Spyder Code‑Completion Engine Decision Guide: Jedi vs. Rope

Decision: Which Code‑Completion Engine to Use in Spyder

Spyder 5.x+ ships with two built‑in completion engines: Jedi and Rope. The choice influences how quickly suggestions appear, how well type hints are understood, and whether you need extra dependencies. This guide helps you decide which engine fits your project’s needs, compares the options, and shows how to enable and verify the selected engine.

Constraints and Environment

• Spyder version: 5.x or newer is required for Jedi support. Spyder 4.x only supports Rope.
• Python interpreter: Jedi requires Python 3.6+; Rope works with 2.7 and 3.x but is less feature‑rich on newer syntax.
• Dependencies: Jedi pulls in pygments and its own parsing libraries; Rope depends on rope and pyflakes.
• Project type: Modern projects with type annotations, async code, and large libraries benefit from Jedi; legacy codebases or minimal dependencies may favor Rope.

Comparison Table

Feature	Jedi	Rope	Notes
Speed of suggestions	Fast, LSP‑style, real‑time	Slower, batch‑based	Jedi uses a lightweight AST parser.
Support for type hints	Excellent, resolves annotations in real time	Limited, may ignore hints in newer syntax	Jedi understands typing module constructs.
Compatibility with older Python	Python 3.6+ only	Python 2.7 & 3.x	Rope is the only option for legacy code.
Dependency footprint	High: pygments, jedi	Low: rope only	Consider environment size.
Accuracy on modern syntax (async, f‑strings)	Excellent	Partial	Jedi parses new language features natively.
Installation effort	Automatic with Spyder 5.x	Automatic with Spyder 5.x	Both engines are pre‑installed but may need pip updates.

Trade‑Offs

• Performance vs. Accuracy: Jedi is faster and more accurate for modern code, but may consume more memory. Rope is lightweight but can lag behind on large projects.
• Type Hint Support: If your code heavily uses typing, Jedi will offer richer completions. Rope may not surface type‑hinted names until you run a full analysis.
• Environment Constraints: In constrained or embedded Python environments, the lighter Rope may be preferable.
• Compatibility Issues: Some third‑party packages (e.g., C extensions) can confuse Jedi’s parser, leading to missing suggestions.

Choosing Jedi for Modern Projects

For projects targeting Python 3.8+ with extensive type hints, async functions, or large third‑party libraries, Jedi provides the most responsive and accurate experience. It integrates with the Language Server Protocol (LSP) style completion, giving you suggestions as you type.

Choosing Rope for Legacy or Minimal‑Dependency Projects

If you maintain a codebase that still runs on Python 2.7 or you need the smallest possible footprint, Rope is the safe bet. It’s also useful when you encounter conflicts with Jedi’s dependencies.

Step‑by‑Step Configuration

1. Open Spyder and go to Tools > Preferences.
2. Navigate to Completion > Code Completion Engine.
3. From the drop‑down list, select Jedi or Rope.
4. Click Apply and then OK.
5. Spyder will prompt you to restart. Click Restart Spyder to activate the new engine.

Permissions and Environment

Ensure you have write access to the Spyder config directory (typically ~/.spyder-py3 on Unix or %APPDATA%\Spyder on Windows). If you’re using a virtual environment, the engine will use the interpreter associated with that environment.

Validation: Quick Tests in the Console

After restarting, open the IPython console and perform the following tests. The results confirm the engine is active and functioning.

Test 1: Third‑Party Library Completion

# In the IPython console
import numpy as np
np.   # Place cursor after the dot and press Ctrl+Space
# Expected: list includes 'array', 'zeros', 'ones', etc.

Test 2: Type‑Hinted Function Suggestion

Create a file named utils.py with the following content:

def greet(name: str) -> str:
    return f"Hello {name}"

Then open another script and import it:

import utils
utils.   # Trigger completion
# Expected: 'greet' appears in the suggestion list.

Test 3: Rope‑Only Features

Switch to Rope in Preferences and repeat the above steps. Note any differences:

• Suggestion speed may be slower.
• Type‑hinted names might not appear until you run a full analysis (press Ctrl+Shift+F5 to re‑index).

Common Pitfalls and How to Avoid Them

• Missing Dependencies: If completions are not working, run pip install jedi pygments (for Jedi) or pip install rope (for Rope) in your active environment.
• Conflicting Extensions: Extensions like spyder‑extensions can override the completion engine. Disable them temporarily to test.
• Running Spyder with a different interpreter than your project can cause mismatched completions. Verify the interpreter via Tools > Preferences > Python interpreter.
• Large projects may trigger Jedi’s “indexing” delay. Allow a few seconds after startup for the engine to build its cache.

Summary

Use Jedi when you need fast, accurate completions for modern Python features and you’re comfortable with the extra dependencies. Choose Rope for legacy code, minimal footprints, or when Jedi’s performance is unacceptable. Follow the configuration steps, run the quick console tests, and monitor completion behavior to ensure the chosen engine meets your workflow needs.