eval·lens

evaluate your Python, through a lens.
_{Evalens is a VS Code extension.}

A notebook's feedback loop, inside an ordinary .py file.
Run a line when you choose. See its answer where you wrote it.

Put the cursor on a line. Press Cmd+Enter (Ctrl+Enter on Windows/Linux). The answer appears beside the code, and stays there as you move on.

The fourth line is the surprise. You changed b, yet a now contains 3 too. Both names refer to the same list.

Earlier readings stay beside their statements: a = [1, 2] still shows [1, 2], while the final a shows [1, 2, 3]. You can see what changed, and where. Four lines make an abstract idea concrete.

Who it's for

For students and anyone learning Python, Evalens makes it easier to connect the code you write with the values it produces. Predict an answer, run a statement, and compare. Teachers and mentors can use the same view to explain a tricky idea.

As you gain experience, use it to try expressions, inspect data, and explore unfamiliar code. Keep the debugger for breakpoints and stepping through function calls.

A total that keeps starting over

Suppose you want to add 2, 4, and 6. You expect 12, but this program gives 6.

Look at total beside the loop: 2, 4, 6. Each time around, total = n replaces the previous total with the next number. The earlier numbers are forgotten.

Change the line inside the loop to:

total = total + n

Read that as: take the total so far, add the next number, and store the new total. Evaluate the whole file again with Cmd+Alt+Enter (Ctrl+Alt+Enter on Windows/Linux), starting again from total = 0. Its history becomes 2, 6, 12, and the final answer is 12. Python also lets you write this more briefly as total += n. Keeping a running total like this is called accumulating; the recorded values let you see it happen.

That is useful in your first Python course. It is still useful when you know the language and are trying to understand someone else's code.

Your first evaluation

You need VS Code 1.90 or later and Python 3.9 or later. Evalens can use the interpreter selected in the Python extension, or find python3 / python on your PATH. No additional Python packages are required.

Install Evalens from the VS Code Marketplace, published by mbjarland, or run:

code --install-extension mbjarland.evalens

Open a Python file and try the four-line example:

a = [1, 2]
b = a
b.append(3)
a

Start on the first line. Press Cmd+Shift+Enter (Ctrl+Shift+Enter on Windows/Linux) to evaluate and move to the next statement. Before each press, predict what will appear.

With focus in the Python editor:

Nothing runs as you type. Evaluation runs a complete enclosing statement: placing the cursor inside a loop runs that loop. Your file stays ordinary Python, with no cell markers or saved outputs added to it.

Every command is also in the Command Palette: Cmd+Shift+P (Ctrl+Shift+P on Windows/Linux), then type Evalens. If a shortcut does nothing, try the command there and use Evalens: Fix Keybinding Conflict. See the installation and keybinding guide for other setups.

When the answer needs more room

Hover for a longer value or an error's traceback. For larger results, open Evalens: Show Values Panel: the source sits beside its recorded values and printed: output. Move between code and results and the matching row follows. Latest result marks what just finished, even after Evaluate and Advance moves on.

Fold a tall result back to the height of its code. Open a large recording in an editor tab to find, select, and copy the part you need. Browsing results does not run the program again.

A loop tells you what happened

Each pass through the outer loop sets base to ten times x and prints it. The inner loop adds each y to base, stores the result in v, and prints the pair x, y:

for x in range(2):
    base = x * 10
    print("base:", base)
    for y in range(3):
        v = base + y
        print(x, y)

Each loop keeps its own value history beside its for header. Here in the panel, Iteration 1 is folded and Iteration 2 is open: x is 1, base is 10, and the inner loop records v as 10, 11, 12. The short guide leads to that inner loop. Variables and Printed output share one pair of headings, so you can follow both down the page.

Short loops stay compact. Longer histories and output fold or page, with the same navigation at each level. If details were not saved, the panel says so. About these values explains when the values were read; the loop guide covers capture limits and missing readings.

Room to grow

For a place to start, choose Evalens: Open Learning Walkthrough, or Try a guided example in the empty Values panel. Five editable exercises invite you to predict a value, follow successive statements, explore aliasing, fix an accumulator, and recognize a stale result. Opening one never runs it.

Keep the guides nearby while you need them. Dismiss the introduction when you know your way around; Help and learning brings it back. The same tool is ready for inspecting an unfamiliar data structure, trying a comprehension, or adding an expression watch inside a loop. There is no beginner mode to outgrow.

When the question needs a breakpoint, a function call followed step by step, or a look at the call stack, use the Python debugger. Evalens runs statements and keeps their answers; the debugger lets you pause inside an execution. Both earn their place.

What stays, what resets

Each result records what that statement produced when it ran. It does not promise that Python still holds that value. Stale markers flag edits and some dependency changes; alias mutations are not fully tracked. Evaluate again when you want a fresh reading.

Statement evaluations share a Python session. Clear Inline Results clears the display and keeps variables. Restart Kernel discards variables and saved input answers, while earlier recordings remain visible. Evaluate File resets variables by default. A debugger starts its own execution. The session guide explains the choices.

Reference and support

User guide · Commands · Settings and colors · Release notes · Report an issue

MIT licensed. See LICENSE. The inline rendering approach builds on Calva.