MLIR thunk magic, 0 GC, 0 latency, 0 allocations
So the honest claim is: you can drive AOT-compiled GPU kernels from Julia with AOT-compiled host marshalling, at zero allocation, through a C++ API, with no C shim and no C++ host program.
Only in Julia
# ImguiDemo
Dear ImGui driven entirely from Julia — window, GL context, widgets, and real
pixels — through the RepliBuild wrapper in `lib/`. No C++ host program.
```julia
julia --project=/path/to/RepliBuild.jl
julia> include("src/ImguiDemo.jl"); using .ImguiDemo
julia> ImguiDemo.main() # visible window, 120 frames
julia> ImguiDemo.main(visible = false) # hidden; returns the pixel count
```
Needs a display. Under Hyprland that means Xwayland:
`GLFW_PLATFORM=x11 DISPLAY=:1`.
## What it demonstrates
`lib/libimgui.so` is Dear ImGui **v1.92.9b** core plus the upstream GLFW
platform backend and OpenGL3 renderer backend, compiled from source and wrapped
from DWARF. Every imgui call below is a Tier-2 MLIR AOT thunk; the GLFW and GL
show original
