Skip to content

Latest commit

 

History

History
81 lines (51 loc) · 3.3 KB

README.md

File metadata and controls

81 lines (51 loc) · 3.3 KB
Raw

ReVault

ReVault is a peer-to-peer self-hosted file synchronization project.

The main goal of the project is for me to spend time, fiddle with things, and explore a few random concepts.

The secondary goal of the project is to allow seemless, mostly-offline, and transparent update of directories and files whenever possible. Basically, provide something that could be a bit of a DropBox alternative where you don't actually have to set anything up with a third party to work, and that could work as well between all-local computers, or using a laptop and a VPS.

If you really are looking for production-ready software, you probably want syncthing.

Rationale

This is a side project done for fun and because I find it useful. It's non-trivial to get right, but I do not want to turn this into a business or a full-time job, and I am interested in getting collaboration from all kinds of interested parties without limiting who can use it too much (hence the LGPL 3.0 license).

Currently, there's little code written. I've had various sample apps where I tried to build the same thing and mostly focused on the synchronization mechanism but had nothing available. For this project, I decided to take an open approach over GitHub, using projects and boards to have an entirely open process to see how it goes.

Approach

Step 1:

  • Explore, figure things out. Until this is done, stability and safety are hard to guarantee, and performance is likely to be bad.

That's where we are right now.

Step 2:

  • no files should be lost or corrupted, and this project should be able to act as an example good Erlang application for the community
    • solid unit tests
    • comprehensive property-based tests
    • type checks
    • linting
    • commented code
  • Have the software possible to operate and understand before it reaches stability. I'd like to wire up OTel stuff in here.

Step 3:

  • An Emphasis on Clarity: things should be well-documented, and the software should aim to cause no undue user surprises, although we know surprises can't be avoided fully
  • While you may need somewhat technical knowledge to set things up, it should require no special knowledge to have the software running over your regular directories

In General:

  • Respectfully: there is a code of conduct which I (@ferd) will enforce strictly. I will obviously not kick myself out of my repository, but will expect call-outs when/if misbehaving, and will accept forks without a complaint on my part

Invariants to Maintain

  • Never modify a file that a user created other than by synchronization
  • a "dry run" mode cannot touch the tracked directory's filesystem [dry-run mode not existing yet]
  • security is critical (allow cert pinning) though we can't guarantee it yet
  • correctness over performance
  • be portable across Linux, OSX, and Windows (at various efficiency costs, also I don't frequently use Windows)
  • assumes a filesystem that supports UTF-8 file paths

Using

This isn't usable yet, so don't even try.

Roadmap

See the Project boards

Why the ReVault name?

It's a bad pun between a file vault, not wanting a paying service where a third party gets to hold copies of my files, and having no master (nodes).

Build

$ rebar3 as prod release

Tests

$ rebar3 check