Initial support for C wrappers

[scott-wilson]

Description

This is the intial C wrapper/reimplementation of Imath (where it makes sense). The general idea for this project is to create a C and eventually a Rust wrapper. At the moment, this is a minimal commit to start a discussion about how best to implement this, maintaining the C bindings, etc.

The current goals

• Follow style standards set by Imath
• Follow C style standards set by the OpenTimelineIO and OpenImageIO C wrappers (where it makes sense)
• Make the C and the C++ types the same, byte for byte, and all functions produce the same result.
• Create a minimal set of wrappers. For example, if there's helper functions and methods, then don't implement them unless they make sense to be implemented.

Reference C wrappers

• OpenImageIO: Sketch of a C API for comment OpenImageIO#2748
• OpenTimelineIO: https://github.com/PixarAnimationStudios/OpenTimelineIO/tree/c-otio

Todo

• Sign the CLA (Shouldn't be any issues with this, but double checking with company)
• Finish implementing half
• Implement the rest of Imath

[linux-foundation-easycla]

The committers are authorized under a signed CLA.

• ✅ Scott Wilson (5753496)

[lgritz]

Can this not use the C++ version (casting on the way out or forcing a type pun)? It seems to me that we want to wrap the C++ functionality rather than repeat it. It's going to be much harder to maintain multiple copies of the same code.

[scott-wilson]

That's true. I want to try to balance speed with maintainability, so the plan was to reimplement some things where they were trivial. However, if we want to go the type punning route, then I can go for that.

[lgritz]

I would suggest that the test program be C, not C++. Otherwise we aren't really verifying that it compiles and runs properly in a pure C context.

[scott-wilson]

I would suggest that the test program be C, not C++. Otherwise we aren't really verifying that it compiles and runs properly in a pure C context.

That's a good point. I'll see what Anders did in oiio.

[cary-ilm]

Closing this PR, as the plans for Rust have progressed separately.