Skip to content

usage.mdx

Latest commit

 

History

History
226 lines (164 loc) · 4.32 KB

usage.mdx

File metadata and controls

226 lines (164 loc) · 4.32 KB
title Usage Guide
description Examples and common debugging workflows with Pointbreak

Basic Debugging Workflow

1. Set Breakpoints

Ask your AI to set breakpoints:

"Set a breakpoint on line 42 of main.py"
"Add a breakpoint at the start of calculate_total()"

2. Start Debugging

"Start debugging with these breakpoints"
"Run the debugger and pause at the first breakpoint"

3. Inspect State

When paused at a breakpoint:

"Show me the value of user_input"
"What are all the local variables?"
"Display the stack trace"

4. Step Through Code

"Step over the next line"
"Step into this function call"
"Step out of this function"
"Continue until the next breakpoint"

Common Patterns

Finding a Bug

User: "This function returns the wrong value. Help me find the bug."

AI: I'll set a breakpoint at the function entry and step through:
1. Setting breakpoint at line 15...
2. Starting debugger...
3. [Paused at line 15]
4. Current value of input: [1, 2, 3]
5. Stepping through...
6. Found the issue: You're dividing by zero when the list is empty.

Understanding Complex Logic

User: "How does this recursive function work?"

AI: Let me step through a few iterations:
1. Setting breakpoint at recursion entry...
2. Starting with input: 5
3. [First call] n=5
4. [Second call] n=4
5. [Third call] n=3
...

Tracking Variable Changes

User: "When does this variable change?"

AI: I'll set a conditional breakpoint:
"Set breakpoint on line 30 with condition: count > 10"

Advanced Features

Conditional Breakpoints

"Set a breakpoint on line 42, but only break when x > 100"
"Add a breakpoint that only triggers after 5 hits"

Log Points

"Add a log point on line 30 that prints the value of total"

This prints without stopping execution!

Watch Expressions

"Add a watch expression for user.name"
"Monitor the value of count throughout execution"

Evaluate Expressions

"Evaluate: len(data) * 2"
"What's the result of: user.age + 10?"

Language-Specific Examples

Python

def process_data(items):
    result = []
    for item in items:  # Breakpoint here
        processed = item * 2
        result.append(processed)
    return result

Ask: "Set a breakpoint in the loop and show me item values"

Rust

fn calculate_sum(numbers: &[i32]) -> i32 {
    let mut sum = 0;  // Breakpoint here
    for &num in numbers {
        sum += num;
    }
    sum
}

Ask: "Break at the sum initialization and step through the loop"

JavaScript

function fetchUserData(userId) {
  const response = fetch(`/api/users/${userId}`); // Breakpoint here
  return response.json();
}

Ask: "Set a breakpoint and show me the userId value"

Best Practices

✅ Do

  • Be specific about line numbers
  • Set breakpoints before starting the debugger
  • Ask for context (stack trace, variables)
  • Use natural language
  • Step through unfamiliar code

❌ Don't

  • Add console.log or print statements (use the debugger!)
  • Guess at values (inspect them!)
  • Run without breakpoints in complex code
  • Forget to ask for the stack trace when confused

Debugging Strategies

Top-Down

1. Set breakpoint at function entry
2. Step through line by line
3. Inspect variables at each step
4. Find where behavior diverges from expected

Bottom-Up

1. Set breakpoint where error occurs
2. Check stack trace
3. Set breakpoints up the call chain
4. Find where bad data originates

Binary Search

1. Set breakpoint in middle of suspect code
2. Check if bug already present
3. Move breakpoint up or down
4. Repeat until bug location narrowed down

Tips for AI Assistants

Being Clear

❌ "Debug this" ✅ "Set a breakpoint on line 42, start debugging, and show me the variable values"

Providing Context

❌ "It's broken" ✅ "This function should return 10 but returns 0. Set breakpoints and find why."

Iterating

User: "Set a breakpoint at line 15"
AI: [Sets breakpoint and starts debugging]
User: "Now step into the function call"
AI: [Steps in]
User: "What's the value of result?"
AI: [Shows value]

Questions?