reactions}
🔥
[03/02/2026 23:33] altiami
we know that mario is being turned around for the wall hit action
[03/02/2026 23:33] altiami
...by what?
[03/02/2026 23:33] xluigigamerx
I was literally at that line lmao
[03/02/2026 23:33] blocky.cmd
it might just be a side effect of the walls interaction being doubled in general
[03/02/2026 23:33] xluigigamerx
Test it out, turn on bouncy level bounds
[03/02/2026 23:33] blocky.cmd
so he turns around twice
[03/02/2026 23:33] altiami
he's not supposed to turn around though
[03/02/2026 23:33] peachypeachsm64
omg
it does return a value
BUT
it returns the value of `set_character_animation(m, CHAR_ANIM_START_WALLKICK)`
WHICH CAN BE 0
[03/02/2026 23:34] peachypeachsm64
why would you use the animation current frame as a way to cancel out an action
wtf
[03/02/2026 23:35] altiami
and even if he bonked the same wall twice, that would mean he reflects twice off the same normal, so he'd end up being in the initial direction again
[03/02/2026 23:35] xluigigamerx
Firsties?
[03/02/2026 23:35] emeraldlockdown
Why would it sometimes be 0 though? That's sort of where my lead dried up, because this happens seemingly randomly, or due to an init issue. It just returns the animFrame, so why would that change in the first place?
[03/02/2026 23:35] altiami
it'd be 0 if it fails right?
[03/02/2026 23:35] xluigigamerx
AnimFrame starts at 0
[03/02/2026 23:36] peachypeachsm64
is that how firsties work?
no way
[03/02/2026 23:36] xluigigamerx
No
[03/02/2026 23:36] emeraldlockdown
Perhaps I need to rephrase it, the issue is why would the animation change when no mods are being ran? Why would how it set change? Maybe we should try returning 1 on the function and try to recreate the issue to see if it's the problem
[03/02/2026 23:37] emeraldlockdown
What inputs are done to recreate the issue? It looks like you access the pause menu in the level selector
[03/02/2026 23:37] altiami
doesn't it change the animation when you bonk though
[03/02/2026 23:37] xluigigamerx
The action change resets the animation to frame 0
[03/02/2026 23:37] blocky.cmd
firsties are immune to double bonks
[03/02/2026 23:37] emeraldlockdown
That's good to know
[03/02/2026 23:38] xluigigamerx
I know
[03/02/2026 23:38] xluigigamerx
It makes sense because firsties are firsties
[03/02/2026 23:38] blocky.cmd
ah yes
[03/02/2026 23:38] xluigigamerx
They're done on exactly the same frame as hitting the wall
[03/02/2026 23:38] blocky.cmd
i'm looking into why getting hit would reset the bug rn
[03/02/2026 23:38] xluigigamerx
Otherwise it's not a firstie
[03/02/2026 23:38] blocky.cmd
maybe we could trace it backwards 🤷
[03/02/2026 23:39] altiami
so we know it only happens if mario reaches the knnockback action
[03/02/2026 23:39] altiami
so, is something in the knockback action making mario turn around when it shouldn't?
[03/02/2026 23:39] emeraldlockdown
If I understand it correctly, it only happens if the second if statement is ran in `act_air_hit_wall`, which, sets the action to `ACT_BACKWARD_AIR_KB`
[03/02/2026 23:40] peachypeachsm64
and it happens when you enter bob and restart the server (???)
[03/02/2026 23:40] altiami
it has to be related to something stale
[03/02/2026 23:40] altiami
or maybe not being initialized correctly in the first place?
[03/02/2026 23:40] emeraldlockdown
It can still be something in `act_air_hit_wall`, since the double turn around happens when you hit the wall
[03/02/2026 23:41] altiami
peachy, what does the call stack look like for each break when it doubles up?
