Consistency of Execution Counts across Different Kernel Versions

Execution Count State Management

Jupyter Notebooks utilize a ZeroMQ-based messaging protocol to synchronize state between the frontend and the kernel. While the kernel manages the active memory state, the frontend tracks the sequence of cell execution via execution counts stored in the notebook metadata.

A challenge arises when notebooks are shared across diverse environments or different kernel versions. Because the execution count is an incremental integer tied to a specific session's history, the metadata may not align with the actual state of a kernel in a new environment, potentially leading to confusion regarding the order of operations during reproduction.

What is the documented behavior for reconciling execution counts when a notebook is opened in a kernel version that differs from the one used during the original save? Does the protocol provide a mechanism to validate if the current kernel state matches the recorded execution sequence?