Samuli,
Please peek this up.
Thanks.
On Mon, Sep 13, 2010 at 6:21 PM, Alon Bar-Lev <alon.bar...@gmail.com> wrote:
> Hello,
>
> Does anyone know why the socklet_t detection was added this way?
>
> I mean the TYPE_SOCKLEN_T m4 macro.
>
> First, it should be sufficient to just try to see if socklen_t is defined.
> Next, on all platforms it would be int in all other platforms.
> And, there is no need to check for the type in monolitic environment
> like Windows, as it will be always the same.
>
> Currently it fails cross compile windows in mingw-w64 compiler, as
> winsock.h is as follows:
> """
> /* define WINSOCK_API_LINKAGE and WSAAPI for less
> * diff output between winsock.h and winsock2.h, but
> * remember to undefine them at the end of file */
> """
> And the macro uses these macros which are internal winsock macros and
> should not be used anyway.
>
> Also, when did the autodefs.h went mandatory? Why is it in
> tap-win32/common.h while no constant is actually used?
>
> The use of WSA_IO_INCOMPLETE without including winsock2.h is invalid!
> Look at [1].
>
> Attached is a patch to solve all these issues for now. Apply to both 2.1, 2.2.
>
> The whole build environment of openvpn is malformed, using two
> separate build systems for windows and posix, invalid use of autoconf,
> and the mixture of different projects: openvpn, tap, windows
> installer, easy-rsa.
>
> Alon.
>
> [1] http://msdn.microsoft.com/en-us/library/aa921087.aspx
>