[03/02/2026 23:42] altiami
it's just occurred to me we sound like an investigation firm trying to track down a serial killer
[03/02/2026 23:42] altiami
not entirely incorrect; I'm sure this bug has killed many players
[03/02/2026 23:42] blocky.cmd
only if leaving right after the act select
[03/02/2026 23:43] blocky.cmd
doesn't work on levels without the act select
[03/02/2026 23:43] peachypeachsm64
perform_air_step runs only once per frame even when the double bonk occurs, I can't explain the double speed (yet)
[03/02/2026 23:44] altiami
try looking at the call stack when the breaks happen. That should let us track down the execution flow
[03/02/2026 23:44] emeraldlockdown
The weird thing to me is why does it only happen in `ACT_BACKWARD_AIR_KB`, and not `ACT_SOFT_BONK`? Both of these actions call `common_air_knockback_step`, and both call `check_wall_kick`. I'm going to be very interested in what the issue ends up actually being
[03/02/2026 23:45] xluigigamerx
Soft bonk doesn't have the star particles
[03/02/2026 23:45] xluigigamerx
Which seems to be a shared trait with all affected wall actions
[03/02/2026 23:45] xluigigamerx
Impact particles
[03/02/2026 23:47] xluigigamerx
It must be whatever triggers the particles
[03/02/2026 23:48] peachypeachsm64
nothing worth
1st is during long jump
2nd is during air knockback action (when double bonk happens only)
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468377650996183295/image.png?ex=6983ccc1&is=69827b41&hm=95e9bbcafdb75a76a05f52692fc0fe68081f04bf27e54cd212fc9e9587f6fb78&https://cdn.discordapp.com/attachments/1264640346957021195/1468377651419943084/image.png?ex=6983ccc1&is=69827b41&hm=33a2207d5c14b0fe4068427156f1c7f410fd9451b0a389e1cfbcd08f928807de&
[03/02/2026 23:49] emeraldlockdown
I remember cross checking the 2nd line with sm64ex, incase anyone is curious, I believe they are the exact same
[03/02/2026 23:49] xluigigamerx
What about decomp
[03/02/2026 23:49] emeraldlockdown
I have not, I can check right now though
[03/02/2026 23:52] emeraldlockdown
Yea they're the same, so it's something outside of it
[03/02/2026 23:52] xluigigamerx
Can you compare the airborne files entirely
[03/02/2026 23:52] emeraldlockdown
Sure why not
[03/02/2026 23:53] altiami
I just zeroed in on that part of the code myself for the knockback action
[03/02/2026 23:54] peachypeachsm64
after further investigation
there is no double step
Mario just retains his speed, but it gets reverted (dive speed 48 becomes -48 for example)
[03/02/2026 23:55] emeraldlockdown
Here's the diff for the decomp vs coop mario_actions_airborne.c file. I'll scan through this as well, but you can look aswell if you'd like
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468379324104315132/diff.patch?ex=6983ce50&is=69827cd0&hm=b7403326958873d63f17458cbdb9fb8289f3203c0260dec05076c444db97f4bd&
[04/02/2026 00:00] blocky.cmd
something in take_damage_from_interact_object fixes the bug
[04/02/2026 00:00] altiami
wait
[04/02/2026 00:00] altiami
would that involve the knockbackTimer
[04/02/2026 00:00] blocky.cmd
```lua
function take_damage_from_interact_object(m)
if (not m or not m.interactObj) then return 0 end
local shake
local damage = m.interactObj.oDamageOrCoinValue
if (damage >= 4) then
shake = SHAKE_LARGE_DAMAGE
elseif (damage >= 2) then
shake = SHAKE_MED_DAMAGE
else
shake = SHAKE_SMALL_DAMAGE
end
if ((m.flags & MARIO_CAP_ON_HEAD) == 0) then
damage = damage + (damage + 1) / 2
end
if (m.flags & MARIO_METAL_CAP) ~= 0 then
damage = 0
end
-- disable player-to-player damage if the server says so
if (m.interactObj and (m.interactObj.oInteractType & INTERACT_PLAYER) ~= 0 ) then
if (gServerSettings.playerInteractions ~= PLAYER_INTERACTIONS_PVP) then
damage = 0
end
end
m.hurtCounter = m.hurtCounter + 4 * damage
queue_rumble_data_mario(m, 5, 80)
if (m.playerIndex == 0) then set_camera_shake_from_hit(shake) end
return damage
end
local function mario_update(m)
if m.playerIndex ~= 0 then return end
if m.controller.buttonPressed & X_BUTTON ~= 0 then
local test = obj_get_first_with_behavior_id(id_bhvDoorWarp)
test.oDamageOrCoinValue = 1
m.interactObj = test
take_damage_from_interact_object(m)
end
end
hook_event(HOOK_MARIO_UPDATE, mario_update)```
[04/02/2026 00:00] blocky.cmd
i made this and the bug was consistently fixed
[04/02/2026 00:01] blocky.cmd
had to translate the function since it wasn't exposed
[04/02/2026 00:02] altiami
there's that interact player thing again...
