Tired of uninformative error output? Probably this will be more convenient:

package main

import (
	"io/ioutil"

	"github.com/ztrue/tracerr"
)

func main() {
	if err := read(); err != nil {
		tracerr.PrintSourceColor(err)
	}
}

func read() error {
	return readNonExistent()
}

func readNonExistent() error {
	_, err := ioutil.ReadFile("/tmp/non_existent_file")
	// Add stack trace to existing error, no matter if it's nil.
	return tracerr.Wrap(err)
}

Find more executable examples in examples dir.

import "github.com/ztrue/tracerr"

err := tracerr.New("some error")

Or:

err := tracerr.Errorf("some error %d", num)

If err is nil then it still be nil with no stack trace added.

err = tracerr.Wrap(err)

Stack trace will be printed only if err is of type tracerr.Error, otherwise just error text will be shown.

This will print error message and stack trace if any:

tracerr.Print(err)

This will add source code:

tracerr.PrintSource(err)

It's able to set up number of lines of code to display for each frame, which is 6 by default:

tracerr.PrintSource(err, 9)

Or to set up number of lines before and after traced line:

tracerr.PrintSource(err, 5, 2)

The same, but with color, which is much more useful:

tracerr.PrintSourceColor(err)

tracerr.PrintSourceColor(err, 9)

tracerr.PrintSourceColor(err, 5, 2)

It's also able to save output to variable instead of printing it, which works the same way:

text := tracerr.Sprint(err)

text := tracerr.SprintSource(err)

text := tracerr.SprintSource(err, 9)

text := tracerr.SprintSource(err, 5, 2)

Stack trace will be empty if err is not an instance of tracerr.Error.

frames := tracerr.StackTrace(err)

Or if err is of type tracerr.Error:

frames := err.StackTrace()

Unwrapped error will be nil if err is nil and will be the same error if err is not an instance of tracerr.Error.

err = tracerr.Unwrap(err)

Or if err is of type tracerr.Error:

err = err.Unwrap()

Stack trace causes a performance overhead, depending on a stack trace depth. This can be insignificant in a number of situations (such as HTTP request handling), however, avoid of adding a stack trace for really hot spots where a high number of errors created frequently, this can be inefficient.

Benchmarks done on a MacBook Pro 2015 with go 1.11.

Benchmarks for creating a new error with a stack trace of different depth:

BenchmarkNew/5    200000    5646 ns/op    976 B/op   4 allocs/op
BenchmarkNew/10   200000   11565 ns/op    976 B/op   4 allocs/op
BenchmarkNew/20    50000   25629 ns/op    976 B/op   4 allocs/op
BenchmarkNew/40    20000   65833 ns/op   2768 B/op   5 allocs/op