The Mergening #13

* Plus a CME workaround until I can figure out why it's happening

Fixes screen flashing when chunks are being built

* Make each tesselator use its own ByteBuffer * Don't make RecyclingList a static variable 🤦 * Add some additions via ITessellatorInstance

* Stop using builtin updateRenderers * Remove a few unused things in MixinTessellator * Remove useless casting due to implementing the interface

* Currently just proxies to the World... need to get it to use the local cache

* block, meta, and light. Still need TODO biome * Start removing unused items from BlockRenderView

* Remove last unsafe this.world accesses * Copy biomeData * Implement getTileEntity(), and getBiomeGenForCoords() * Remove unused 1.16+ classes - Biome, Colorizer, and ColorBlender related

* Use one instance on the main thread, and individual instances on every other thread * Prevent draw() and buffer.reset() on non singleton instance

…GLStateManager transformer

… for the distance at which entities will render. Hope it's not too bad mitch :)

* Remove some uses of blockstate + patch FluidState * Fix lerp * Update FluidRenderer.java * Enable Sodium fluid rendering * Update FluidRenderer.java * Temporarily disable Sodium fluid rendering

Matches vanilla

This reverts commit bbe1aba.

Pretty useless since the integrated server stalls instantly, but our renderer supports it

Partially fixes wrong biome colors, but at chunk edges it's still wrong because neighboring chunks aren't accessed

The old code seemed to work fine but this should be safer

LWJGL3's MemoryUtil#memRealloc does this, see its javadoc. Fixes screen flashing and this error getting printed when indirect buffer grows: GL_INVALID_OPERATION error generated. Bound draw indirect buffer is not large enough.

Fixes wrong colors at chunk edges

* Avoids errors when usin lwjglxdebug

* Add Mixin to disable MC's builtin checkGLError * Add configs for iris/sodium

* Add the commits where code was used from

* Update README.MD

* Move some StateNotifiers to GLStateManager * Remove some redundant mixins based on sodium/angelica work and rendering state captures * Add Iris DebugScreenHandler

* Ported Sodium's GUI * Correct AO * Fixed Vignette

* Adds some code from lwjgl3 in CompatMemoryUtil.java * Re-add *Unsafe classes

Reduces jar size by 11 mb

If you actually want to work on it, pass -Dsodium.enableFluidRenderer=true

* Correct useNoErrorGLContext option * Added translucency sorting option * Implemented Particle Culling gives real fps boost, however, it's not as big as i thought * Begin implementing animateOnlyVisibleTextures * Correct Donation page opening * Fixed Shift + P Now it uses NotFine's GUI, but it think that NotFine's GUI should be rewrited for using other config options * Added Shaders button * Correct URL's for Linux * second code cleanup in particle culling

Fixes CaffeineMC/sodium#740

Remove most of ArchaicFix * Move a few mixins over to angelica * Keep AF's Config classes for now Remove CompileStubs * Comment out anything using them (shadersmod) SplashProgress - Stop errors * Disable if org.lwjgl.util.Debug is set * Only enable Iris/lwjgl debug callbacks if org.lwjgl.util.Debug is set

…atching RenderBlocks

…mixins

(Can reintroduce them when Iris is wired up, if needed)

* More CompatMemoryUtil functions * Fix Sodium GLProgram to use ResourceLocation instead of Identifier * rm GUI options for now

* Get icon sanely * Update WorldSlice.java * Update SodiumConfig.java oops * Init and pass Sodium LightPipelines * Update FluidRenderer.java actually save the lpp * Remove registry lookup and refactor WorldUtil * Hook up lighting pipeline * Patch fluid velocity * Update WorldUtil.java oops * It's alive! Sodium Fluid Renderer is alive!

…ot file

…rch for Tessellator references

…search for class references

…sformer

* Initialize GLStateManager via Mixins instead of all in the static block * Exclude two known safe methods from the GLStateManager check (startGame & intializeTextures)

…ghbors sometimes

Backport of CaffeineMC/sodium#755

Partially reverts the previous chunk tracking commit, because that logic will reintroduce CaffeineMC/sodium#605 As long as we only rebuild with neighbors present, we shouldn't see the visual seams

…RedirectorTransformer

…ethod in GLStateManager

This will allow NotFine to use Sodium menu elements and vice versa in a future commit.

* Hacky workaround for glDepthFunc not being applied properly after splash * better glColor() support with helper function. (Still not complete)

oops

…y working in isolation

…name to handle obfuscation but not only

…o the same blend fonction in GLStateManager

…in thread. * Log instead of crash, for now

Done: - A batching vertex buffer builder for font output - Reduction in drawcalls from 1-3 per character to 1 per font atlas switch - Mixin to draw the entire F3 screen in one batch, instead of separate batches per line TODO: - Switch drawing to a VBO - Cache string meshes across frames - Cache string width calculations used for layouts - Multidraw to reduce draw calls even further

TODO: Check if this impacts performance

* Keep track of inGLNewList * Clear binding on delete texture * Push/Pop Attrib Hacks -- to be evolved later

Why it happened: - setWorldAndLoadRenderers has a hook at the beginning that initializes Sodium if it hasn't been already - loadRenderers has a hook at the end that always initializes Sodium - setWorldAndLoadRenderers calls loadRenderers at the end This caused a race condition sometimes because Sodium's thread scheduler is not prepared to be able to stop the instant after it has started. The issue was resolved by moving the setWorldAndLoadRenderers hook to the end of the method (the same place as it is in Sodium/Embeddium 1.16)

No reason to since it's a compile-time dep. Fixes ASMModParser yelling about module-info.class

* Add license * Register the events from the angelica clientproxy to avoid registration errors

…ead blows up

* Unclear if needed, but can remove it later if it's not

…ing anything visible.

* We'll need to replace, remove, or rewrite these * It's also likely silently hiding something that's preventing the Iris backport from rendering anything useful

…25)

* Why would someone run Angelica on a server?

…need immediate cleanup

Fixes Iris camera transform

* First shadow pass renders; however it moves all around when moving the camera... * Second shadow pass not running because it makes everything very very dark instead of just rendering translucents

…ss...

* Shaders sun location is still wrong, but the main sun image now renders correctly

…ders.xxx

* Return a block instead of a BlockMatch (at least for now) for populating mc_entity

* Potential hud-caching fix * Adding config for testing * adding comments * Removing redundant calls * More comments * Adding some error handling, I really don't know exactly what I should be doing for errors or how to know when to throw one

…wasn't the z level...

** Clicking on everything doesn't yet work, and the visuals are slightly off

* Scrolling by clicking isn't currently wired up

* Stuff is sort of working * A little more progress

Sodium pipeline is still very buggy, so isolate it to Iris-only for now

* Still needs testing on OSX/Windows

…d to stick around.

…default

Should make launches slightly faster in dev

Will be fixed upstream in GTNewHorizons/ExampleMod1.7.10#244

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

The Mergening #13

The Mergening #13

Commits on Dec 25, 2023