Hi there,
I've just upgraded our live555 version (from 2011.06.12 to 2012.05.17)
and RTSPClient::connectToServer is failing on Windows. The following
check fails:
__int const err = envir().getErrno();
if (err == EINPROGRESS || err == EWOULDBLOCK) {
...
}
I read Ross Finlayson's response to a previous post
<http://lists.live555.com/pipermail/live-devel/2012-January/014432.html>
by Jeff Shanab, where he added EWOULDBLOCK as a platform-independent
solution. However, it turns out EWOULDBLOCK != WSAEWOULDBLOCK.
_winerror.h_
#define WSAEWOULDBLOCK 10035L
_errno.h_
#define EWOULDBLOCK 140
_winsock.h / winsock2.h_
#if 0
#define EWOULDBLOCK WSAEWOULDBLOCK
See this feedback entry
<http://connect.microsoft.com/VisualStudio/feedback/details/509380/errno-h-socket-return-codes-now-inconsistent-with-wsagetlasterror>
and MSDN Error Codes
<http://msdn.microsoft.com/en-us/library/windows/desktop/ms737828%28v=vs.85%29.aspx>.
Seems like in VS2010 they broke compatibility with Berkeley? The #if 0
is a bit of a lol. :)
There used to be a workaround in NetCommon.h ln 38, however it is now
commented out in 2012.05.17:
//#define EWOULDBLOCK WSAEWOULDBLOCK
This results in Windows builds being unable to connect.As a temporary
workaround, I've changed my NetCommon.h to read:
#undef EWOULDBLOCK
#define EWOULDBLOCK WSAEWOULDBLOCK
#undef EINPROGRESS
#define EINPROGRESS WSAEWOULDBLOCK
#undef EAGAIN
#define EAGAIN WSAEWOULDBLOCK
#undef EINTR
#define EINTR WSAEINTR
This appears to work again. Is this solution correct?
Thanks for your help,
Scott Taylor
immersaview.com
<http://lists.live555.com/pipermail/live-devel/2012-January/014432.html>
_______________________________________________
live-devel mailing list
live-devel@lists.live555.com
http://lists.live555.com/mailman/listinfo/live-devel