[04/02/2026 00:02] blocky.cmd
i thought so too but i couldn't find anything
[04/02/2026 00:02] blocky.cmd
relevant
[04/02/2026 00:02] blocky.cmd
i'll try to disect this function further if notyhing is found
[04/02/2026 00:04] emeraldlockdown
Does it work if you don't set the hurt counter?
[04/02/2026 00:06] altiami
that would be CRAZY
[04/02/2026 00:06] blocky.cmd
yes it does
[04/02/2026 00:06] blocky.cmd
i checked if it was the camera shake JUST IN CASE
[04/02/2026 00:07] blocky.cmd
(it wasn't)
[04/02/2026 00:07] blocky.cmd
i'm just going line by line rn
[04/02/2026 00:07] emeraldlockdown
The only 2 things it can be then is something with interactObj or, my favorite theory, `queue_rumble_data_mario`.
[04/02/2026 00:07] altiami
rumble data is a bit weird iirc
[04/02/2026 00:07] blocky.cmd
tried ``queue_rumble_data_mario`` on its own it didn't do anything
[04/02/2026 00:07] blocky.cmd
at least not when isolated
[04/02/2026 00:07] peachypeachsm64
I FOUND IT
{Reactions}
👀 (3) 🔥 (2) mariopog beware_the_polygonal ✅🍕🍞 🇬 🇴 🅰️ 🇹 eyesshaking
[04/02/2026 00:07] emeraldlockdown
BIG
[04/02/2026 00:08] altiami
tell us
[04/02/2026 00:08] blocky.cmd
TELL
[04/02/2026 00:08] blocky.cmd
🐟
[04/02/2026 00:09] peachypeachsm64
THIS
you know that every Mario has oInteractType INTERACT_PLAYER
and when you enter a level, you "interact" with the inactive players (maybe for 1 frame idk)
resetting the server doesn't clear the interactObj, so it's always a remote Mario, unless you interact with something else, for example, a door
SO
this weird ass pvp stuff is the cause of the double bonk
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468383008741654658/image.png?ex=6983d1be&is=6982803e&hm=612eddaac7e7c7855619c9ead1db86b6d15e3b09b16d59a0b871ac07e6cb128a&
[04/02/2026 00:10] altiami
this only strengthens what I offhandedly mentioned the other day about how we need to properly handle unloading network players
[04/02/2026 00:10] blocky.cmd
IN YOUR FACE
[04/02/2026 00:11] blocky.cmd
STUPID
[04/02/2026 00:11] xluigigamerx
Bro, you can activate it without prior pvp, how is it pvp related
[04/02/2026 00:11] altiami
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468383477488681241/DAAAMN.mp4?ex=6983d22e&is=698280ae&hm=496c9ff4ea5e6aa145accc5e3bf543c9b39a0e8d011d5ed2b3720c4260bb3487&
[04/02/2026 00:11] peachypeachsm64
read bro
[04/02/2026 00:11] emeraldlockdown
Thank you peachy for your incredible service to finding this bug that's plagued this game for centuries. Now I'm curious how long it's been around. What file is that and what line?
{Reactions}
⬆️ (2)
[04/02/2026 00:11] xluigigamerx
I can't im doing something for college
[04/02/2026 00:11] altiami
what's really funny is
[04/02/2026 00:12] altiami
I saw that condition
[04/02/2026 00:12] altiami
and was like
[04/02/2026 00:12] altiami
hmmmmmmmmmmmmmmmmmmmmmmmmmm
[04/02/2026 00:12] xluigigamerx
Read it
[04/02/2026 00:12] peachypeachsm64
`mario_action_airborne.c`
`common_air_knockback_step`
line 1240
[04/02/2026 00:12] emeraldlockdown
Thanks
[04/02/2026 00:12] xluigigamerx
(Past tense)
[04/02/2026 00:12] altiami
yup, that's the one
[04/02/2026 00:12] xluigigamerx
Who added that line
[04/02/2026 00:12] emeraldlockdown
Yup, 3 years ago
[04/02/2026 00:13] altiami
but, something's still weird. Doesn't that only set mario's forward velocity when that check passes?
