[EXTERNAL-RIFFCC] Use Case For Codex - Streaming Platform Demo

[Zorlin]

tl;dr:

codex has discussed Riff.CC, an external project related to Riff Labs, and that it would be a great use case for Codex to demonstrate a viable use case.

We want to prepare a demo of Riff that could easily swap the IPFS component (Bunny.net CDN with tuning in front of standard Kubo IPFS) of the network for Codex with trivial work, that wings can outsource to the Riff team with no impact on DST timelines or priorities.

It will demonstrate a working streaming, distribution and curation platform, and Codex will be used to provide key parts of it.

I'm folding it in as something I've explicitly:

A) Scheduled over a weekend
B) Developed at no cost to Vac/DST
C) Confirmed would be a good demo and use case for Codex and gauged the Codex team's interests
D) Minimised the efforts needed to ship this (Riff staff will be working on it, not DST, just with guidance from DST, and this won't impact VacLab operation)

to help Codex's PR efforts
and to give them a great demo of their tech.

To do this task we essentially need to do the following:

• Add Codex support to Orbiter (Add Orbiter retrieval support for the Codex network (as persistence layer) riffcc/orbiter#44)
• Create scripting to pull artifacts from IPFS, transform them by reuploading them to Codex and getting a Codex-style CID
• Create a library based on Riff's library at the time, using the above script to save any real effort building it
• Test, demo, then ship a finished working and stable demo
• Deploy it properly with highly available Codex gateways behind it.

it's definitely not a 2 day project in itself, and the timing will slip and be readjusted over time, but my hope is that I can start laying the foundation for this and get it shipped around the time I feel Codex will be needing a really great demo.

As a fallback, we can write a paper on exactly how we could add Codex support to Riff.CC, indicate the desire for it, make a mention that IPFS is a neutral technology and is kinda becoming a standard like S3 did; and then simply use an IPFS based demo which can provably become Codex-ified later...

or perhaps easier, convince the Codex team that it's worthwhile to support IPFS CIDs including their format (or possibly Iroh style CIDs which are frankly far superior and we need to support for Lis anyways for the official Riff persistence library); which may actually be a good trade - the effort they may save us would probably make this trivial to ship since we can then just basically change the IPFS gateway we point at in order to add Codex support formally to Riff.

I will be pushing to get Riff production ready by November 5th to help give us time to ramp up to marketing it at DevConnect. It'll be a great demo of the potential of Codex and also great crosspromotion for Riff.

cc @jessiebroke and @dryajov