tracer
Trace wrapper package - wraps the boilerplace code for the standard
runtime/trace.
Getting started
Minimalistic example:
import (
"log"
"github.com/rusq/tracer"
)
func main() {
flag.Parse()
if err := run("trace.out"); err != nil {
log.Fatal(err)
}
}
func run(tracefile string) error {
t := tracer.New(tracefile)
if err := t.Start(); err != nil {
return err
}
defer t.Close()
// do something
return nil
}
For complete example see example/basic.
But why all this?
Adding tracing to your program is a great opportunity to get the insights on the
performance and locks.
Initialising tracing is usually done by adding the similar code to the beginning
of your main
function:
tf, err = os.Create(tracefile)
if err != nil {
log.Fatalf("failed to create trace output file: %s", err)
}
if err := trace.Start(tf); err != nil {
log.Fatalf("failed to start trace: %s", err)
}
Then, adding similar code to the end of the main:
if trace.IsEnabled() {
trace.Stop()
}
if err := tf.Close(); err != nil {
log.Fatalf("trace file failed to close: %s", err)
}
But what happens if the programme crashes and burns? Thanks for asking! If the
program crashes and burns, i.e. by calling the log.Fatal()
before the trace
file is properly closed, it will be corrupt, and go tool trace
will not be
able to parse it.
The easiest solution is to extract the logic into a run function:
func main() {
if err:=run(); err!=nil{
log.Fatal(err)
}
}
func run() error {
tf, err = os.Create(tracefile)
if err != nil {
log.Fatalf("failed to create trace output file: %s", err)
}
defer tf.Close()
if err := trace.Start(tf); err != nil {
log.Fatalf("failed to start trace: %s", err)
}
defer tf.Stop()
// call your functions
// do your magic
// stop all wars
// make this world a happier place
return nil
}
With this approach, unless the programme panics somewhere, Stop
and Close
would be inevitably called (in this order).
This package simply wraps the boilerplate code of opening/closing the trace file
and starting/stopping the tracer.
I just got tired of adding the same code over and over and over again, that is why.