Quickly sketching diagrams on black/whiteboards is a great way to understand and explain complex systems. If we could digitize the experience, that knowledge could be stored for future reference. Doing so would also allow easier colloboration and extension. These are the motivations behind Pannus et Circulos (cloth and circles):

Once digitized it is possible to make dynamic illustrations. E.g. in an emergency instead of trying to visualize the mental model of a broken system, why not automate chart updates:

curl -X PATCH -d '{ "op": "add", "path": "/nodes/server-2/className", "value": "failed server" }' http://api-server/docs/$DOCID

Building on a simple API would allow knowledge to be programmed into the chart:

curl -X PATCH -d '{ "op": "add", "path": "/edges/lb-1-server-3/className", "value": "red-edge" }' http://api-server/docs/$DOCID
curl -X PATCH -d '{ "op": "add", "path": "/edges/lb-2-server-3/className", "value": "red-edge" }' http://api-server/docs/$DOCID

With a chart data model based on simple standards (css and json) accessed simultaneously through a UI and API allowing the best of automated and human input:

{
  "nodes": {
    "server-1": {
      "id"   : "server-1",
      "label": "server 1",
      "x"    : 10,
      "y"    : 20
    },
    "mysql": {
      "id"       : "database-1",
      "label"    : "mysql",
      "className": "mysql",
      "x"        : 30,
      "y"        : 40
    }
  },
  "edges": {
    "server-1-mysql": {
      "id"       : "connection-1",
      "source"   : "server-1",
      "target"   : "database-1",
      "className": "red-edge"
    }
  },
  "style": {
    "nodes": {
      "mysql": {
        "backgroundColor": "#DDF",
        "borderColor"    : "#369",
        "borderRadius"   : "20px",
        "borderWidth"    : "10px",
        "height"         : "32px",
        "width"          : "32px"
      }
    },
    "edges": {
      "red-edge": {
        "paintStyle": {
          "stroke": "red"
        }
      }
    }
  }
}

git clone [email protected]:typingduck/petc.git
cd petc
npm install
npm start  # this starts the UI/API server and DB.

• See db/README.md to setup initial database (just some tables to create).
• When ready the UI will be visible at http://localhost:5000

• DB: 5002
• API: 5001
• UI: 5000

A React app, node API server and couchdb database. Uses socket.io websockets to allow realtime collaboration/updates. Each of the components db, server, client are self contained in their own directory. See their respective directories for more details, running them independently, testing, etc.

Started as a prototype, but went better than expected. Having reached the intial milestone, I consider it done pending a good reason to take it to the next level. Would be nice to make the whiteboard-esque behaviour complete and productionize the backend.

This project currently uses comic sans, a font known to the State of California to potentially cause seizures in individuals with prior sensitivity to taste. But remember: Autoritätsdusel ist der größte Feind der Wahrheit.