Today, we are excited to announce the release of drasyl 0.11.0. This maintenance release brings performance improvements, fixes and the new CLI sub-command sdon for the creation of dynamic overlays.

Changelog

🚀 Added

  • CLI: Option --rc-start-node to sub-command node added.
  • CLI: Option --unavailability-cause to sub-command version added.
  • CLI: Sub-command node-rc peers added.
  • CLI: Sub-command sdon added.
  • EventTypeDrasylNode added.
  • DrasylNode will now check local time is correct on start as this is mandatory for drasyl protocol to work.
  • InboundExceptionEvent now contains exception in string presentation.
  • DrasylConfig: Option drasyl.remote.tcp-fallback.connect-port added.
  • Boolean system property org.drasyl.pub-key.interning added to disable interning of IdentityPublicKeys, as this could have a negative impact on performance..

💅 Changed

  • Dependencies have been updated.
  • TCP fallback now connects to all super peers.
  • Application messages are not longer passed through the DrasylServerChannel (this was a performance bottleneck).
  • CLI: Option --ack-interval from sub-command wormhole receive removed (the new reliability layer determines this value automatically).
  • CLI: Options --window-size and --window-timeout from sub-command wormhole send removed (the new reliability layer determines this value automatically).
  • DrasylConfig: Options drasyl.remote.message.mtu, drasyl.remote.message.max-content-length, drasyl.remote.message.composed-message-transfer-timeout, drasyl.remote.message.arq removed (the new reliability layer determines this value automatically).
  • DrasylConfig: Options drasyl.remote.tcp-fallback.timeout, drasyl.remote.tcp-fallback.address removed.
  • Improved performance in inbound and outbound message processing by reducing task scheduling overhead. Instead of creating a separate task for each message, a single task now processes entire batches of reads and writes. Additionally, fewer Java objects are created, and zero-copy techniques are applied more frequently where possible.

🐛 Fixed

  • CLI: ensure wormhole sender/receiver handlers gracefully stop when connection is lost.
  • CLI: fix tunnel sub-command.
  • Receiving multiple UNITE messages will now trigger new HELLO messages only if new endpoints have been reported.

Photo by Anton Filatov on Unsplash

Related Posts