FlowStorm: A Powerful Tool for Debugging and Understanding Clojure Code

• 0:00:08 Initial Plan and Change of Agenda: The speaker initially planned to talk about FlowStorm's internals and new features in FlowStorm 4. However, they realized that much of this content would overlap with a previous talk.
• 0:00:58 New Title and Agenda: The talk's title was changed to "The Power of the Storm," and the agenda shifted to focus on practical usage and demos, including:
  • FlowStorm refresher
  • Instrumentation and recording
  • Workflow tips and tricks
  • New features like the outputs panel and data windows
• 0:02:07 FlowStorm Overview: FlowStorm is a development tool for Clojure that records program execution and visualizes data.
• 0:02:22 Key Difference: Unlike other tools, FlowStorm records program execution as a "flow," which is a timeline of all thread activities.
• 0:02:54 Local vs. Remote Mode: FlowStorm can operate in local mode (everything on the same JVM) or remote mode (instrumented code and recordings on a separate process).
• 0:05:34 Instrumentation with ClojureStorm: ClojureStorm, a fork of the official Clojure compiler, adds instrumentation to emit extra bytecode for tracing, enabling detailed analysis without modifying the original code.
• 0:08:43 Recording in FlowStorm: The tool hooks into ClojureStorm and stores all events, function calls, and intermediate object references with pointers to source code coordinates.
• 0:10:07 Thread Timelines: FlowStorm organizes recordings into thread timelines, which are arrays of events, providing a detailed view of each thread's execution.
• 0:10:59 Demo: Web Applications: The speaker demonstrates how to use FlowStorm to analyze a web application, showing features like:
  • Recording requests and responses
  • Naming and recalling specific requests
  • Searching for values within recordings
  • Defining and running print statements on recordings
• 0:21:20 Demo: Data Windows for HTML Visualization: The speaker introduces data windows, a new feature for visualizing data, and demonstrates how to view HTML output directly within FlowStorm.
• 0:22:17 Demo: Editor Integration: The speaker shows how to jump from the editor to recordings and back, using features like:
  • Quick jump to function
  • "debugger;" statement for fast positioning
  • Defining all frame variables in the REPL
  • Opening source code from recordings
• 0:27:45 Demo: Codebase Exploration: The speaker demonstrates how to use FlowStorm to explore a new codebase (Clojure's core.async flow library), showing how to:
  • Record and analyze the creation of flows
  • Identify and analyze threads spawned by the library
  • Use bookmarks to mark interesting points in the recordings
• 0:43:21 Accessing Recordings from the REPL: FlowStorm allows programmatic access to recordings via the REPL using the get-timeline function.
• 0:44:13 Demo: Outputs Panel: The speaker introduces the outputs panel, a replacement for the REPL console that provides enhanced visualization and navigation of evaluation results.
• 0:46:24 Demo: Data Windows: Data windows are a system for data inspection and visualization, supporting:
  • Lazy navigation of nested data structures
  • Infinite sequences
  • Multiple visualizations
  • Custom visualizations
  • Clojure datafy and navigation
  • Real-time data visualization
• 0:50:34 Custom Visualizations: The speaker demonstrates how to create custom visualizations for specific data types, using a chessboard example.
• 0:52:25 Datafy and Navigation Integration: FlowStorm integrates with Clojure's datafy and navigation protocols, allowing seamless exploration of data structures that support these protocols.
• 0:56:20 Built-in Visualizations: FlowStorm provides built-in visualizations for numbers, byte arrays, and other common data types.
• 0:56:27 EQL Queries: Data windows support EQL (Extensible Query Language) queries for filtering and analyzing data.
• 0:59:14 Real-time Visualization: The speaker demonstrates how to create data windows that update in real time, using a sine wave example.
• 1:01:04 Ants Demo: The speaker revisits the classic "ants" demo to showcase how data windows can be used to visualize and debug agent-based simulations.
• 1:03:05 Conclusion: The speaker concludes by encouraging the audience to explore FlowStorm and its features.