From: Minero Aoki <aamine@...>
Date: 2002-04-06T17:48:15+09:00
Subject: Re: SOAP query

Hi,

  In mail "Re: SOAP query"
    Holden Glova <dsafari@xtra.co.nz> wrote:

> > > /usr/lib/ruby/1.6/net/protocol.rb:345:in `new': Operation already in
> > > progress - - "connect(2)" (RuntimeError)

> I'm still stumped over this, spent more hours today trying to remove the 
> problem which I am not sure is soap related, my code related or Ruby related 
> as you have implied may be possible above. If this behaviour was a sign of a 
> defect, does this imply that my version of Ruby may have a defect in the 
> socket area?

Does the attaching patch changes anything?
This is the patch for ext/socket/socket.c.

-- Minero Aoki

--- socket.c.org	Sat Apr  6 17:56:42 2002
+++ socket.c	Sat Apr  6 17:57:10 2002
@@ -711,9 +711,7 @@ ruby_connect(fd, sockaddr, len, socks)
 {
     int status;
     int mode;
-#if defined __CYGWIN__
     int wait_in_progress = -1;
-#endif
 
 #if defined(HAVE_FCNTL)
     mode = fcntl(fd, F_GETFL, 0);
@@ -748,15 +746,12 @@ ruby_connect(fd, sockaddr, len, socks)
 	      case EAGAIN:
 #ifdef EINPROGRESS
 	      case EINPROGRESS:
-#if defined __CYGWIN__
 	      case EALREADY:
 		wait_in_progress = 10;
 #endif
-#endif
 		thread_write_select(fd);
 		continue;
 
-#if defined __CYGWIN__
 	      case EINVAL:
 		if (wait_in_progress-- > 0) {
 		    struct timeval tv = {0, 100000};
@@ -764,7 +759,6 @@ ruby_connect(fd, sockaddr, len, socks)
 		    continue;
 		}
 		break;
-#endif
 
 #ifdef EISCONN
 	      case EISCONN: