The command stack is used for long running tasks that need to maintain state.
An example is a threaded conversation with a user than ends in a task being performed.
The command stack is built up of root commands and a stack of current commands.
Commands can be either added as root commands or pushed on to the stack of current commands. Root commands are commands that should always be available. A module will usually have one root command that is the entry point for all its functionality. Commands added to the stack represent discrete pieces of work that will end at some point. e.g. A command that PMs a question and then saves the next number between 1 and 10 in reply via pm maybe added as part of a command that collects OKR stats.
When the command stack receives a message each command starting from the top of the stack chooses whether to handle the message. When a command elects to handle a message no futher commands are given the opportunity. If none of the current commands handles the message, then each root command is given a chance in the order that the root commands were added, regardless of whether another root command chose to handle it.
Any command may push children commands to the stack. When the child is popped from the stack the parent command is notified. Typically a command is responsible for popping itself from the stack. When a command is popped from the stack all children of that command are also popped but their parents are not notified.
With this structure complex long running command trees can be composed together
from simple primitive commands. For example the user module may add a root
command that handles any private messages from an admin user exactly equal to
"add user", "remove user", or "update user".
When this root command recieves an "add user" message it then creates and pushes an add user command to the stack. This command in turn pushes a subcommand which when added to the stack asks the user a question and saves the input. e.g. "Please type the email address for the new user". When the user responds the command pops itself from the stack notifying the parent add user command. The add user command then pushes a new subcommand for a first name and so forth. Finally, once the add user command has collected all the required information it adds the user to the database, notifies the user, and pops itself from the current stack.
- Add user
- Set user role
- Delete user
- Identify user from chat id
- Be able to add OKR questions for a user
- Questions have replaceable tokens
- Answers can be boolean, range, or string
- Answers can be exported or perhaps viewed on web
- Questions are asked on a cron schedule
- Questions are posed via private message and replies parsed