Federating a domain name... how?

I’ve thought a lot about what we could do to avoid it and it’s non-trivial. It is a project in itself and outside of the scope of fedeproxy, IMHO. But maybe there are relatively simple temporary workarounds? In any case that’s a discussion that needs to happen. And I suspect that by creating bridges between forges we are likely to discover it’s not as hard as we think it is.

@misc do you mind if I create a new topic out of this message to keep exploring the options on this particular point of centralization?

Feel free to split, yup. I am a bit less optimistic than you, as I think the solution exist, but would requires to use something else than git. Current CS theories discussion point to CRDT or Operational Transform. After all, real time editing of a shared document can also be done in non real time, and the main property is to be able to merge stuff and reconcile modification, but that doesn’t mean you can’t keep a 2nd copy.

But the problem, from a quick reading, is that this would requires to have a semantic understanding of the document (eg, source code), which could mean using your DVCS on the AST which in turn become a fun problem for languages with macros and pre processors, and also a fun problem for everything dropped at the AST level (spaces, comment,etc). And since you store the AST, you need to convert it back to source code somehow.

1 Like

It requires visitors to download something. There is however a centralization point: whoever has the private key to update the content controls it.

I think the problem that needs solving could be expressed like this:

  • there are N people creating digital assets (code, web pages, discussions, etc.)
  • they all claim it is related to X (the name of a project)
  • there are M people looking for assets related to X

How can these M people find these assets? How can they not be confused by the fact that N people produced heterogeneous assets that all claim to be related to X although they sometime contradict each other?

You might raise this question to the DREAM forum or matrix chatroom. Part of dream specs include e.g. ERIS for content-addressing that avoid use of domain names.

1 Like

Finding a given blob of data in a content addressable storage system (such as a peer-to-peer network) is definitely possible. But how does someone name such a blob of data? How do people find the data given the name they heard or bookmarked somewhere?

I don’t know if ERIS is appropriate. Just mentioned as an example of the stuff DREAM is involved with.

Re:ERIS… @cj spent some time on coding intending a Golang impl. (see: https://github.com/cjslep/eris )

1 Like