Ethereum Registry

[cathalgarvey]

Let's pick this back up: Ethereum is now the second-most-significant blockchain and gaining on Bitcoin, and it's got good client and library support. It wouldn't take a very complex contract to create a gittorrent registry, at all.

So, let's stop waiting for Bitcoin and just use Ethereum? :)

[cathalgarvey]

Proposal:

• A canonical Ethereum registry contract, perhaps implemented as a "contract pointer" to enable "upgrades", which all Gittorrent users use as a reference for mapping usernames to infohashes.
• The contract should probably map SSH keys and alternative sources for the hosted repositories, for permanence's sake.
• The contract should record a link to the "bugtracker", wherever that is, for a repository, as well as the "homepage".
• The contract could also contain a way for forks to announce themselves, and to announce other engagements.
• The contract should not be constructed so as to facilitate censorship of the code by force, although it might be constructed to allow a "curator group" to vote on removal of certain repositories according to a clear code. In this case, the curators should live in several countries to provide jurisdictional arbitrage, and the group should be mutable.

The contract could be ownerless, for censorship-free-ness, or multi-owner for resilience-with-backup, or it could be owned by @cjb outright for ease (but that may put him at risk). Even in the case of "censorship" the blockchain will store prior state.

Thoughts welcome; just shelling out concepts here that might make Gittorrent happen. :)

[cjb]

Cool, sounds good to me. We could add it as one of several sources of name resolution (e.g. DNS is another). By the way, are you familiar with Mango?

https://medium.com/@alexberegszaszi/mango-git-completely-decentralised-7aef8bcbcfe6#.66kxuibma

[cathalgarvey]

Oh, Mango looks very interesting. Thanks for the pointer! I'll read up on that a bit more before commenting further.

And yes, offering pluggable name resolution is good. :)