Version 0.19.1

• Added margo_migrate_progress_loop

Version 0.19.0

• Adds optional support for the mochi-plumber library to automatically select optimal network cards for a given process
  • enabled via the --with-mochi-plumber configure option or +plumber spack variant
  • currently only supports Slingshot (CXI) networks

Version 0.18.3

• Adds margo_set_progress_when_needed

Version 0.18.2

• Bug fix: margo_top_(pre)finalize_callback should not remove callbacks;
• Added feature: spindown logic to improve performance of bursts of RPCs.

Version 0.18.1

• Moves execution of finalization callbacks from release time to finalize time

Version 0.18.0

• Better support for refcounting margo instances
• Added refcounting for margo-managed pools and xstreams

Version 0.17.3

• Fixes bug in earliest_first pool
• Adds the possibility to specify the default log level via MARGO_LOG_LEVEL

Version 0.17.2

• Make the progress loop yield more frequently to account for the possibility that the ES it runs on has other pools to check
• Adds warning related to some configurations that could cause starvation or deadlocks

Version 0.17.1

This patch fixes a bug that makes applications crash when monitoring (time series) is enabled and the user adds/removes pools.

Version 0.17.0

• Adds margo_monitor_dump function