Re: [lmi] Possible wx-2.9.0-RC6 regression

[Vadim Zeitlin]

On Mon, 20 Jul 2009 14:51:08 +0000 Greg Chicares <address@hidden> wrote:

GC> Here's a vector of an incomplete type [see C++03, last
GC> line of 17.4.3.6/2]:
GC> 
GC> /* clipbrd.h */
GC> 24  class WXDLLIMPEXP_FWD_CORE wxDataFormat;
GC> 117     wxVector<wxDataFormat> m_formats;

 Indeed, this doesn't risk to work with std::vector<>, of course. I somehow
only though our own wxVector implementation when writing this code :-(

GC> I suppose this should be fixed by adding
GC>   #include "wx/dataobj.h"
GC> to 'clipbrd.h', replacing these two lines:
GC> 24 class WXDLLIMPEXP_FWD_CORE wxDataFormat;
GC> 25 class WXDLLIMPEXP_FWD_CORE wxDataObject;

 I remember that I didn't want to introduce dependency of this header on
wx/dataobj.h (and there also comments to this effect in
src/common/clipcmn.cpp) but now that I think about it again I wonder if
it's really such a big deal as it doesn't seem possible to use wxClipboard 
without including this header anyhow. So I went ahead and did just this in
http://trac.wxwidgets.org/changeset/61485.

 Thanks for reporting this problem!
VZ