Sounds interesting, and promising! I'm no expert in streaming videos, but your shell reminded me of mosh [1]. Mosh's auth approach is also interesting - it leverages ssh for authentication, something that you might consider too. Mosh unfortunately has shortcomings though - mainly, it lacks tunneling ability.
Seems like we’re barking up the same tree:
https://github.com/proxylity/raptor
> Perhaps most importantly, it does so using a tiny fraction of the CPU time, saving energy and keeping our datacenters (and planet) a little cooler.
But then:
> A decoder backend on AWS (SQS + Lambda + DynamoDB + S3) reassembles objects from incoming encoded packets delivered via Proxylity UDP Gateway.
:( those microservice invocation will burn up the DC more.
The real sell looks like offloading s3 upload latency.
After having used it a while I tend to agree — the snappy send time is such a nice change.
Maybe at some point S3 will have a native implementation of something like this.
Sounds interesting, and promising! I'm no expert in streaming videos, but your shell reminded me of mosh [1]. Mosh's auth approach is also interesting - it leverages ssh for authentication, something that you might consider too. Mosh unfortunately has shortcomings though - mainly, it lacks tunneling ability.
[1] https://mosh.org
Really cool, but the thing I really appreciate is that you had this towards the top of the GH README: "Built with Codex using GPT-5.4 (Low)."
I am fine with using AI to code something, but I really like it when people are honest about it so that I know what I am looking at.
that's interesting, what are the use cases you have in mind with this protocol?
Just tried, works well