Commanded Process Map

This tool is meant to help people which are working with elixir Commanded apps. The solution isn't some silver bullet and is based on as simple (and silly) solution as was possible to quickly get some satisfying result related to projects I am working with.

I was build this tool on project which is including now 65 nodes (aggregates, handlers, ...) and I hope this could help us get better picture about whole system.

Example

This example is generated for https://github.com/slashdotdash/conduit

Installation

1. copy config/dev.tmp.exs as config/dev.exs and set properly path to your project. The path_wildcard is used to get better files what we want to analyse.

2. execute ./bin/setup. Setup mix and install required node packages for frontend client.

3. execute ./bin/generate_data. This command will generate required json file for FE in frontend_client folder.

4. execute ./bin/run_client. Will run server (serving generated json file to FE) and open FE react application in browser.

Configuration & Base Logic

The main app logic is based on configuration and regular expressions. In config/dev.exs you can play with it.

• config :commanded_process_map, :project

  • path: set path to your project
  • path_wildcard: set wildcard pattern to anaylyse only files which matters. (eg. /lib/**/*.ex or /apps/*/lib/**/*.ex in umbrella apps)

• config :commanded_process_map, :types are representing nodes if FE flow graph. The key is type of node and following are

  • regexp regular expression to regognise node
  • not_one_of list of regular expressions to exclude file if some is evaluated as true
  • output what type of output is node emitting
  • targets what types of nodes are receiving output from this type of node

• config :commanded_process_map, :regexp are used for analysing each node what edges (commands and events) are they using. It is including regexp for named capturing and parsing and for scan parsing.

You can define as many nodes as want. They will be automatically grouped by colour in graph and edged by rules in config (output, targets). You can even create your own pseudo language for spiking ideas and update config regexp by it. It will generate you graph for presenting that logic/structure.*

TODO

• after regenerating data let server returning new data (no needs to restart)
• get rid of react compile warnings
• investigation/play with graph options to reach better UX
• update README by UI possibilities description (selecting, source code view, ...)
• improve selectors (suggestion filter, ...)