README
¶
taskgroup
A *taskgroup.Group represents a group of goroutines working on related tasks.
New tasks can be added to the group at will, and the caller can wait until all
tasks are complete. Errors are automatically collected and delivered to a
user-provided callback in a single goroutine. This does not replace the full
generality of Go's built-in features, but it simplifies some of the plumbing
for common concurrent tasks.
Here is a working example in the Go Playground.
Rationale
Go provides excellent and powerful concurrency primitives, in the form of goroutines, channels, select, and the standard library's sync package. In some common situations, however, managing goroutine lifetimes can become unwieldy using only what is built in.
For example, consider the case of copying a large directory tree: Walk through the source directory recursively, creating a parallel target directory structure and starting a goroutine to copy each of the files concurrently. That much is conceptually simple:
func copyTree(source, target string) error {
err := filepath.Walk(source, func(path string, fi os.FileInfo, err error) error {
adjusted := adjustPath(path)
if fi.IsDir() {
return os.MkdirAll(adjusted, 0755)
}
go copyFile(adjusted, target)
return nil
})
if err != nil {
// ... clean up the output directory ...
}
return err
}
Unfortunately, it isn't quite sufficient: How will we detect when all the
file copies are finished? Typically we will use a sync.WaitGroup:
var wg sync.WaitGroup
...
wg.Add(1)
go func() {
defer wg.Done()
copyFile(adjusted, target)
}()
...
wg.Wait() // blocks until all the tasks signal done
In addition, we need to cope with errors: Copies might fail (the disk might fill up, or there might be a permissions error, say). For some applications it might suffice to log the error and continue, but often in case of error you'd like to back out and clean up your mess. To do this, we need to capture the return value from the function inside the goroutine―and that will require us either to add a lock or plumb in another channel:
errs := make(chan error)
...
go copyFile(adjusted, target, errs)
Since multiple copies can be running in parallel, we will also need another goroutine to drain the errors channel and accumulate the results somewhere:
var failures []error
go func() {
for e := range errs {
failures = append(failures, e)
}
}()
...
wg.Wait()
close(errs)
But now we need to also detect when the error collector is finished, so that
we can examine the failures without a data race. We'll need another channel
or wait group to signal for this:
var failures []error
edone := make(chan struct{})
go func() {
defer close(edone)
for e := range errs {
failures = append(failures, e)
}
}()
...
wg.Wait() // all the workers are done
close(errs) // signal the error collector to stop
<-edone // wait for the error collector to be done
All of this is tedious, but works fine. Now, however, if something fails we
really don't want to wait around for all the copies to finish―if one of the
file copies fails, we want to stop what you're doing and clean up. So now we
need a way to signal cancellation, such as a context.Context:
ctx, cancel := context.WithCancel(context.Background())
defer cancel()
...
copyFile(ctx, adjusted, target, errs)
and then copyFile will have to check for that:
func copyFile(ctx context.Context, source, target string, errs chan<- error) {
if ctx.Err() != nil {
return
}
// ... do the copy as normal, or propagate an error
}
Finally, we need some way to limit the number of concurrent copies. Even if the host has plenty of memory and CPU, unbounded concurrency is likely to run us out of file descriptors. As a rule of thumb, unbounded concurrency is almost always a poor choice. We might use a semaphore or a throttling channel:
throttle := make(chan struct{}, 64) // allow up to 64 concurrent copies
go func() {
throttle <- struct{}{} // block until the throttle has a free slot
defer func() { wg.Done(); <-throttle }()
copyFile(ctx, adjusted, target, errs)
}()
In case you've lost count, we're up to four channels (including the context).
The point here is that, while Go's concurrency primitives are more than powerful enough to express these relationships, it can be tedious to wire them all together.
The taskgroup package exists to handle the plumbing for the common case of a
group of tasks that are all working on a related outcome (e.g., copying a
directory structure), and where an error on the part of any single task may
be grounds for terminating the work as a whole.
