but without central points of failure. The aim is to create continuous distributed software development processes, without having to rely on complex or centralized services.
**You need only know the nostr ID or bech32 form of a repo you will be able to discover all versions, its origins and its history.**
The git object data is not designed to be stored in nostr, but rather, the signaling, update, version and commit data. Git data will live in existing git repositories, github, gitlab, gitea, gogs, ssh, file system etc.
**`tags`** MUST contain tag "c" which is equal to the hex-encoded git commit hash. Typically this will be a 40 character sha-1 hash, but git will soon be allowing sha-2 hashes too
MAY contain an entry identifying the previous commit, if one exists, in the form `["e", "<event_id>"]`. Further tags can be developed to capture informative git information such as major/minor versioning. TBD: it will be possible to hint at a recommended git URL to download the repos (e.g. from github), but it is not yet decided whether this would be a tag itself or an additional entry in an existing tag.
**`pubkey`** SHOULD be associated with a given repo. This could be communicated out of band, or, for example MAY be included in a file in the root directory of the repo, `nostr.json`
It will also be possible signal a recommended repo, just like it is possible to recommended a relay. In this way, a codebase could start on github, move to gitlab, then to an own hosted instance, and finally back to github. All, without any breaks in service. TODO: flesh out this section, perhaps include relay discovery, and bech32 encoding
nostr.json
==========
The design of this NIP is such that each repo will have a public key used to send updates to a relay. That key SHOULD appear in the top level directory of the repo in a file called `nostr.json`