Non-chair hat on: I wanted to share that I have been implementing generalized transcoding following the patterns in this proposal, and I have positive feedback to report.

My implementation handles transcoding between legacy octet-sized untyped data and char32_t. The error handling extends well to the WHATWG set of encoders and works acceptably with iconv-based systems (though iconv interface adaptation remains a challenge).

You can find the implementation here: https://github.com/steve-downey/transcode

I have also started a very early draft paper regarding this work: https://github.com/steve-downey/transcode/blob/main/papers%2Ftranscode-view.md

This is not intended for the current agenda; I simply wanted to provide positive feedback on the patterns in this proposal, specifically regarding error handling and the problems of literals versus arrays for ranges not being much of a practical problem using the null adapter.