[Mingw-users] Re: Re: MinGW GLib

[Soren A <sor...@fa...>]

Danny Smith <dan...@ya...> wrote around 27 Sep 2002
news:200...@we...: 

>> @@ -1625,7 +1660,7 @@
>>    *-*-cygwin*)
>>       G_LIBS_EXTRA="-luser32 -lwsock32 -lkernel32"
>>      ;;
>> -  *-*-mingw*)
>> +  *mingw*)
>>       G_LIBS_EXTRA="-lwsock32"
>>      ;;
>>    *)
> 
> 
> Please be aware that unless <winsock.h> is included before <windows.h>
> then winsock2.h will be included by default.  With winsock2.h
> interface, you should link in libws2_32.a., not libwsock32.a, else you
> can have some unpleasant surprises.  
> 
> Why is winsock2.h the default in mingw's w32api?  Because MSDN docs no
> longer talk much about winsock 1.1 interface.

Thanks for looking over the patch, Danny. Hopefully Tor will read this. I 
only changed the 'case' statement simple regex to *mingw* from *-*-mingw* 
(there was a recent posting about this in another thread that gave me the 
impression that this was a good idea, although the way autoconf presently 
works, it shouldn't matter practically speaking). I didn't look into any 
library issues aside from checking that "libiconv" and "libintl" could be 
successfully detected by 'configure' on my system once I downloaded and 
installed them.

So the winsock issue is completely outside my awareness when I made the 
patch, and based on your advice what needs to be done is combing thru the 
sources checking for order of inclusion. Is that right? And maybe configure 
needs to try to find libws2_32.a and there should be something in config.h 
that #defines HAVE_WINSOCK2 or like that.

  Best,
   Soren A