The good news is that I've succeeded in replacing the Windows-specific
code in the beta version with platform-independent GTK code. This is
what Jonathan Brandmeyer started with originally but abandoned when it
seemed very difficult for developers (including me) to install GTK on
Windows. In the interim installing GTK and GTKMM has gotten easy, and
there are of course advantages to being able to debug and use the same
code on all platforms. (The relevant GTK binaries can be packaged with
Visual, so it needn't be necessary for end users to install these
materials.)
I also hoped that this would be sufficiently different to shake loose
some of the serious bugs that the beta version has on Windows. The bad
news is that it seems to have made no difference. The good news is that
this eliminates the possibility that the problems lay in the
Windows-specific code. One area that is attracting particular suspicion
is the new rendering scheme for curves, which was changed in the beta
version but which I have reason to think is causing problems on Windows.
I have not checked in my changes to CVS. Basically they consist of
eliminating from src/win32 all files but winrate.cpp and wintimer.cpp,
commenting out most conditional includes in various files that were
based on the compiler flag _WIN32, and making some minor adjustments in
src/Makefile.in to the lists of .lo files for Linux and Windows.
Bruce Sherwood
|
