Error Context & Backtraces

Attach context as errors bubble up. Enable backtraces (RUST_BACKTRACE=1) for panic and some error types to locate failure origins.

Recipe

use anyhow::Context;
 
fn load(path: &str) -> anyhow::Result<String> {
    std::fs::read_to_string(path)
        .with_context(|| format!("failed to read config at {path}"))
}

When to reach for this: Production debugging - know what operation failed, not only low-level cause.

Working Example

fn pipeline(cfg_path: &str) -> anyhow::Result<u16> {
    let raw = load(cfg_path)?;
    let port: u16 = raw.trim().parse().context("port must be integer")?;
    Ok(port)
}
 
fn main() {
    if let Err(e) = pipeline("missing.toml") {
        eprintln!("Error: {e:#}");
        for cause in e.chain() {
            eprintln!("  caused by: {cause}");
        }
    }
}

What this demonstrates:

• Layered context messages
• {e:#} alternate display with causes
• Chain iteration for logs

Deep Dive

std::backtrace::Backtrace capture on nightly/stable features varies - RUST_BACKTRACE=1 for panics. anyhow backtrace feature adds to errors.

Gotchas

• Context without actionable info - "failed" useless. Fix: Include paths, ids, operation names.
• Logging same error multiple layers - Duplicate noise. Fix: Log once at boundary with chain.
• Backtrace in hot path - Expensive capture. Fix: Capture on error only, not success.
• Secrets in context - Tokens in paths. Fix: Redact sensitive fields.
• thiserror without source - Broken chain. Fix: #[source] attribute on variants.

Alternatives

FAQs

Related

• anyhow
• thiserror
• Custom Error Types
• Observability

Stack versions: This page was written for Rust 1.97.0 (edition 2024), Tokio 1.x, Axum 0.8, serde 1.0, sqlx 0.8, clap 4, and Polars 0.46+.