php editor Apple will give you the answer: In the Go language, when an error occurs in a goroutine, it will not be automatically propagated to the main coroutine. Instead, it is silently ignored, which may result in you receiving only partial errors instead of errors in all goroutines launched. This is because the original intention of the Go language design is to keep the program stable and efficient, and the entire program will not be stopped immediately even in the event of an error. If you want to catch all errors, you can use a channel or other mechanism to pass error information explicitly. This way you can ensure that all errors are handled correctly.
I defined a cycle class to handle concurrent tasks. What I want is to run two functions, each in a goroutine, wait for them to complete and merge their output errors together. But I only get an error. The responsibilities of each method are as follows:
run
-Run a function in a goroutine and collect its errors
waitalldone
- merge all function errors together and wait for all functions to complete
do1, do2
- Test function
import ( "fmt" "go.uber.org/multierr" "sync" "testing" ) type Cycle struct { errChan chan error wg sync.WaitGroup } func NewCycle() *Cycle { return &Cycle{ errChan: make(chan error), wg: sync.WaitGroup{}, } } // run fn and collect its error into error channel func (c *Cycle) Run(fn func() error) { c.wg.Add(1) go func() { defer c.wg.Done() if err := fn(); err != nil { c.errChan <- err } }() } // wait all fn finish and combine their error together func (c *Cycle) WaitAllDone() error { var err error go func() { for { if tmpErr, ok := <-c.errChan; ok { err = multierr.Append(err, tmpErr) } else{ break } } }() c.wg.Wait() close(c.errChan) return err } func Do1() error { return fmt.Errorf("ERR1") } func Do2() error { return fmt.Errorf("ERR2") } func Test41(t *testing.T) { c := NewCycle() c.Run(Do1) c.Run(Do2) if err := c.WaitAllDone(); err != nil { t.Log(err) } }
Eventually t.log(err)
outputs err1
or err2
, but I want it to output err1 err2
. Why does it miss an error.
This is because (*cycle).waitalldone
will not wait for the goroutine collecting errors to complete. If you run your code with the -race
flag, sometimes it may report several data race errors. This is one of them:
$ go test -race . ================== warning: data race write at 0x00c0000a0610 by goroutine 10: m.(*cycle).waitalldone.func1() /home/zeke/src/temp/76370962/main_test.go:40 +0xb6 previous read at 0x00c0000a0610 by goroutine 7: m.(*cycle).waitalldone() /home/zeke/src/temp/76370962/main_test.go:48 +0x14e m.test41() /home/zeke/src/temp/76370962/main_test.go:63 +0xa4 testing.trunner() /snap/go/current/src/testing/testing.go:1576 +0x216 testing.(*t).run.func1() /snap/go/current/src/testing/testing.go:1629 +0x47
This change will resolve the issue:
func (c *cycle) waitalldone() error { var err error + done := make(chan int) go func() { for { if tmperr, ok := <-c.errchan; ok { err = multierr.append(err, tmperr) } else { break } } + close(done) }() c.wg.wait() close(c.errchan) + <-done return err }
And you can use the range
clause to simplify the for loop:
func (c *Cycle) WaitAllDone() error { var err error done := make(chan int) go func() { for tmpErr := range c.errChan { err = multierr.Append(err, tmpErr) } close(done) }() c.wg.Wait() close(c.errChan) <-done return err }
The above is the detailed content of Why am I only getting some errors and not all errors from the goroutine I started?. For more information, please follow other related articles on the PHP Chinese website!