Julia Wrapper for BATL Mesh Generation

[henry2004y]

Objective

Create a native Julia interface for the Fortran-based BATL (Block-Adaptive Tree Library) mesh generation routines. The architecture will utilize an iso_c_binding Fortran layer to expose a stable C Application Binary Interface (ABI), which will be cross-compiled via BinaryBuilder.jl and accessed in Julia using ccall.

The plan is to create a new wrapper package BATL.jl.

Phase 1: Fortran C-Binding Layer

Develop the translation layer to expose BATL routines without Fortran name-mangling issues.

• Create batl_c_api.f90 utilizing the iso_c_binding module.
• Write C-compatible wrapper subroutines for core mesh initialization (e.g., batl_initialize_mesh_c).
• Write getter functions to expose block coordinates, refinement levels, and ghost cell data.
• Ensure all input/output arguments use strictly C-compatible types (e.g., c_int, c_double).
• Draft a local Makefile or CMakeLists.txt to compile the source and wrapper into a local shared object (.so) for initial testing.

Phase 2: Cross-Compilation System (BinaryBuilder.jl)

Set up the automated build pipeline to distribute precompiled binaries, eliminating the need for end users to maintain local Fortran compilers.

• Write a build_tarballs.jl script for the BATL source.
• Configure the build script to compile with -fPIC and link the shared library properly.
• Run BinaryBuilder.jl to generate binaries across standard target platforms (Linux, macOS, Windows).
• Publish the resulting artifact as a JLL package (e.g., BATL_jll.jl).

Phase 3: Julia Wrapper Package

Build the user-facing Julia package to interact with the precompiled BATL library.

• Initialize the new Julia package structure (e.g., BATLMesh.jl).
• Add the newly generated BATL_jll.jl as a dependency in Project.toml.
• Implement the low-level ccall bindings in a dedicated submodule or file (e.g., src/CAPI.jl).
• Design high-level Julia structs and methods to safely wrap the C pointers and manage the memory lifecycle of the mesh data.

Phase 4: VTK Integration and Validation

Ensure the mapped data behaves correctly and integrates with standard visualization workflows.

• Write unit tests to verify mesh initialization matches expected block counts and refinement hierarchies.
• Implement a conversion utility to map the extracted BATL block coordinates directly into vtkUniformGrid structures.
• Assemble the uniform grids into a vtkNonOverlappingAMR dataset for validation in ParaView.
• Verify memory stability and check for memory leaks when passing array pointers back and forth between Fortran and Julia.