[04/02/2026 00:13] blocky.cmd
2023 :faceholdingbacktears:
[04/02/2026 00:13] emeraldlockdown
Ah it may have been even earlier, let me check real quick
[04/02/2026 00:13] xluigigamerx
I want to know who's walls I need to live in
[04/02/2026 00:14] altiami
or is the problem that it's *not* passing the condition
[04/02/2026 00:14] peachypeachsm64
that's the problem
if the fvel is not set to -16, you continue bonking into the wall
or, for the dive, get sent backwards violently
[04/02/2026 00:14] altiami
ah, that's where I was mistaken
[04/02/2026 00:15] altiami
I thought the speed has been set already
[04/02/2026 00:15] altiami
yeah I see, it's not here in the hit wall action
[04/02/2026 00:16] altiami
alright gang, let's lift the mask and see whose kneecaps need to be gently tapped
[04/02/2026 00:16] altiami
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468384692775555277/image.png?ex=6983d350&is=698281d0&hm=0806f2c53e9b8c4649e2bf21833e160928f6dfbd696d7add1e0269441a644c2f&
[04/02/2026 00:16] blocky.cmd
FIXED?
[04/02/2026 00:16] blocky.cmd
# FIXED??
[04/02/2026 00:16] altiami
<bf435d5aa9>
[04/02/2026 00:17] altiami
it clearly had good intentions
[04/02/2026 00:17] altiami
but dormant network players are scuffed
[04/02/2026 00:17] blocky.cmd
aw it was when too much knockback setting got nerfed :(
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468384968592855165/image.png?ex=6983d391&is=69828211&hm=64f8675f995e4fabc9145eaca84d8069e3b7f9aa628c99d7a84eb08b1d0938f2&
[04/02/2026 00:17] blocky.cmd
i remember that
[04/02/2026 00:17] emeraldlockdown
No I don't know if that one is it, since, if I understand it correctly, ` if (m->interactObj == NULL || !(m->interactObj->oInteractType & INTERACT_PLAYER)) {` caused the issue, and that existed beforehand
[04/02/2026 00:18] altiami
the git blame tracks it back to this commit
[04/02/2026 00:18] altiami
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468385134880362551/image.png?ex=6983d3b9&is=69828239&hm=fe08108418b3326b8a5952f8002fcb5d057515f199c55a5c126813406e684788&
[04/02/2026 00:18] emeraldlockdown
51940d6a82
[04/02/2026 00:19] emeraldlockdown
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468385345015124245/Screenshot_2026-02-03_at_5.19.04_PM.png?ex=6983d3eb&is=6982826b&hm=c834fdd510a80f007c61b2ababf197312930b85c4b2a4b402b4e6cc72f7a4215&
[04/02/2026 00:19] altiami
oh wait I see what you mean
[04/02/2026 00:19] altiami
it was already in there in the diff
[04/02/2026 00:19] emeraldlockdown
Dang my memory is vividly off
[04/02/2026 00:19] emeraldlockdown
2020 is crazy
[04/02/2026 00:19] blocky.cmd
yeah
[04/02/2026 00:19] blocky.cmd
i did remember it existing since the beggining
[04/02/2026 00:20] theincredibleholc
https://tenor.com/view/renuu-thanos-rest-peaceful-gif-20831938
[04/02/2026 00:20] theincredibleholc
It's all ogre now 🥹
[04/02/2026 00:21] altiami
yeah emerald you found it
[04/02/2026 00:21] peachypeachsm64
well, some update
it's not when entering a level
it's specifically when restarting the server
it somehow interacts with a random mario object (13?)
(object pointer | is mario obj | playerIndex)
{Attachments}
https://cdn.discordapp.com/attachments/1264640346957021195/1468385977859969185/image.png?ex=6983d482&is=69828302&hm=59f49bd2f11d4a48eaa781966d24cf45affadf285983114e52a34ab6db79bd4e&
[04/02/2026 00:22] theincredibleholc
Huge shoutout to @ColbyRayz! for finding a way to replicate it. That was a seriously obscure find!
