Stellar Core Pre Release v10.0.0rc1

  • This release includes changes to the protocol as well as some stability improvements.

    As it changes the meta data format, a newer version (0.14.0 or greater) of Horizon is required.

    Stability improvements

    • Avoid 100% CPU usage when running certain commands (--newhist for example).
    • Handshake code in overlay had issue that would lead to peers not able to an error “unexpected MAC”.
    • Align publish with network consensus, this avoids potentially publishing bad ledgers in case of corruption.
    • Align catchup with network consensus, this avoids having to rely on timing (which may cause significant delays in catchup).
    • Properly fail work when spawning a process throws an exception.
    • Transactions are better flooded on the network.
    • SCP: synchronize nodes over time to reduce chance of timeout
    • SCP: nomination was sometimes computing “round leader” properly, leading to timeouts
    • SCP: ensure latest quorum information is used when whitelisting validators
    • Move to a more recent version of asio.
    • Reduced number of warnings.
    • Properly kill subprocesses.
    • Moved to Catch2 (testing).
    • Moved to c++ 14 for the toolchain.

    Protocol Changes (v10)

    • BumpSeqOp w/ seqnum processing changes
    • Made offer rounding behavior more intuitive
    • Moved signature verification to transaction apply step
      • CAP0002
    • Fast fail not allowing to change trust on native assets

    New features

    • Improved private network load testing
      • updated loadgenerator to allow for more scalable and parallel submission of synthetic traffic.
      • Documented how to do performance testing
    • New metrics:
      • archive health
      • performance characteristics of validators and network
    • Validators can now be configured to upgrade to protocol versions other than the latest supported.
    • Support for running tests in parallel

    Source Code

Looks like your connection to Cryptocentral was lost, please wait while we try to reconnect.