Re: [lmi] wx-2.7, MinGW gcc-3.4.5, inline + dllimport

[Greg Chicares]

On 2006-4-2 12:58 UTC, Vadim Zeitlin wrote:
> On Sun, 02 Apr 2006 02:09:37 +0000 Greg Chicares <address@hidden> wrote:
> 
> ...
> GC> Vadim, is there any disadvantage to removing WXDLLEXPORT from these
> GC> three inline functions?
> 
>  The problem is that while dllexport/import declaration is clearly
> unnecessary when the function is indeed inlined, we can't know for sure if
> this is the case. Of course, we do expect the compiler to be able to inline
> an empty function, but what about the other ones, e.g. this one:
> 
> GC> wxWidgets/include/wx/dnd.h
> GC> 
> GC>   48  inline WXDLLEXPORT bool wxIsDragResultOk(wxDragResult res)
> GC>   49  {
> GC>   50      return res == wxDragCopy || res == wxDragMove || res == 
> wxDragLink;
> GC>   51  }
> 
>  Will it be inlined or not? If not, then we do need a WXDLLEXPORT here as
> the compiler would put the body of the function in the DLL and you must use
> dllimport to get it from there.
> 
>  I also wonder about what happens when inlining is disabled (as is usually
> the case in debug builds). All in all I'm almost certain that removing
> WXDLLEXPORT from here will break some DLL builds, unfortunately.

Reported:
  http://sourceforge.net/mailarchive/message.php?msg_id=15297302
So far at least, I've seen the unwanted diagnostic only with
MinGW gcc-3.4.5, and I don't know of any reason why we'd need to
upgrade to that compiler version. We're using 3.4.2 in production.