rust-error

Solution Patterns

Pattern 1: Option for Normal Absence

// Lookup operations where "not found" is normal
fn find_user(id: u32) -> Option<User> {
    users.get(&id)
}

// Usage patterns
match find_user(123) {
    Some(user) => println!("Found: {}", user.name),
    None => println!("User not found"),
}

// Or convert to Result for propagation
let user = find_user(123).ok_or(UserNotFoundError)?;

When to use: Queries, lookups, optional configuration values.

Key insight: None carries no information, just absence.

Pattern 2: Result for Expected Failures

// File might not exist (expected failure)
fn read_config(path: &Path) -> Result<String, io::Error> {
    std::fs::read_to_string(path)
}

// Network request might timeout
fn fetch(url: &str) -> Result<Response, reqwest::Error> {
    reqwest::blocking::get(url)
}

When to use: I/O operations, parsing, validation, network calls.

Key insight: Error type carries information about why it failed.

Pattern 3: Custom Error Types (thiserror)

use thiserror::Error;

#[derive(Error, Debug)]
pub enum ParseError {
    #[error("invalid format: {0}")]
    InvalidFormat(String),

    #[error("missing required field: {0}")]
    MissingField(&'static str),

    #[error("IO error")]
    Io(#[from] io::Error),

    #[error("parse error: {0}")]
    Parse(#[from] serde_json::Error),
}

// Use in library code
pub fn parse_config(input: &str) -> Result<Config, ParseError> {
    let raw: RawConfig = serde_json::from_str(input)?;  // Auto-converts
    validate_config(raw)
}

When to use: Library code, public APIs, need type-safe error handling.

Trade-offs: More boilerplate, but precise error types.

Pattern 4: Flexible Errors (anyhow)

use anyhow::{Context, Result, bail};

fn process_request() -> Result<Response> {
    let config = std::fs::read_to_string("config.json")
        .context("failed to read config file")?;

    let parsed: Config = serde_json::from_str(&config)
        .context("failed to parse config as JSON")?;

    if !parsed.is_valid() {
        bail!("invalid configuration: missing API key");
    }

    Ok(build_response(parsed))
}

When to use: Application code, rapid development, error context matters more than types.

Trade-offs: Loses type information, but gains flexibility and context.

Workflow

Step 1: Classify the Failure

Is absence normal?
  → Option<T>

Is failure expected and recoverable?
  → Result<T, E>

Is this a bug or invariant violation?
  → panic!() or assert!()

Step 2: Choose Error Representation

Library code (public API)?
  → thiserror (typed errors)

Application code (internal)?
  → anyhow (flexible errors)

Need error conversions?
  → Implement From traits

Step 3: Propagate or Handle

Can caller handle this?
  → Return Result, use ?

Need to add context?
  → .context("why it failed")?

Must handle here?
  → match / if let / unwrap_or

Error Propagation Best Practices

✅ Good Patterns

// Clear error types
fn validate() -> Result<(), ValidationError> {
    if name.is_empty() {
        return Err(ValidationError::EmptyName);
    }
    Ok(())
}

// Add context during propagation
let config = File::open("config.json")
    .context("failed to open config.json")?;

// Use ? operator
let data = read_file(&path)?;

// Provide defaults
let timeout = config.timeout.unwrap_or(Duration::from_secs(30));

// Pattern match for complex handling
match parse_input(input) {
    Ok(value) => process(value),
    Err(ParseError::InvalidFormat(msg)) => log_and_retry(msg),
    Err(e) => return Err(e),
}

❌ Anti-Patterns

// ❌ unwrap() on operations that can fail
let content = std::fs::read_to_string("config.json").unwrap();

// ❌ Silently ignore errors
let _ = some_fallible_operation();

// ❌ Generic error messages
Err(anyhow!("error"))  // Too vague

// ❌ Converting all errors to strings
.map_err(|e| e.to_string())?  // Loses type info

// ❌ Panic for expected failures
let num: i32 = input.parse().expect("parse failed");  // User input!

