Re: include: Only include winuser.h in oleidl.idl as a replacement for windows.h when compiling the Wine source.

[Jacek Caban]

Hi Francois,

On 12/28/11 10:39, Francois Gouget wrote:
> -cpp_quote("#include<winuser.h>")
> +cpp_quote("#ifdef __WINESRC__")
> +cpp_quote("# include<winuser.h>")
> +cpp_quote("#endif")
>
> /*****************************************************************************
>    * IOleTypes interface

How about entirely removing that include? We can always include it in C 
files that need it.

Jacek

include: Only include winuser.h in oleidl.idl as a replacement for windows.h when compiling the Wine source.

[Francois Gouget]

On Wed, 28 Dec 2011, Jacek Caban wrote:
> Hi Francois,
>
> On 12/28/11 10:39, Francois Gouget wrote:
> > -cpp_quote("#include<winuser.h>")
> > +cpp_quote("#ifdef __WINESRC__")
> > +cpp_quote("# include<winuser.h>")
> > +cpp_quote("#endif")
> >
> > /*****************************************************************************
> >    * IOleTypes interface
>
> How about entirely removing that include? We can always include it in C files
> that need it.

Substituting an alternative header for windows.h is an approach that is 
used in a few other places, specifically in dshow.h, msctf.h, pdh.h, 
rpc.h, snmp.h and winsock.h. So I think it's ok to follow the pattern 
here, it should just be a bit clearer.

But if the consensus is that it (and the others?) should be removed I 
can do that too. The patch will be quite a bit bigger though.

-- 
Francois Gouget <fgouget(a)free.fr>              http://fgouget.free.fr/
                     Avoid the Gates of Hell - use Linux.

include: Only include winuser.h in oleidl.idl as a replacement for windows.h when compiling the Wine source.

[Alexandre Julliard]

Francois Gouget <fgouget(a)free.fr> writes:
> Substituting an alternative header for windows.h is an approach that is 
> used in a few other places, specifically in dshow.h, msctf.h, pdh.h, 
> rpc.h, snmp.h and winsock.h. So I think it's ok to follow the pattern 
> here, it should just be a bit clearer.
>
> But if the consensus is that it (and the others?) should be removed I 
> can do that too. The patch will be quite a bit bigger though.

Maybe it's time to reconsider the windows.h prohibition. Our headers are
pretty stable now so we don't need full rebuilds very often; and if we
define WIN32_LEAN_AND_MEAN, avoiding windows.h probably doesn't gain us
all that much anymore.

-- 
Alexandre Julliard
julliard(a)winehq.org