Drasyl

Added

- Each node now belongs to a certain network ID. Connections between nodes will occur only if both.
peers use the same network ID. The main network has ID 1 (see configuration `drasyl.network.id`).
- Added drasyl groups plugin for membership management.
- Added support for netty's Epoll channels and event loops.
- Plugins can now access the node's identity.

Changed

- All `drasyl.marshalling` config properties will now distinct between allowed outgoing and ingoing
types. Config must be updated according to our documentation.
- `drasyl-core` now only depends on slf4j and no longer on logback. As part of this, the methods.
`DrasylNode.getLogLevel()` and `DrasylNode.setLogLevel()` have been removed and the configuration.
`drasyl.loglevel` has been removed.
- Nodes now only accept children joins if they are configured as a
super-peer (`drasyl.super-peer.enabled = false`).
- Removed grandchildren nodes. The hierarchy is now limited in depth.
- All-new plugin interface.
- Changed docker base image from `drasyl/drasyl-build-images:jre-11-curl-7.64.0`
to `openjdk:11-jre-slim` and removed default HEALTHCHECK.
- Enhanced JavaDoc.
- Changed `TypeValidator` to distinguish between in- and outbound messages.
- EmbeddedPipeline was generalized to simplify handler testing.
- Node now checks if the permissions of the identity file are too open (POSIX file systems only).
- Changed `Pipeline.processInbound` to accept `RelayableMessage`.
- Endpoints defined in `drasyl.super-peer.endpoints` must now always include the super peer's public
key.

Fixed

- Fixed memory-leaks of certain immutable classes.
- Fixed `DefaultPipeline.addBefore` method.
- Fixed error in IntraVM discovery which has delivered events around the pipeline directly to the
application.
- Several minor bug fixes.

Added

- Plugins can now be defined in the configuration `drasyl.plugins`. Plugins can be used to change
the behavior of the node when interacting with the network.
- `DrasylNode.send()` can now send arbitrary objects (make sure that a suitable serialization codec
is defined in the configuration `drasyl.marshalling`).
- `DrasylNode.identity()` returns own identity.
- Node will now try to expose itself via UPnP-IGD/NAT-PMP/PCP (can be disabled in the configuration
via `drasyl.expose.enabled = false`).
- Local Host Discovery: Node can use the file system to discover other drasyl nodes running on the
local computer (`drasyl.local-host-discovery`).
- CLI: `wormhole` sub-command added.

Changed

- `wss://production.env.drasyl.org025fff6f625f5dee816d9f8fe43895479aecfda187cb6a3330894a07e698bc5bd8`
is now the default super peer (`drasyl.super-peer.endpoints`).
- Configuration `drasyl.super-peer.public-key` removed (public key can now be specified in
configuration `drasyl.super-peer.endpoints`).
- `DrasylNode.send()` now returns a future, which complements when the message has been processed.
- `DrasylNode.send()` no longer throws exceptions, but writes them to the future.
- `DrasylNode.send()` is no longer synchronized.
- The own identity is now created in the `DrasylNode` constructor and not at node start.
- The `DrasylNode` constructor now throws an error if the own identity is invalid.
- Number of threads required by drasyl reduced by internal switch to `DrasylScheduler`.
- Chat example and CLI now stops the JVM after shutting down the node.
- Javadoc improved.
- Changed to a non-blocking logger.
- Chunking is now disabled by default in the configuration, because it is not fully implemented
yet (`drasyl.message.max-content-length`).
- The drasyl thread pools are now lazily created when the first node is started.
- `MessageEvent.getMessage()` replaced by `MessageEvent.getSender()` and `MessageEvent.getPayload()`
.
- Messages and Events objects are now immutable.
- Non-started node now throws a `NoPathToPublicKeyException` for all outgoing messages.
- Removed `PeerUnreachableEvent` and `NodeIdentityCollisionEvent`.

Fixed

- Fixed concurrent modification error in `PeersManager`.
- Fixed concurrent modification error in `DirectConnectionsManager`.
- Fixed a bug that causes endless loops when a wanted handler was not the next handler in the
pipeline.
- Fixed a bug that causes a `NullPointerException` when the node is started and stopped too quickly.
- Fixed a bug that results in a partially started node if it was started and stopped too quickly.
- Make sure that monitoring is also shut down in case of an error.
- Reduce time till sending ping messages to prevent problems with too aggressive reverse proxies (I
mean you, nginx).

Added

- Release for Maven Central

Added

- Release for Maven Central

Added

- First release