Still allow the pandas DataFrame backend, but also allow using a MongoDB backend to store everything except binary files.
Have a way to import and export between mesmerize pandas DataFrame and Mongo.
For caiman extensions we can just use pandas as a front end and query the mongodb. This should be possible, needs more thought.
Advantages: can work with or run multiple items simultaneously, don't have to worry about the DataFrame file being constantly overwritten. Can add multiple items simultaneously and keep working on then concurrently from different processes, such as realtime gridsearch during live experiments.
Still allow the pandas DataFrame backend, but also allow using a MongoDB backend to store everything except binary files.
Have a way to import and export between mesmerize pandas DataFrame and Mongo.
For caiman extensions we can just use pandas as a front end and query the mongodb. This should be possible, needs more thought.
Advantages: can work with or run multiple items simultaneously, don't have to worry about the DataFrame file being constantly overwritten. Can add multiple items simultaneously and keep working on then concurrently from different processes, such as realtime gridsearch during live experiments.