I'm writing a go-wrapper for this implementation; BTW the C wrapper makes that really convenient, thanks for that!

I struggle to completely understand the API though. What I'm trying to do is quite basic: given n original data buffers or symbols, I want to generate n "parity symbols" s.t. that I can loose any n (of the now 2*n total) symbols and still recover the data.

I mainly experimented with the API in banchmarks.cpp: OK, sounds like I would need to call leo_encode with original_count == recovery_count for that, right? leo_encode_work_count returns 2*original_count for these params and calling leo_encode yields encode_work_data with the size 2*original_count as expected. So far so good. What is surprising though: I would have expected encode_work_data to contain the original data too instead it does seem to contain 2*original_count recovery or parity symbols. So is there no way I can achieve what described above using the public C API or am I using recovery_count wrong?

Thanks in advance :-)