Zero-config peer-to-peer encrypted live folder syncing tool that respects your .gitignore
. Built on top of the Hypercore protocol with emphasis on being zero-config, secure, and decentralized.
Pre-packaged binary, no external dependencies required! You can find the binaries on the latest release here.
# Requires node >=v12.22.1
$ npm i -g portal-sync
# Start using portal
$ portal new
# or
$ portal join [sessionID]
portal
, no further content can be downloaded from it. No data is stored anywhere except on the host device.portal
server that all data is routed through. portal
only uses public servers to maintain a DHT (distributed hash table) for peer discovery.portal
tracks which files have changed to avoid resyncing entire folders wherever possible. A priority queue is used to optimize concurrent operations.portal
session ID can view the data you share.Portal relies on a publish-subscribe event model to drive its render and update cycles. File tree structure and individual file statuses are stored in a trie structure known as the Registry. On the host side, there is a local Registry that listens to file changes on the host machine and broadcasts them to an append-only Hypercore that is used as an event log. A drive syncing hook listens for changes in the local registry and streams file changes from disk to a Hyperdrive. On the client side, a remote Registry listens for changes in the event log and replicates changes locally. A drive download hook listens for changes in the remote registry and streams file changes from the Hyperdrive to the local disk.
Portals are identified by unique* 32-byte keys. When a client 'joins' a portal, portal
looks up the session key using Hyperswarm and establishes a connection to the host using UDP holepunching.
*8.63x10-78 chance of collision
Might seem similar to another similar project built on top of the Hypercore protocol called Dat but there are a few key differences.
portal
focuses purely on being one-to-many for file sharing/syncing.portal
is designed to be zero-config, it doesn't leave any dotfiles laying around, whereas Dat stores secrets and metadata in a ~/.dat
folder.portal
runs on the same underlying protocols, I haven't found a need to utilize version histories yet.portal
respects your .gitignore
so it doesn't sync anything you don't want (like pesky node_modules
)node >= v12.22.1
yarn
to install deps and yarn dev
to enable hot-reloadyarn link
to register portal
as a valid executable