The package provides a taskgroup.Group type that has built-in support for
some of these concerns:
- Limiting the number of active goroutines.
- Collecting and filtering errors.
- Waiting for completion and delivering status.
A taskgroup.Group collects error values from each task and can deliver them
to a user-provided callback. The callback can filter them or take other actions
(such as cancellation). Invocations of the callback are all done from a single
goroutine so it is safe to manipulate local resources without a lock.
A group does not directly support cancellation, but integrates cleanly with the
standard context package. A context.CancelFunc
can be used as a trigger to signal the whole group when an error occurs.
The API is straightforward: A task is expressed as a func() error, and is
added to a group using the Go method,
g := taskgroup.New(nil).Go(myTask)
Any number of tasks may be added, and it is safe to do so from multiple goroutines concurrently. To wait for the tasks to finish, use:
err := g.Wait()
This blocks until all the tasks in the group have returned (either
successfully, or with an error). Wait returns the first non-nil error
returned by any of the worker tasks.
An example program can be found in examples/copytree/copytree.go.
Filtering Errors
The taskgroup.New function takes an optional callback that is invoked for
each non-nil error reported by a task in the group. The callback may choose
to propagate, replace, or discard the error. For example:
g := taskgroup.New(func(err error) error {
if os.IsNotExist(err) {
return nil // ignore files that do not exist
}
return err
})
Controlling Concurrency
The Limit method supports limiting the number of concurrently active
goroutines in the group. For example:
// Allow at most 25 concurrently-active goroutines in the group.
g, start := taskgroup.New(nil).Limit(25)
// Start tasks by calling the function returned by taskgroup.Limit:
start(task1)
start(task2)
// ...
Solo Tasks
In some cases it is useful to have a solo background task to handle a separate concern, such as collecting the results from a batch of concurrent workers.
For example, suppose we have a group of tasks processing a complex search, and we want to aggregate the results. We can't do this from the same group that the workers are using, since it needs to run until they have all completed. On the other hand, creating two full groups is overkill, since we only need one additional goroutine.
To handle such cases, the Single constructor is helpful: It manages a single
background goroutine with a separate wait:
var total int
results := make(chan int)
s := taskgroup.Single(func() error {
for v := range results {
total += v
}
})
g := taskgroup.New(nil)
for i := 0; i < numTasks; i++ {
batch := newBatch(i)
g.Go(func() error {
return search(batch)
})
}
// Wait for the workers to finish.
g.Wait()
// Signal the collector to stop, and wait for it to do so.
close(results)
s.Wait()
// Now it is safe to use the results.
fmt.Println(total)
Documentation
¶
Overview ¶
Package taskgroup manages collections of cooperating goroutines. It defines a Group that handles waiting for goroutine termination and the propagation of error values. The caller may provide a callback to filter and respond to task errors.
Index ¶
Examples ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
This section is empty.
Types ¶
type ErrorFunc ¶
An ErrorFunc is called by a group each time a task reports an error. Its return value replaces the reported error, so the ErrorFunc can filter or suppress errors by modifying or discarding the input error.
func Listen ¶
Listen creates an ErrorFunc that reports each non-nil task error to f. The resulting ErrorFunc returns task errors unmodified.
Example ¶
package main
import (
"errors"
"fmt"
"github.com/creachadair/taskgroup"
)
func main() {
g := taskgroup.New(taskgroup.Listen(func(e error) {
fmt.Println(e)
}))
g.Go(func() error { return errors.New("heard you") })
fmt.Println(g.Wait()) // the error was preserved
}
Output: heard you heard you
func Trigger ¶
func Trigger(f func()) ErrorFunc
Trigger creates an ErrorFunc that calls f each time a task reports an error. The resulting ErrorFunc returns task errors unmodified.
