Hi.

A few thoughts:

1. bot/chat/user_data are vailable in your main as updater.dispatcher.bot/chat/user_data. keep in mind, that updater.dispatcher.chat/user_data are dictionaries containing for each chat/user_id the data available as context.chat/user_data in callbacks for that chat/user.
2. if you store your custom class as updater.dispatcher.bot_data['some_key'] = my_class_instance and update it in your callbacks, it will be persisted both after each update and on shutdown, so there is no need for you to persist anything manually
3. we already have [FEATURE] Customization of context logic #2068 open, which will, when implemented, allow you to access your custom class e.g. as context.message_tracker

I think 1.+2. are a good way to achieve what you want for now.

1. Also the predefined function I mentioned, to call any function just after start and just before exiting of bot, will add a great functionality to any bot just like constructor and destructor of any object. These can be used for other functionalities also other the Pickling.

You can do whatever you want before calling updater.start_polling(), e.g. send a message via updater.bot.send_message. I'm not sure if I see a value in doing something on exit. Persistence is already updated on exit to make sure the data is stored, but the usual concept of bots is that they should not shut down (unless they need an update). If you have a specific use case in mind, please try to describe it :)

While exiting we can just add that custom class to updater.dispatcher.bot_data in one step only. No need to update that updater.dispatcher.bot_data class copy every time my object mutate. That will be helpful. Anyways, I will try using your solution and provide feedback.

Python passes objects by reference:

lst = [1,2]
context.bot_data['list'] = list
lst.append(3)
print(context.bot_data['list']) # this gives [1, 2, 3]

so no need to update the custom object before exit. its current state will be saved.

Thanks for help!