This is the mail archive of the
cygwin@cygwin.com
mailing list for the Cygwin project.
Re: cvs broken on win2k after recent cygwin upgrade: "The descriptor is a file, not a socket"
- To: cygwin <cygwin at cygwin dot com>
- Subject: Re: cvs broken on win2k after recent cygwin upgrade: "The descriptor is a file, not a socket"
- From: Corinna Vinschen <cygwin at cygwin dot com>
- Date: Wed, 2 May 2001 13:23:49 +0200
- References: <f5bofu2cm9s.fsf@cogsci.ed.ac.uk> <07fd01c0c3dc$cf05f730$0200a8c0@voyager> <f5bhezs4e2c.fsf@cogsci.ed.ac.uk> <20010414161237.A4149051@MELON> <f5b8zl23731.fsf@cogsci.ed.ac.uk> <20010415090728.B8359@redhat.com> <20010415184450.V20490@cygbert.vinschen.de> <20010415153631.D9015@redhat.com> <20010415225227.W20490@cygbert.vinschen.de> <3AEE22F0.2E034CE@ece.gatech.edu>
On Mon, Apr 30, 2001 at 10:44:00PM -0400, Charles S. Wilson wrote:
> Corinna Vinschen wrote:
> > > >That's not quite correct. Cygwin uses ws2_32.dll if available.
> > > >The reason is the Winsock2 functionality used for duplicating
> > > >sockets. The old method is still implemented as a fallback, though.
> > >
> > > How is that implemented? I don't see it in the code. Where is
> > > that implemented below?
> >
> > It's not that obvious. Since WsaStartup is called with a requested
> > version of 2.2 in wsa_init(), the wsock32 code asks if winsock 2 is
> > available on the system. If so, it will be used. The reason for
> > loading ws2_32 explicitly in Cygwin is that the function pointers of
> > the special winsock 2 functions couldn't be resolved otherwise.
> >
> > > Also the two tests for ws32_handle and wsock32_handle look reversed.
> > >
> [snip]
> >
> > It is. WsaStartup should only be called once regardless whether the
> > first winsock function is a version 1.1 or version 2 function. The
> > wsa_init function is only called if it's not already called from the
> > other entry point.
>
> I'm confused. Corinna, does your statement mean that the code in
> autoload.cc is correct as it is now (which is the same as it was at the
> time this email was written) --- or does autoload.cc need to be
> "corrected" to unreverse the ws32_handle and wsock32_handle tests?
> (I'm going through my cygwin inbox trying to tie up some open threads
> w.r.t "my" packages)
Sorry for being unclear. What I meant was, it is correct as it is.
That's what I meant to say but actually, it's not really correct.
Unlike the other LoadDLLxxx functions a ipreliminary check for
handle != NULL is missing. I will add it now.
Corinna
--
Corinna Vinschen Please, send mails regarding Cygwin to
Cygwin Developer mailto:cygwin@cygwin.com
Red Hat, Inc.
--
Want to unsubscribe from this list?
Check out: http://cygwin.com/ml/#unsubscribe-simple