Proposal: Add Codex32 (BIP93) as a BIP85 Application

[BenWestgate]

This proposal has been submitted as a PR to bitcoin/bips
bitcoin/bips#1958
Bitcoin Mailing list and BIP93 author review feedback seems positive.
As BIP-0085 champion I would very much appreciate your review.

The BIP-0085.mediawiki diff is here:
bitcoin/bips@master...BenWestgate:bips:codex32

My bip85 reference implementation for the proposal is here:
ethankosakovsky/bip85@master...BenWestgate:bip85:master

[akarve]

@BenWestgate howdy and sorry this was closed without progress. i am down to review and discuss this and can look again this weekend. mostly need to understand BIP-93 (not familiar yet) but as a rule new apps to BIP-85 should not be a problem as long as they don't break existing apps.

so feel free to reopen this and #68 and i'll try to respond faster; been busy with day job.

[BenWestgate]

Hello @akarve! The proposal is alive and well. I moved to the Bitcoin Mailing list https://groups.google.com/g/bitcoindev/c/--lHTAtq0Qc and I chatted with the BIP93 authors yesterday. Feedback seems positive. A review from the BIP85 champion would be very appreciated.

I closed this as it became out of date as I ended up forking the older ethankosakovsky/bip85 reference implementation to add the new applications including codex32 to compute my test vectors.

[BenWestgate]

@akarve I opened a PR against bip-0085.mediawiki and it requires your sign-off.
bitcoin/bips#1958

[akarve]

sweetness, i commented there. hopefully the protocol (not merged yet so malleable) makes your PR easy:
bitcoin/bips#1958 (comment)

https://github.com/akarve/bipsea/blob/protocol/src/bipsea/app_protocol.py

[BenWestgate]

Yes, I've looked at the diff and it will make the Bipsea PR easier.

bitcoin/bips#1958 is currently waiting on bitcoin/bips#2040. We made the decisions of how codex32 will behave with different human-readable parts, I wrote test vectors covering the general HRP case, my implementation passes on them, I just need to finish edits to the text to reflect what we decided.

[akarve]

Copy. Lmk if I can help with anything. I see you are getting back to 2040 in Feb, so no worries.

[BenWestgate]

Yes. Since this BIP93 app proposed to do any HRP we want that behavior standardized.

On the other hand, some unproposed codex32 encodings like extended private master keys would have some non-random payload components in their serialization. In that case it new HRPs must define how they use the BIP85 entropy.

For example "ms" directly encodes 16-64 master seed bytes as the payload, "cl" encodes a 32-byte private key.

There is a chance reviewers may deprecate certain "ms" lengths or change the switch between regular/long checksum.

In the meantime, would you like me to add codex32 support to your xprv command?

So that a codex32-encoded master seed (or set of k shares) piped is decoded and its xprv printed?

The codex32 authors particularly like the use-case of using codex32 as the backup of the root key for BIP85.