Feature Request: (maybe) Use Raptorq for large payloads

[pmespresso]

Pulling this out into its own issue - #320 (comment)

https://github.com/cberner/raptorq

[pmespresso]

so luckily it seems Qualcomm has updated their License to allow the use of RaptorQ without claiming rights to royalties, etc. https://datatracker.ietf.org/ipr/1957/

Ah, yeah, it's Apache 2.0 https://app.fossa.io/projects/git%2Bgithub.com%2Fcberner%2Fraptorq/refs/branch/master/61875c4640d8ea4339b606fd078d644822e033e2/preview

[burdges]

As an aside, I believe the only "research" question here is: Should we simply nest a new erasure coding scheme inside the QR code's erasure coding scheme? Or should we do some new erasure coding scheme for the QR code itself so that partially received QR codes help reconstruct? The issue is that partially received also contain wrong data which also harms reconstruction. In other words, the question is: Is there actually anything better than fully correcting at the frame level before going on into the stream level?

We do need to answer this question for it to be worth implementing this protocol using any desired erasure coding scheme. We already use https://github.com/darrenldl/reed-solomon-erasure in the availabity scheme, btw.

[Slesarew]

Done

[burdges]

Awesome! :)