Cleanup goroutine structure of autonat #41
Conversation
willscott
force-pushed
the
feat/thread-cleanup
branch
from
88c84f7 to
bb33600
Compare
|
pausing here for sanity check review of structure. If this organization seems plausible, there are a couple additional tests that should go with it. |
* incoming connections post a channel event - fix libp2p#40 * inbound connections reduce the frequency of probes - address libp2p#35 waiting on libp2p/go-libp2p#747 for detecting local address changes
willscott
force-pushed
the
feat/thread-cleanup
branch
from
bb33600 to
1f8ed1d
Compare
…nat into feat/thread-cleanup
|
@aarshkshah1992 I merged your #51 into this branch in fe1f5df. A sanity check review before I merge this would be great :) |
|
@willscott Will take a look first thing tomorrow. |
| delay = AutoNATRetryInterval | ||
| } | ||
| timer := time.NewTimer(delay) | ||
| timerRunning := true |
There was a problem hiding this comment.
defer timer.Stop ?
| private int | ||
| public int | ||
| pubaddr ma.Multiaddr | ||
| if timerRunning && !timer.Stop() { |
There was a problem hiding this comment.
From the timer docs, just this should be fine:
if !t.Stop() {
<-t.C
}
We can then get rid of the timerRunning variable.
There was a problem hiding this comment.
The reason for timerRunning is that if the timer fires above, then the channel will have already been drained. t.Stop() on a fired timer will return false, but in that code path we don't want to attempt to drain t.C again.
…ibp2p/go-libp2p-0.5.2 Bump github.com/libp2p/go-libp2p from 0.5.1 to 0.5.2
waiting on libp2p/go-libp2p#747 for detecting local address changes