:claw honing - First milestone

Huge first milestone for :claw honing is reached - I’ve run C++ library routines (PhysX) from Common Lisp without writing any C/C++ myself, without writing any bindings manually. :claw generated C adapter and CL bindings, I only built the former and called the latter.

This is an important proof of the concept:

• IFFI approach works
• C wrapper generation for C++ is possible with libresect (libclang) - no need to dive into llvm/clang development libraries any deeper, or so it seems so far.

99% work done, so only another 99% left.

Wait.. PhysX?

I’ve switched from Filament to PhysX wrapping, because the latter one requires much less work to get it actually running. Filament comes after.

What’s next

Next I’ll slowly start bringing :claw level of C++ wrapping support to that of C while adding any missing C++ features I’ll stumble upon. Then polishing and code cleanup - it’s got a bit messy in the process. Hopefully, I won’t be burnt out by then completely and will be able to add proper documentation, so anyone else would be able to enjoy :claw bugsfeatures.

I also decided to add system generation capability to :claw. At the moment, it works similar way cl-autowrap operates - you have primary macro with configuration that expands into bindings. For that to work, you need to load whole :claw bundle and that’s totally an overkill - you don’t need any parsing and autogeneration when only thing you want is to load pregenerated bindings. So :claw will have an option to generate an ASDF system for you to do just that - load bindings for the architecture of your choice.

Basically, you would use system with primary macro to generate bindings and an ASDF system to load them, then in your project that needs those bindings you would depend directly on this generated system rather than on :claw.

:claw is performance-oriented first and it also cares about image size of your application.