Serialization

[easyasme]

Concept Historian passes records containing potentally non-JSON-serializable objects to MasterSerializer. MasterSerializer recursively traverses the record and replaces any objects that needs serialization with dictionary {_ms_type, args, kwargs}. Dictionary is passed to Blobstorage - simply stores and retrieves these dictionaries without any custom serialization logic - simple implementation.

Modified MasterSerializer - automatically serialize objects by capturing their class name and constructor arguments (args, kwargs). for deserialization - it can reconstruct the object by importing the class and calling its constructor with the stored arguments.

PersistenHistory - use MasterSerializer to serialize records before passing them to the blobstorage and to deserialize records after retrieving them

Historian, Blobstorage, workflowmanager - do not need to change

The test file has not been implemented yet; it will be added after confirming that I am on the right track.

[easyasme]

Modifications MasterSerializer - modified to return the object as-is if no custom serialization is registered.

• register_serializer function implemented.
• serialize and deserialize method is changed to async.
• accepts both class and function. utils.py - implemented for functions that is used in other files as well. persistence.py - await serialize and deserialize methods. manager.py - passes the MasterSerializer when calling _create_history. init.py - _create_history function modified to have MasterSerializer as parameter

Questions persistence.py - I noticed that await can't be used in the init method. I used asyncio.run for the deserialization process. Do you think this is an acceptable solution, or should I refactor the code to use await instead?

serializer.py - In the deserialize function, I think that the factory might be a coroutine function. Do you think I need to handle this case by adding code to support asynchronous factories?

[easyasme]

Concept

• Serialization process is initiated by the PersistentHistory.append method which uses MasterSerializer to convert objects.
• Deserialization process occurs when loading workflow history via PersistentHistory.load_history. MasterSerializer is used to restore the serialized objects.
• Historian is aware of the serialization and deserialization process calling methods like load_history, _append_event_record

Modifications Updated Historian, PersistentHistory, workflowmanager, and init.py to pass and utilize MasterSerializer

init.py:

• Added the import for MasterSerializer to ensure it's available for the WorkflowManager
• Modified create_filesystem_manager function to initialize and pass MasterSerializer when creating PersistentHistory
• Adjusted create_history to include master_serializer

manager.py:

• Updated the WorkflowManager to instantiate MasterSerializer
• Modified the _start_workflow method to pass MasterSerializer to both Historian and PersistentHistory

persistence.py:

• load_history method to utilize MasterSerializer for deserialization of stored event records
• Updated the append method to serialize new event records using MasterSerializer
• Updated remove method

Questions

• If I could get some feedback on my PR for especially the following files if I'm applying async patterns correctly, or modifying it as the right concept: init.py, manager.py, historian.py -> load_history, _append_event_record, _record_version_event, handle_step