When to Panic

✅ Acceptable Panic Scenarios

// ✅ Acceptable: initialization check
let api_key = std::env::var("API_KEY")
    .expect("API_KEY environment variable must be set");

// ✅ Acceptable: test assertion
#[test]
fn test_user_creation() {
    let user = create_user("Alice");
    assert_eq!(user.name, "Alice");
}

// ✅ Acceptable: invariant violation
let first = queue.pop().expect("queue should never be empty at this point");

❌ Unacceptable Panic Scenarios

// ❌ User input validation
let num: i32 = input.parse().unwrap();  // Use Result instead

// ❌ Network operations
let response = reqwest::blocking::get(url).unwrap();  // Use Result

// ❌ File operations
let config = std::fs::read_to_string("config.json").unwrap();  // Use Result

Error Type Design

Enum for Multiple Error Cases

#[derive(Error, Debug)]
pub enum ConfigError {
    #[error("file not found at {path}")]
    FileNotFound { path: String },

    #[error("invalid syntax at line {line}: {message}")]
    InvalidSyntax { line: usize, message: String },

    #[error("missing required field: {0}")]
    MissingField(String),

    #[error(transparent)]
    Io(#[from] io::Error),

    #[error(transparent)]
    Parse(#[from] serde_json::Error),
}

Nested Errors with Context

#[derive(Error, Debug)]
pub enum AppError {
    #[error("configuration error")]
    Config(#[from] ConfigError),

    #[error("database error")]
    Database(#[from] DatabaseError),

    #[error("authentication failed: {0}")]
    Auth(String),
}

Common Pitfalls

Quick Reference

Review Checklist

When reviewing error handling code:

• All fallible operations return Result or Option
• Error types are meaningful (not just String)
• Error context is preserved through propagation
• unwrap() used only with justification (comments)
• panic!() used only for bugs or unrecoverable states
• Library code uses typed errors (thiserror)
• Application code adds context (anyhow .context())
• Error messages are actionable for users/operators
• No silent error swallowing (let _ = ...)
• Tests cover error paths, not just happy paths

Verification Commands

# Check for unwrap/expect usage
cargo clippy -- -W clippy::unwrap_used -W clippy::expect_used

# Check for panic in production code
cargo clippy -- -W clippy::panic

# Run tests including error paths
cargo test

# Check for unused Results
cargo clippy -- -D unused_must_use

# Verify error types implement Error trait
cargo check

Conversion Patterns

Option ↔ Result

// Option → Result
let result: Result<T, E> = option.ok_or(error_value)?;
let result: Result<T, E> = option.ok_or_else(|| compute_error())?;

// Result → Option
let option: Option<T> = result.ok();

// Result<Option<T>, E> → Result<T, E>
result.and_then(|opt| opt.ok_or(error))?;

Error Type Conversions

// Manual conversion
.map_err(|e| MyError::from(e))?;

// Automatic with #[from]
// Requires: #[derive(Error, Debug)] with #[from] attribute
result?;  // Auto-converts if From impl exists

// Add context
.map_err(|e| MyError::Wrapped(e.to_string()))?;

// Use anyhow for flexibility
.context("operation failed")?;

Advanced: Error Source Chains

use std::error::Error;

fn print_error_chain(e: &dyn Error) {
    eprintln!("Error: {}", e);

    let mut source = e.source();
    while let Some(e) = source {
        eprintln!("  Caused by: {}", e);
        source = e.source();
    }
}

// Usage
if let Err(e) = dangerous_operation() {
    print_error_chain(&e);
}

Related Skills

• rust-error-advanced - Advanced error patterns (thiserror, anyhow, error chains)
• rust-anti-pattern - Error handling anti-patterns to avoid
• rust-coding - Error handling coding standards
• rust-web - Error handling in web contexts
• rust-async - Error handling in async code

Localized Reference

• Chinese version: [SKILLZH.md](./SKILLZH.md) - 完整中文版本，包含所有内容