Richard van der Hoff wrote:
> Anders Broman wrote:
>> Hi,
>> The buildbot is failing. I'm not sure which of the recent changes that
>> caused this.
>
> Indeed. I think it was r21330, by Sebastien:
>
> http://anonsvn.wireshark.org/viewvc/viewvc.py?view=rev&revision=21330
>
> Specifically, this bit:
>
> http://anonsvn.wireshark.org/viewvc/viewvc.py/trunk/epan/nstime.h?r1=21330&r2=21329&pathrev=21330
>
> I think it's actually failing to build epan.c, which includes proto.h,
> which includes frame_data.h, which includes nstime.h, which now includes
> wtap.h, which includes winsock2.h, which has conflicting definitions
> with the winsock.h dragged in from somewhere else. At least, that's my
> best guess, currently.
>
> The question really is where the #include of winsock.h is coming from.
> I'll have to defer to someone with Win32 knowledge on that...
It's coming from the include of windows.h. Moving the #ifdef _WIN32
block down to below the other includes fixes it as including winsock2.h
first will stop the winsock.h attempting to redefine things.
Another similar occurrence is in gui_utils.c. I've checked in fixes in
r21335.
--
Regards,
Graham Bloice