[04/02/2026 00:22] peachypeachsm64
wait
[04/02/2026 00:22] peachypeachsm64
no
[04/02/2026 00:22] emeraldlockdown
Starting the server, or closing the server?
[04/02/2026 00:22] peachypeachsm64
it doesn't interact
it uses whatever was in that object slot
which happens to be a mario
[04/02/2026 00:22] peachypeachsm64
stale pointer lmao
[04/02/2026 00:22] altiami
so I was right in a super roundabout way
[04/02/2026 00:22] altiami
goddamn
[04/02/2026 00:23] altiami
I didn't expect stale object pointer though
[04/02/2026 00:23] peachypeachsm64
tldr
mario was never reset properly
[04/02/2026 00:23] blocky.cmd
i'm close to doing a pr for a bunch of star road fixes does a romhack port repo still exist
[04/02/2026 00:23] blocky.cmd
coop-romhacks-blahblah
[04/02/2026 00:23] altiami
<RTD sound effect>
Mario HAS A TUMOR!
[04/02/2026 00:24] altiami
okay well
[04/02/2026 00:24] altiami
that was fun
[04/02/2026 00:24] emeraldlockdown
Indeed
[04/02/2026 00:24] altiami
roleplaying as bloodhounds
[04/02/2026 00:24] peachypeachsm64
https://github.com/coop-deluxe/rom-hacks
[04/02/2026 00:25] altiami
https://tenor.com/view/bear-sniffing-spartan370-big-blue-house-gif-22010873
[04/02/2026 00:25] altiami
this us finding the bug
[04/02/2026 00:25] blocky.cmd
thanks!
[04/02/2026 00:25] emeraldlockdown
The only other bug I can think of that has been in the community for waay too long is the doors disappearing bug, but something tells me that won't be as easy to fix. But seeing how colby somehow found a way to reproduce this bug, I'll hold out hope
[04/02/2026 00:25] peachypeachsm64
we found it
now, let's find a way to fix it
[04/02/2026 00:26] altiami
oh I thought you had an idea now
[04/02/2026 00:26] altiami
since you found it was a stale pointer
[04/02/2026 00:26] peachypeachsm64
reset interactObj
but
where
[04/02/2026 00:26] peachypeachsm64
and when
[04/02/2026 00:26] emeraldlockdown
Can't you just reset it on `network_shutdown`, or am I thinking in super simple ways
[04/02/2026 00:26] altiami
level load imo
[04/02/2026 00:26] altiami
when the mario is initialized anyway
[04/02/2026 00:26] altiami
at least, I hope he is
[04/02/2026 00:26] emeraldlockdown
Oh yea
[04/02/2026 00:26] altiami
he better be
[04/02/2026 00:27] altiami
or I'm throwing hands with miyamoto himself
[04/02/2026 00:27] emeraldlockdown
Wait can't you just reset it in `init_single_mario`?
[04/02/2026 00:27] sharen462
thats what i was gonna suggest
[04/02/2026 00:27] sharen462
that function doesnt reset ot
[04/02/2026 00:27] sharen462
you could just add it
[04/02/2026 00:27] altiami
really, *all* marios should be initialized on level load if they aren't already
[04/02/2026 00:27] emeraldlockdown
```c
m->heldObj = NULL;
m->heldByObj = NULL;
m->riddenObj = NULL;
m->usedObj = NULL;
m->bubbleObj = NULL;
```
They were so close!
[04/02/2026 00:27] altiami
LMAO
[04/02/2026 00:27] sharen462
it even resets other obj fields like ``riddenObj`` and stuff
[04/02/2026 00:28] sharen462
yeah lmfao
[04/02/2026 00:35] peachypeachsm64
yeah that works
now all I have left to do is making a comically long commit description for a one-line change
[04/02/2026 00:36] altiami
just screenshot this entire discord conversation, put it in a google doc, and attach it to the commit message
* Fix all double bonk related bugs
Here's some explanation about the infamous wall double bonk glitch.
First, why does it happen seemingly randomly?
When the server resets or when entering a level, Mario is fully initialized
again, but, for some reason, interactObj is not reset to NULL.
