make wxArtProvider pure virtual (I was not so on

[wxWidgets.git] / wxPython / CHANGES.txt

diff --git a/wxPython/CHANGES.txt b/wxPython/CHANGES.txt
index 5d91cdb7c2c9069dbd2589383b2bbd5ecf6b1f80..157dcec7f6da343aac55705f480540e7479e51fe 100644 (file)
--- a/wxPython/CHANGES.txt
+++ b/wxPython/CHANGES.txt
@@ -22,11 +22,67 @@ Changed the img2py tool to use PNG instead of XPM for embedding image
 data in Python source code, and the generated code now uses streams to
 convert the image data to wxImage, wxBitmap, or wxIcon.
 
-Added wxPython.lib.rcsizer which contains RowColSizer.  This sizer is
-based on code from Niki Spahiev and lets you specify a row and column
-for each item, as well as optional column or row spanning.  Cells with
-no item assigned to it are just left blank.  Stretchable rows or
-columns are specified and work the same as in wxFlexGridSizer.
+Added the wxPython.lib.rcsizer module which contains RowColSizer.
+This sizer is based on code from Niki Spahiev and lets you specify a
+row and column for each item, as well as optional column or row
+spanning.  Cells with no item assigned to it are just left blank.
+Stretchable rows or columns are specified and work the same as in
+wxFlexGridSizer.
+
+Updated XRCed from Roman Rolinsky
+
+Added wxBufferedDC.
+
+Upgraded wxSTC from Scintilla 1.40 to Scintilla 1.45
+
+UNICODE!
+
+    wxWindows/wxPython can be compiled with unicode support enabled or
+    disabled.  Previous to wxPython 2.3.3 non-unicode mode was always
+    used.  Starting with 2.3.3 either mode is supported, but only if
+    it is also available in wxWindows on the platform.  Currently
+    wxWindows only supports unicode on MS Windows platforms, but with
+    the recent release of GTK+ 2.0 it is only a matter of time until
+    it can be done on wxGTK (Linux and other unixes) as well.
+
+    Unicode works best on platforms in the NT branch of the Windows
+    family tree (NT, win2k, XP) but it is now also possible to use the
+    same unicode binaries on win95/98/ME platforms as well!  This is
+    done by using a special library and DLL with the application
+    called MSLU, (Microsoft Layer for Unicode).  It simply gets out of
+    the way if the app is run on an NT box, otherwise if run on a
+    win9x box it loads a special DLL that provides the unicode
+    versions of the windows API.  So far I have not been able to get
+    this to work perfectly on win9x.  Most things work fine but
+    wxTaskBarIcon for example will cause a crash if used with the
+    unicode build on win95.
+
+    So how do you use it?  It's very simple.  When unicode is enabled,
+    then all functions and methods in wxPython that return a wxString
+    from the C++ function will return a Python unicode object, and
+    parameters to C++ functions/methods that expect a wxString can
+    accept either a Python string or unicode object.  If a string
+    object is passed then it will be decoded into unicode using the
+    converter pointed to by wxConvCurrent, which will use the default
+    system encoding.  If you need to use a string in some other
+    encoding then you should convert it to unicode using the Python
+    codecs first and then pass the unicode string to the wxPython
+    method.
+
+Added wxListCtrlAutoWidthMixin from Erik Westra.
+
+Added wxIconBundle and wxTopLevelWindow.SetIcons.
+
+Added wxLocale and wxEncodingConverter.
+
+A little black magic...  When the C++ object (for a window or
+whatever) is deleted there is no way to force the Python shadow object
+to also be destroyed and clean up all references to it.  This leads to
+crashes if the shadow object tries to call a method with the old C++
+pointer.  The black magic I've done is to replace the __class__ in the
+Python instance object with a class that raises an exception whenever
+a method call (or other attribute access) is attempted.  This works
+for any class that is OOR aware.