file descriptor leak about cocoaasyncsocket HOT 6 CLOSED

I forgot to mention that I am still non arc. 7.0.X

from cocoaasyncsocket.

I noticed there was an update that closes the file descriptors in closeWithError,
but I don't know if they cover the same path.

I believe it does. I tested it, and the file descriptor is getting closed. I added some extra logs just now for testing.

Am I missing something? Are you sure the file descriptor isn't getting closed?

from cocoaasyncsocket.

I a sure the file descriptor was not getting closed, but that was before we pulled Kyle Fuller's fix:

Fix an issue where a refused connection will not be cleaned up properly. 5c28a2a/

The particulars, as best I understand it, happened when we tried to connect to a host using GCDAsyncSocketMsgController:

• (id)initWithDelegate: ( id ) inDelegate
  queue: ( dispatch_queue_t ) inDispatchQueue
  interface: ( NSString * ) inInterface
  host: ( NSString * ) inHost
  port: ( UInt16 ) inPort ;

When the host wasn't on the interface specified, or was explicity rejected, the connection would fail, causing the file descriptor leak.

Please let me know particularly if our patch will cause problems, or if it's safe for us to remove our patch because Kyle's fix covers the same problem in a better place.

Ken

On May 14, 2012, at 3:05 AM, Robbie Hanson wrote:

I noticed there was an update that closes the file descriptors in closeWithError,
but I don't know if they cover the same path.

I believe it does. I tested it, and the file descriptor is getting closed. I added some extra logs just now for testing.

Am I missing something? Are you sure the file descriptor isn't getting closed?

---

Reply to this email directly or view it on GitHub:
#52 (comment)

from cocoaasyncsocket.

You meant this fix right: 6f46421

Yes, this fix should also cover the same problem.

from cocoaasyncsocket.

sorry, i copied the parent hash.

Yes, that is the fix I was referring to.

On May 14, 2012, at 11:43 AM, Robbie Hanson wrote:

You meant this fix right: 6f46421

Yes, this fix should also cover the same problem.

---

Reply to this email directly or view it on GitHub:
#52 (comment)

from cocoaasyncsocket.

Cool. Marking the issue as closed. But let me know if you see any further problems with this, and I reopen the issue.

from cocoaasyncsocket.