Because of that, it keeps the pointer of the last object Mario interacted
with, which becomes a stale object pointer (the object it points to no
longer exists).
Because objects are stored in a static buffer rather than being
allocated dynamically on the heap, this pointer always points to valid
memory (it cannot crash the game), but could potentially point to a
random Mario object depending on various conditions I won't list here.
Why interactObj pointing to a Mario object has anything to do with
wall bonks?!?
Because PVP.
sm64coopdx adds PVP. And the way it adds it is kinda questionable.
But that's a topic for another day.
To handle player vs player interactions, all Mario objects get the
unique INTERACT_PLAYER interaction type.
For the sake of code reusability, all PVP interactions use regular
interactions functions, with a Mario's interactObj temporarily set
to another Mario.
When the interaction is done, the interactObj is properly reset to NULL.
Meaning in regular gameplay, Mario's interactObj is NEVER another
Mario object.
Unless stale pointer as mentioned above.
What's important here is the interaction type INTERACT_PLAYER.
Because PVP is about punching and kicking other people's face with
force, it would be a shame if someone threw one of their best blows just
to see their opponent slightly move backwards like nothing happened.
It had to be impactful. For that, knockbacks had to be changed.
In SM64, ground knockbacks have capped speed and air knockbacks set the
speed to a predefined value. Not really usable for in the context of PVP.
So, for the sake of code reusability again, these limits had to be
temporarily ignored during PVP knockbacks.
Now the question is... how do you recognize a PVP knockback?
...
What about checking m->interactObj->oInteractType?!?!?
Because of this stupid check that does nothing most of the time because
Mario's interactObj is NEVER another Mario object, BUT suddenly treats
all knockbacks as PVP attacks because of a stale pointer, the following
happens:
- Mario bonks into a wall, entering the backwards air knockback action
- Due to the check above, Mario's speed is not set properly, keeping
him moving forward instead of backwards
- Mario bonks into the wall again, but he's no longer facing the wall
(Firsties are not affected by the bug, because they allow Mario to wall
kick before he enters the backwards knockback.)
Just remove the check and for proper measure, reset interactObj during
Mario init and that's it, right?
...
NO.
Now, why does it happen AGAIN?
There is still an obscure bug that probably nobody ever encountered
before, but exists.
During a 10-frame window after exiting a PVP knockback, the next
knockbacks are treated as PVP knockbacks.
To prevent players from being knockbacked again and again, a
knockbackTimer is set and starts counting down as soon as Mario exits
any knockback action.
But!
When not zero, this timer also makes the knockback actions ignore
their corresponding speed cap.
Not one, but TWO broken checks, that can't even properly identify a PVP
knockback, causing one of the most infamous glitch of the game.
Solution?
First, use a specific actionArg to flag a knockback action as a PVP
attack.
Second, ignore the knockback speed caps only if this flag is set.
Third, because some mods implement their own PVP knockbacks in a bad
way (not their fault, just tried to mimic what was already in place),
set the default knockback speed during set_mario_action if the PVP attack
flag is not set.
With this, no more double bonk, no more janky knockbacks.
That's all, folks!
* I'M A CHUCKSTER
Hue, saturation, brightness, contrast, exposure, dithering, posterization
Global = usually an object in the world
Non global = usually a screen element
Also:
* Changed skyboxes to use environment color instead of vertex color, visually identical
* Changed hud head texture to be non global
* changed power meter to be non global
* changed sparkles to be global
* changed some BBH vertex types to global
OpenGL renderer supported only for now.
* Allow multiple returns in autogen functions
* condense function member names
* This shouldn't be here
* Relocate these functions
LVT_FIELD shouldn't be there yet
* multi-return part 2: most functions are usable now
---------
Co-authored-by: PeachyPeachSM64 <72323920+PeachyPeachSM64@users.noreply.github.com>
- `djui_hud_render_texture(_tile)` now accepts properly any `TextureInfo`
- Removed `bitSize`, added `format` and `size` to `TextureInfo`
- `texture_to_lua_table` can now convert any `Texture` into an RGBA table
This command sets the RGB values of a material's environment color without setting the alpha. This is important, as functions like `geo_update_layer_transparency()` use the environment color alpha to control a model's overall opacity. However, if one of the model's materials sets its own environment color alpha, then that will take priority. This command allows these models to still use environment colors in their materials.
