Use the REPL in Daily Clojure Workflow

REPL-driven development is not “type random code until it works.” A good Clojure workflow uses the REPL for fast feedback while keeping source files, tests, and Git as the durable record.

The Daily Loop

    flowchart LR
	    A["Edit source"] --> B["Evaluate form"]
	    B --> C["Inspect value"]
	    C --> D["Adjust code"]
	    D --> B
	    C --> E["Write or update test"]
	    E --> F["Commit source truth"]

The REPL shortens the loop, but the commit still contains source and tests. That distinction keeps the workflow collaborative.

Connect the Editor

A productive REPL workflow usually evaluates code from the editor, not by pasting large blocks into a terminal.

Choose the editor that gives you reliable evaluation and clear feedback. Brand loyalty matters less than the ability to evaluate the right form quickly.

Work From Source, Not Terminal History

Prefer this loop:

1. Write or edit a function in a source file.
2. Evaluate that function or top-level form.
3. Call it with sample data.
4. Move useful sample data into a test.
5. Commit the source and tests.

Avoid this loop:

1. Define ten functions directly in the REPL.
2. Forget which definitions matter.
3. Restart the REPL and lose the working state.
4. Reconstruct behavior from history.

The first loop is fast and reviewable. The second loop is private and fragile.

Use Sample Data Deliberately

Keep small domain-shaped examples nearby:

1(def sample-order
2  {:order/id 1001
3   :order/customer {:customer/id 42}
4   :order/lines [{:sku "A1" :qty 2 :price 10}
5                 {:sku "B2" :qty 1 :price 25}]})

Then evaluate source-backed functions:

1(orders.core/order-total sample-order)

When the sample reveals a real requirement, move it into a test fixture or test body.

Convert REPL Wins Into Tests

The REPL answers “does this work right now?” A test answers “will this keep working?”

This is the same engineering discipline Java teams already value; Clojure just gives you a faster way to find the right assertion.

Keep the Running System Honest

Long REPL sessions can drift. Use checkpoints:

A Practical Java-to-Clojure Rule

If you would normally add a breakpoint, log line, or temporary controller endpoint in Java, first ask whether you can evaluate the relevant function directly in Clojure.

That question pushes you toward:

• smaller functions
• explicit data
• fewer hidden dependencies
• better tests
• faster feedback

Key Takeaways

• A good REPL workflow is source-backed, not history-backed.
• Evaluate from your editor whenever possible.
• Use the REPL to discover behavior, then encode durable behavior in tests.
• Restart when dependency or classpath changes make the live JVM suspect.
• The REPL complements Git, CI, and tests; it does not replace them.

Quiz: REPL Workflow