0.1.10

Memory improvements

• A new flushing mechanism has been implemented allowing aggregates in CRDT variables to behave as moving averages, adding replicated aggregation on top of local computations.
• First experimental deployments show near constant memory footprint
• Data has been split into specific variables for each type and each node, removing the need to store repetitive identification tags in variables.

State convergence

• Improved recovery after crashes
• Successful retrieval of data after reboot without durable storage
• ⚠️ Divergent behavior has not been completely resolved yet. Currently ongoing work #8