* djui_hud_get_mouse_locked()
Adds a total of 1 function:
# `bool djui_hud_get_mouse_locked()`
Returns the DJUI mouse lock status
Tip: This won't return the true state of mouse lock, so make sure to use `get_first_person_enabled()` and `camera_config_is_mouse_look_enabled()` as well
---
This also fills in some descriptions left out by Isaac in his viewport/scissor functions
* djui_hud_is_mouse_locked()
Same as before but is
* this is better imo
* zip + json properties; check existing file in create file
* smlua_audio_utils_replace_sequence
* audio_stream_load, audio_sample_load, smlua_model_util_get_id
* get_texture_info + can also load PNG files
* smlua_collision_util_get
* add wildcard in properties files + set text mode
* filepath restrictions
* Some mod_storage improvements
- Cache mod storage files into a map to reduce file I/O
- Fix a bug in mod_storage_save
- Add mod_storage_load_all that returns all keys/values as a table
* shutdown; fix buffer overflow; fix warnings; lua table
* reject binary files starting with MZ or ELF
* function members
* better doc
* adding file rewind
* ModFS guide; replace yaml by ini; read string buffer changes
Fix `textureArray` type to be recognized by autogen.
Painting textures can be replaced with the following code:
```
local tex1 = get_texture_info("texture_bottom")
local tex2 = get_texture_info("texture_top")
gPaintingValues.bob_painting.textureArray[1] = tex1.texture
gPaintingValues.bob_painting.textureArray[2] = tex2.texture
```
Texture format must be RGBA 16-bit for built-in textures or
RGBA 32-bit for mod loaded textures.
* De-sort fields + Add `protected`
This fixes the atrocious ordering in autocomplete (a,b,g,r 😭)
also adds the protected scope for read-only fields
_ReadOnlyTable revision
* Restored OOB death (when alone)
also bubbles the player
* Rerun autogen
* Peachy Review
* kill goto
* Curly Brace
added these new djui hud tools for Lua:
- djui_hud_set_viewport
- djui_hud_reset_viewport
- djui_hud_set_scissor
- djui_hud_reset_scissor
- djui_hud_render_line - this code could already be done exactly. this is just to make rendering lines more efficient.
Also made nametags use a viewport for scaling in the credits.
Shared graph nodes were incorrectly interpolated, only the first object with a shared graph node was interpolated properly.
Thanks to @Cooliokid956 for noticing that most of node types were **never** interpolated.
- Use a double hashmap to store interpolated data for each graph node and object. All translations, rotations and scales are now interpolated correctly.
- Add `GraphNodeScaleXYZ` type to scale to all 3 dimensions; Add `GEO_SCALE_XYZ` command.
gMtxTbl was also reaching it's limit, so dynamic allocation was added.
* wip
* get rid of seg2_dialog_table and fix crashes
* remove unused include
* change more types to s32
* remove dialog tables from eu_translation.h
* fix dialog_table_get signature
* Change `oToadMessageDialogId` to s32
* remove exit from dialog_table_add
* warning
* calloc allocated dialogs
* avoid memcpy on allocation fail
* Move dialog entry text field init to dialog_table_init
* Free original dialog entries' text field
* Put all reset logic in one place
* Run autogen
* Remove ``get_dialog_text_ascii`` from autogen
* minor fix
* free dialog->str if replaced
* nevermind
* remove get_dialog_unmodified from autogen
* Change -1 to DIALOG_NONE
* update stuff to s32
* use growing array functions
* Change limit to something more reasonable
* add null check
* remove explicit cast
* Minor fixes
* more type fix
* peachy review
* Fix weird enums in autogen
* remove extra newline :p
* add dialog restore
* bump max dialogs just a lil bit 🥺
* add lua behavior params
* all OBJECT fields are now Lua variables
* Revert lua behavior params changes
* isaac review
* fix segfault
---------
Co-authored-by: PeachyPeachSM64 <72323920+PeachyPeachSM64@users.noreply.github.com>
Lights are also zero indexed now which I could change to being one indexed but I think this is better anyway.
I restored ambient light objects setting it every frame since it broke a mod I was working on.
Sorry if any of these changes break anything you're working on dj.
Kaze would be proud though I bet
