From: Kenneth J. Davis (jeremyd_at_computer.org)
Date: Sat Nov 22 2003 - 22:06:57 EST
On Sun, 23 Nov 2003 12:25:24 +1100 Stephen Viles <sviles_abi_at_iinet.net.au> wrote:
...
SV> >In file included from d:/mingw/include/windows.h:55,
SV> > from ut_sleep.cpp:24:
SV> >d:/mingw/include/winuser.h:3411: type specifier omitted for parameter
SV> `
SV> > BLENDFUNCTION'
SV> >d:/mingw/include/winuser.h:3411: parse error before `*' token
SV> >make[4]: ***
...
SV> This is a known problem with MinGW Win32API 2.4:
SV> http://sourceforge.net/tracker/index.php?func=detail&aid=825438&group_i
SV> d=2435&atid=102435
...
SV> There's some mention of a possible fix in this bug report, but I
SV> haven't tried it.
...
SV> Cheers - Stephen.
That is the best fix at the moment. It's what I did (though I didn't
look it up on the forum) for tinderbox. I thought I mentioned it on the list,
but with working around the other 2.4/2.3 win32api differences I guess
I forgot. The #ifdef is what the VC headers do as well.
This is caused by the function prototype using a structure that is only
defined when some GDI includes are also included and our #defining
WIN32_LEAN_AND_MEAN with NOGDI before #including <windows.h> prevents this
inclusion. While we could work around this by including extra Windows
headers, this is clearly a broken header and best fixed by correcting the
header. But feel free to commit a workaround (probably just need to
remove the #define NOGDI in ut_sleep.cpp) to go along with rest of win32api
workarounds in our code.
Jeremy
This archive was generated by hypermail 2.1.4 : Sat Nov 22 2003 - 22:07:17 EST