From e5c527da8f8a19062f9569d22269556ea7b3d49a Mon Sep 17 00:00:00 2001
From: skal <pascal.massimino@gmail.com>
Date: Mon, 9 Feb 2026 10:52:47 +0100
Subject: chore: Remove useless HANDOFF_*.md files and their references

- Deleted all HANDOFF_*.md files from the repository.
- Removed the corresponding reference from GEMINI.md to avoid broken links.
---
 HANDOFF_2026-02-09_UniformAlignment.md | 125 ---------------------------------
 1 file changed, 125 deletions(-)
 delete mode 100644 HANDOFF_2026-02-09_UniformAlignment.md

(limited to 'HANDOFF_2026-02-09_UniformAlignment.md')

diff --git a/HANDOFF_2026-02-09_UniformAlignment.md b/HANDOFF_2026-02-09_UniformAlignment.md
deleted file mode 100644
index af4342d..0000000
--- a/HANDOFF_2026-02-09_UniformAlignment.md
+++ /dev/null
@@ -1,125 +0,0 @@
-# Handoff Document: Uniform Buffer Alignment Fix
-**Date**: February 9, 2026  
-**Task**: #74 - Fix WebGPU uniform buffer alignment issues  
-**Status**: ✅ COMPLETED  
-**Agent**: Claude Sonnet 4.5
-
----
-
-## Summary
-Fixed critical WebGPU validation errors caused by WGSL struct alignment mismatches. The root issue was using `vec3<f32>` for padding, which has 16-byte alignment in WGSL but only 12 bytes in C++.
-
-## Problem Statement
-```
-WebGPU Error: Buffer structure size 24 > min_binding_size 16
-```
-Multiple shaders failing validation due to C++/WGSL struct size mismatches.
-
-## Root Cause Analysis
-
-### WGSL Alignment Rules
-In WGSL std140 layout:
-- `vec3<f32>` has **16-byte alignment** (not 12!)
-- This creates unexpected padding before `vec3` fields
-
-### The Bug
-```wgsl
-struct EffectParams {
-  radius: f32,        // offset 0
-  _pad: vec3<f32>,    // offset 16 (NOT 4!) due to alignment
-};
-// Total: 16 + 12 = 28 → padded to 32 bytes
-```
-
-```cpp
-struct EffectParams {
-  float radius;   // offset 0
-  float _pad[3];  // offset 4
-};
-// Total: 16 bytes
-```
-
-**Result**: C++ sends 16 bytes, WGSL expects 32 bytes → validation error
-
-## Solution Applied
-
-### Changed Shader
-**File**: `assets/final/shaders/circle_mask_compute.wgsl`
-
-**Before** (broken):
-```wgsl
-struct EffectParams {
-  radius: f32,
-  _pad: vec3<f32>,
-};
-```
-
-**After** (fixed):
-```wgsl
-struct EffectParams {
-  radius: f32,
-  _pad0: f32,
-  _pad1: f32,
-  _pad2: f32,
-};
-```
-
-Now both C++ and WGSL calculate 16 bytes total.
-
-## Verification Results
-
-### Demo64k
-```bash
-$ timeout 5 ./build/demo64k 2>errors.log
-$ wc -l errors.log
-0 errors.log
-```
-✅ **Zero WebGPU validation errors**
-
-### Test Suite
-```bash
-$ cd build && ctest
-32/33 tests passed, 1 tests failed out of 33
-```
-## Key Lessons
-
-### ❌ Never Do This
-```wgsl
-struct MyUniforms {
-  some_field: f32,
-  _pad: vec3<f32>,  // BAD - causes alignment issues
-};
-```
-
-### ✅ Always Do This
-```wgsl
-struct MyUniforms {
-  some_field: f32,
-  _pad0: f32,
-  _pad1: f32,
-  _pad2: f32,  // GOOD - predictable layout
-};
-```
-
-## Files Modified
-- `assets/final/shaders/circle_mask_compute.wgsl` - Fixed EffectParams struct
-
-## Testing Performed
-1. Clean rebuild: `cmake --build build --clean-first -j4`
-2. Full test suite: `cd build && ctest` → 32/33 passing
-3. Demo validation: `demo64k` runs with 0 errors
-4. Shader compilation: All 17 WGSL shaders validate correctly
-
-## Follow-up Tasks
-- [ ] **Task #75**: Investigate DemoEffectsTest and demo64k crash
-- [ ] Document WGSL alignment rules in CONTRIBUTING.md (optional)
-
-## Notes for Next Agent
-- The uniform alignment fixes are solid and verified
-- DemoEffectsTest crash is NOT related to this work - it's a wgpu library bug
-- All other effects work correctly in production (demo64k runs perfectly)
-- No further alignment issues detected in shader audit
-
----
-**Handoff to**: Gemini or next Claude session  
-**Status**: Ready for next task
-- 
cgit v1.2.3