node-debugger

This is an implementation of a state machine for troubleshooting the TLA+ modeling work that is happening here.

Initial states

To interact with the state machine, one must first create an initial state. This can be done by using the functionality in lib/initial_state\initial_state.ml or one can simply create an essentially empty initial state with the init function in lib/node_info/node_info.ml.

Once an initial state is created, we can apply actions and see what effect they have on the state.

Viewing states and trace

There is functionality to view the state in lib/node_info/node_info.ml and the execution trace in lib/execution/execution.ml.

Enabling conditions

There is also functionality to print out a list of all enabled actions in the current state. These are the actions that one may apply to advance the state.

Getting started

Clone the repo and use dune utop to load all the modules into the OCaml toplevel.

Now I like to do:

open Basic;;
open Execution;;
open Node_info;;

because it makes the whole experience a lot less painful.

Next, we create an initial state, e.g. let state = init 3, but you can certainly do something fancier with Initial_state. I'll call the state state.

Now we can view the enabled actions in state with enabled state. For example, you will see something like:

Enabled system actions:
  New_chain
  New_branch
  New_block
Enabled node actions:
  Activate
    node 1 :> chain(s) 1
    node 2 :> chain(s) 1
    node 3 :> chain(s) 1

This displays the enabled node and system actions. It says the system actions New_chain, New_branch, and New_block are enabled (these actions are always enabled) and the node actions Activate (1, 1), Activate (2, 1), and Activate (3, 1) are enabled (since the nodes are currently inactive on chain 1, they can become active on chain 1; chain 1 is currently the only chain that exists).

Now we can apply any of these enabled actions, e.g. let's apply Activate (2, 1) (node 2 becomes active on chain 1). To do this, we enter activate' state 2 1. At this point, we can print view state to display the entire state or print (view_node' state) 2 to only display node 2's state or print (view_chain' state) 1 to only display chain 1's state.

Now if we display the enabled actions, we get some new options:

Enabled system actions:
  New_chain
  New_branch
  New_block
  Advertise: Current_branch on chain(s) 1
  Advertise: Current_head
    chain 1 :> branch(es) 0
Enabled node actions:
  Activate
    node 1 :> chain(s) 1
    node 3 :> chain(s) 1
  Deactivate
    node 2 :> chain(s) 1

At any point, we can apply any enabled action to the state or view any part of the state.