Change made by external contributor Ibrar Ahmed (ibrar.ahmad@gmail.com), revi...

Change made by external contributor Ibrar Ahmed (ibrar.ahmad@gmail.com), reviewed by erikkay:
http://codereview.chromium.org/6577/show

ports listen_socket and telnet_server to POSIX

I had to make some changes to get this to work on Mac and to fix a few regressions it caused in Windows, so please take a fresh look at this diff Dan.

Ibrar, please take a look at the changes from your patch to mine (you may need to diff listen_socket_unittest.h with listen_socket_unittest.cc manually since I moved a bunch of code here).  Some were bugs that I should have caught in review, some were bugs that only got tickled on the Mac, others were just cleanup.  Comments welcome.

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=5153

[ibrar, 2008-11-08 18:20:26]

http://codereview.chromium.org/9260/diff/601/810
File net/base/listen_socket.cc (right):

http://codereview.chromium.org/9260/diff/601/810#newcode142
Line 142: // socket closed ignore in case of Windows
// in case of Windows OnObjectSignaled function is responsible to call Close()
but in POSIX we should call Close() here.

http://codereview.chromium.org/9260/diff/601/810#newcode168
Line 168: if (wait_state_ == WAITING_CLOSE)
I think we can set wait_state_ = WAITING_CLOSE without checking.

[Erik does not do reviews, 2008-11-10 21:40:08]

http://codereview.chromium.org/9260/diff/601/810
File net/base/listen_socket.cc (right):

http://codereview.chromium.org/9260/diff/601/810#newcode142
Line 142: // socket closed ignore in case of Windows
On 2008/11/08 18:20:26, ibrar wrote:
> // in case of Windows OnObjectSignaled function is responsible to call Close()
> but in POSIX we should call Close() here.

Good clarification here.  Done.

http://codereview.chromium.org/9260/diff/601/810#newcode168
Line 168: if (wait_state_ == WAITING_CLOSE)
On 2008/11/08 18:20:26, ibrar wrote:
> I think we can set wait_state_ = WAITING_CLOSE without checking.

The reason for this check is to avoid calling DidClose multiple times.  This was
happening in the POSIX unit tests (on the Mac at least), and was confusing the
unit tests.  This may not be happening any more since I moved the Close() call
into Read(), but I left this in here just in case.