Re: PATCH: glib iochannel win32 bug fix

["Andrew J. Lanoix" <alanoix umich edu>]

Peter you are right, that doesn't belong there. I made the change in CVS.

*giowin32.c: g_source_remove()ing an socket iochannel closes
	the socket when it should not. Patch by Peter Zelezny
	<zed linux com>
*giowin32.c: Fix a few typos

Owen, perhaps there should be a win32 "production" branch of glib 1.3 going 
back a bit. Tor created one for gtk-1.3 a while back. The reason why I say 
this is that Tor is unfortunately still offline after his move and he has 
not done any release in while. I have been telling people to keep up with 
CVS on windows but now all sorts of API changes are happing.

I also fixed some typos but the remaining errors have to do with the API 
changes.
Andy Lanoix

giowin32.c: In function `g_io_win32_sock_read':
giowin32.c:956: `G_IO_STATUS_INTR' undeclared (first use in this function)
giowin32.c:956: (Each undeclared identifier is reported only once
giowin32.c:956: for each function it appears in.)
giowin32.c: In function `g_io_win32_sock_write':
giowin32.c:1005: `G_IO_STATUS_INTR' undeclared (first use in this function)
giowin32.c: At top level:
giowin32.c:1090: warning: initialization from incompatible pointer type
giowin32.c:1091: warning: initialization from incompatible pointer type
giowin32.c:1101: warning: initialization from incompatible pointer type
giowin32.c:1102: warning: initialization from incompatible pointer type
giowin32.c:1112: warning: initialization from incompatible pointer type
giowin32.c:1113: warning: initialization from incompatible pointer type
make: *** [giowin32.o] Error 1

--On Monday, July 02, 2001 8:11 AM -0400 Owen Taylor <otaylor redhat com> 
wrote:

> Peter Zelezny <zed linux com> writes:
>
> > Here's a patch that makes the win32 implementation of GIOChannel's more
> > like the unix one. The problem was that after g_source_remove()ing, it
> > would close your socket. I don't know why it ever did this, the win32
> > maintainers might want to take a closer look at it.
>
> The IOChannel interfaces recently changed a lot, and as a consequence,
> the giowin32 code is in need of more extensive care at the moment.  It
> probably doesn't compile. If it does, it certainly doesn't work.
>
> If someone wants to take a look at this, and also check to make sure
> that we haven't made any unforgiveable portability mistakes in the
> interfaces that need to be fixed, that could be quite useful.
>
> (Also see the comments/questions about Win32 and line endings
> from me in:
>
>   http://mail.gnome.org/archives/gtk-devel-list/2001-July/msg00020.html)
>
> Regards,
>                                          Owen