quartznet/raylib-cs
Watch
2
0
Fork
You've already forked raylib-cs
 0
Code Wiki Activity

WASM examples (+ backports of new official examples) (#344)

Browse source 
* chg: New build system that uses the officially distributed binaries, bumped version to 8.0.0, simplified git workflow, removed deprecated OpenGL 1.1 functionality.

* chg: Modernize CI workflow, enable SourceLink

- Bump workflow actions to latest majors (Node 24); drop deprecated softprops/action-gh-release@v1
- Trigger push builds on main instead of master
- Create local nuget feed dir before pack (fixes NU1301)
- Enable Microsoft.SourceLink.GitHub for debugging symbols (ref PR #340)

* fix: centralized version data in Directory.build.props, and fixed various interop details that had incorrect function signatures

* chore: updated readme

* fix: version the native extract marker and chain download via DependsOnTargets

The .extracted marker now includes the raylib package name, so bumping
TargetRaylibTag re-extracts the new archive instead of silently keeping
(and packing/copying) the previous version's files.

_PrepareNativeLibrary and _StageWasmNative now depend directly on
_DownloadAndExtractInternal instead of CallTarget-ing it; dependency
targets run in the same project instance, so the resolved properties
(RaylibPackageName etc.) propagate naturally.

* fix: let the binding build for browser-wasm on both net8.0 and net10.0

The net8-era wasm workload (Microsoft.NET.Runtime.WebAssembly.Sdk 8.0.x,
auto-imported for RID browser-wasm) treats every browser-wasm project as
a wasm app: it forces OutputType=Exe after project evaluation (CS5001
for a classlib) and hooks its app-bundle build after Build, which errors
because a library has no assemblies to bundle. Opt Raylib-cs out via
DisableAutoWasmBuildApp (props time, before the workload defaults its
trigger) and pin OutputType back to Library in Directory.Build.targets
(evaluated after the workload props, so the assignment wins). net10's
wasm SDK needs neither workaround.

* chore: readme updated

* chg: simplifying build logic - a simple line in the documentation should save us the code here

* fix: Wrong signature of FrameBufferComplete

* chore: readme update

* feat: samples default to local project reference, and can optionally use the nuget package

* feat: backporting existing raylib-cs examples and new official raylib examples to WASM, adopting raylib's original code style

* chore: readme, gitignore, and targets backport.

* fix: Examples.csproj runs the download task when building locally

* feat: backporting existing raylib-cs examples and new official raylib examples to WASM, adopting raylib's original code style

* chore: readme, gitignore, and targets backport.

* chore: clean up linter warnings

* feat: html harness focuses the example and allows quick navigation with J/K instead.

* chore: readme mentions the property to use nuget vs. the local project reference

* feat: replaced the J/K navigation with good old HTML buttons

* chore: run dotnet format scoped default (was previously scoped to just 'style')
This commit is contained in:
tiger tiger tiger 2026-07-30 19:34:34 +02:00 • committed by GitHub
commit 8c22e68c2a
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
236 changed files with 40405 additions and 10896 deletions
Download patch file Download diff file Expand all files Collapse all files

30
Examples/resources/shaders/glsl100/eratosthenes.fs
View file

@ -1,6 +1,11 @@
#version 100
// value reaches scale*scale = 1,000,000, beyond mediump's guaranteed precision
#ifdef GL_FRAGMENT_PRECISION_HIGH
precision highp float;
#else
precision mediump float;
#endif
/*************************************************************************************
@ -9,13 +14,16 @@ precision mediump float;
The screen is divided into a square grid of boxes, each representing an integer value
Each integer is tested to see if it is a prime number. Primes are colored white
Non-primes are colored with a color that indicates the smallest factor which evenly divdes our integer
Non-primes are colored with a color that indicates the smallest factor which evenly divides our integer
You can change the scale variable to make a larger or smaller grid
Total number of integers displayed = scale squared, so scale = 100 tests the first 10,000 integers
WARNING: If you make scale too large, your GPU may bog down!
NOTE: GLSL ES 100 requires constant loop bounds, so the factor loop runs to the worst
case (scale) and breaks out at sqrt(value), matching the glsl330 version
***************************************************************************************/
// Input vertex attributes (from vertex shader)
@ -38,23 +46,25 @@ vec4 Colorizer(float counter, float maxSize)
void main()
{
vec4 color = vec4(1.0);
float scale = 1000.0; // Makes 100x100 square grid. Change this variable to make a smaller or larger grid
float scale = 1000.0; // Makes 100x100 square grid, change this variable to make a smaller or larger grid
float value = scale*floor(fragTexCoord.y*scale) + floor(fragTexCoord.x*scale); // Group pixels into boxes representing integer values
int valuei = int(value);
//if ((valuei == 0) || (valuei == 1) || (valuei == 2)) gl_FragColor = vec4(1.0);
//else
if (value <= 2.0) gl_FragColor = vec4(1.0);
else
{
//for (int i = 2; (i < int(max(2.0, sqrt(value) + 1.0))); i++)
// NOTE: On GLSL 100 for loops are restricted and loop condition must be a constant
// Tested on RPI, it seems loops are limited around 60 iteractions
for (int i = 2; i < 48; i++)
float maxFactor = max(2.0, sqrt(value) + 1.0);
for (int i = 2; i < 1001; i++) // Constant bound = scale (worst-case sqrt(value))
{
if (float(i) >= maxFactor) break;
if ((value - float(i)*floor(value/float(i))) <= 0.0)
{
gl_FragColor = Colorizer(float(i), scale);
color = Colorizer(float(i), scale);
//break; // Uncomment to color by the largest factor instead
}
}
gl_FragColor = color;
}
}