Fixes a panic in Metal backend when creating render pipelines with fragment shaders.

[tristanpoland]

Note

I am not much of a graphics dev, found this issue playing with my rendering engine project PLEASE DOUBLE CHECK ME ON THIS

Summary

Fixes a panic in Metal backend when creating render pipelines with fragment shaders.

Issue

The Metal pipeline creation panics with:

index out of bounds: the len is 1 but the index is 1
at blade-graphics/src/metal/pipeline.rs:324

This occurs when creating any render pipeline with a fragment shader on macOS.

Root Cause

When compiling shaders with per_entry_point_map in naga, the output info struct contains only the single requested entry point at index 0, regardless of its position in the original module.

The code was using ep_index (from the original module, where fragment shader is typically at index 1) to access info.entry_point_names, which only has 1 element at index 0.

• Vertex shader: ep_index=0 → info.entry_point_names[0] ✓
• Fragment shader: ep_index=1 → info.entry_point_names[1] ✗ panic

Solution

Use index 0 to access info.entry_point_names since naga's per-entry-point compilation always produces a single entry point at index 0. We still correctly use ep_index for accessing the original module's entry points.

Testing

Any render pipeline creation with a fragment shader now works correctly on macOS/Metal. Previously, all such pipelines would panic at initialization.