Example ¶
package main
import (
"context"
"fmt"
"log"
"time"
"github.com/creachadair/taskgroup"
)
func main() {
ctx, cancel := context.WithCancel(context.Background())
defer cancel()
const badTask = 5
// Construct a group in which any task error cancels the context.
g := taskgroup.New(taskgroup.Trigger(cancel))
for i := 0; i < 10; i++ {
i := i
g.Go(func() error {
if i == badTask {
return fmt.Errorf("task %d failed", i)
}
select {
case <-ctx.Done():
return ctx.Err()
case <-time.After(10 * time.Second):
return nil
}
})
}
if err := g.Wait(); err == nil {
log.Fatal("I expected an error here")
} else {
fmt.Println(err.Error())
}
}
Output: task 5 failed
type Group ¶
type Group struct {
// contains filtered or unexported fields
}
A Group manages a collection of cooperating goroutines. New tasks are added to the group with the Go method. Call the Wait method to wait for the tasks to complete.
The group collects any errors returned by the tasks in the group. The first non-nil error reported by any task (and not otherwise filtered) is returned from the Wait method.
Example ¶
package main
import (
"fmt"
"github.com/creachadair/taskgroup"
)
func main() {
msg := make(chan string)
g := taskgroup.New(nil)
g.Go(func() error {
msg <- "ping"
fmt.Println(<-msg)
return nil
})
g.Go(func() error {
fmt.Println(<-msg)
msg <- "pong"
return nil
})
g.Wait()
fmt.Println("<done>")
}
Output: ping pong <done>
func New ¶
New constructs a new empty group. If ef != nil, it is called for each error reported by a task running in the group. The value returned by ef replaces the task's error. If ef == nil, errors are not filtered.
Calls to ef are issued by a single goroutine, so it is safe for ef to manipulate local data structures without additional locking.
func (*Group) Limit ¶
Limit returns g and a function that starts each task passed to it in g, allowing no more than n tasks to be active concurrently. If n ≤ 0, the start function is equivalent to g.Go, which enforces no limit.
The limiting mechanism is optional, and the underlying group is not restricted. A call to the start function will block until a slot is available, but calling g.Go directly will add a task unconditionally and will not take up a limiter slot.
Example ¶
var p peakValue
g, start := taskgroup.New(nil).Limit(4)
for i := 0; i < 100; i++ {
start(func() error {
p.inc()
defer p.dec()
time.Sleep(1 * time.Microsecond)
return nil
})
}
g.Wait()
fmt.Printf("Max active ≤ 4: %v\n", p.max <= 4)
Output: Max active ≤ 4: true
func (*Group) Wait ¶
Wait blocks until all the goroutines currently active in the group have returned, and all reported errors have been delivered to the callback. Wait returns the first non-nil error returned by any of the goroutines in the group and not filtered by an ErrorFunc.
After a call to Wait returns, the group is ready for reuse.
type Solo ¶ added in v0.4.0
type Solo struct {
// contains filtered or unexported fields
}
A Solo manages a single background goroutine. The task is started when the value is first created, and the caller can use the Wait method to block until it has exited.
Example ¶
package main
import (
"fmt"
"math/rand"
"github.com/creachadair/taskgroup"
)
func shuffled(n int) []int {
rand.Seed(1)
vs := make([]int, n)
for i := range vs {
vs[i] = i + 1
}
for i := n - 1; i > 0; i-- {
p := rand.Intn(i + 1)
vs[i], vs[p] = vs[p], vs[i]
}
return vs
}
func main() {
var total int
results := make(chan int)
// Start a task to collect the results of a "search" process.
s := taskgroup.Single(func() error {
for v := range results {
total += v
}
return nil
})
const numTasks = 25
input := shuffled(500)
// Start a bunch of tasks to find elements in the input...
g := taskgroup.New(nil)
for i := 0; i < numTasks; i++ {
target := i + 1
g.Go(func() error {
for i, v := range input {
if v == target {
results <- i
break
}
}
return nil
})
}
// Wait for the searchers to finish.
g.Wait()
// Signal the collector to stop, and wait for it to do so.
close(results)
s.Wait()
// Now it is safe to use the results.
fmt.Println(total)
}
Output: 5972
Source Files
Directories