Reputation: 8361
What's the best way to broadcast values from a stream to two network destination simultaneously? Here is simplified code:
func main() {
resp, _ := http.Get("http://origin.com/image.jpeg")
var buf bytes.Buffer
// tee, when read, writes to &buf
respBodyTee := io.TeeReader(resp.Body, &buf)
sendToClient(respBodyTee)
uploadeToServer(&buf)
}
A stream cannot be read twice, so TeeReader
is used to populate &buf
with whatever is read from respo.Body
.
However, the functions (sendToClient
and uploadToServer
) will run synchronously while I'd like to them to make their work concurrently.
Solution that I have on mind is to pass a channel to sendToClient
that will populate channel with bytes already sent to client. Later have uploadToServer
read from the same channel. Something along these lines:
func main() {
resp, _ := http.Get("http://origin.com/image.jpeg")
ch := make(chan byte)
go sendToClient(respBodyTee, ch) // pass 'ch' for writing and run in a goroutine
uploadeToServer(ch) // will read from 'ch' (synchronous)
}
I'm new to Go and am not sure if the above is the right direction.
Upvotes: 2
Views: 650
Reputation: 1681
in your scenario, it's better to have 2 independent byte streams for 2 network calls. If they rely on one source stream, when sendToClient
stalls uploadeToServer
will hang. channel wouldn't solve the problem above but introducing locking overhead.
you can try io.MultiWriter
to make 2 independent byte streams
var buf1, buf2 bytes.Buffer
mw := io.MultiWriter(&buf1, &buf2)
if _, err := io.Copy(mw, r.Body); err != nil {
...
}
go sendToClient(buf1)
go uploadeToServer(buf2)
...
Upvotes: 1