We are currently migrating Bugzilla to GitHub issues.
Any changes made to the bug tracker now will be lost, so please do not post new bugs or make changes to them.
When we're done, all bug URLs will redirect to their equivalent location on the new bug tracker.

I first noticed this bug in Wesnoth, but after testing in my own program, I'm fairly certain it's an SDL bug. When you scroll up with a MacBook Pro trackpad (using two fingers), SDL generates some SDL_BUTTON_WHEELUP events, but a lot of SDL_BUTTON_WHEELDOWN events as well. It seems that if you scroll more quickly up, you get a higher ratio of up events to down ones, but you still get the down ones; if you scroll up slowly, it counts as a net down. To test this, I stuck this code in the main loop of a C++/SDL program of mine:
if (SDL_PollEvent(&event))
{
	if (event.type == SDL_MOUSEBUTTONDOWN)
	{
		std::cerr << "Mouse down ";
		if (event.button.button == SDL_BUTTON_WHEELUP) std::cerr << "WHEELUP\n";
		if (event.button.button == SDL_BUTTON_WHEELDOWN) std::cerr << "WHEELDOWN\n";
	} else if (event.type == SDL_MOUSEBUTTONUP)
	{
		std::cerr << "Mouse up ";
		if (event.button.button == SDL_BUTTON_WHEELUP) std::cerr << "WHEELUP\n";
		if (event.button.button == SDL_BUTTON_WHEELDOWN) std::cerr << "WHEELDOWN\n";
	}
}

PS: I am actually using Mac OS X 10.5 on Intel, but that wasn't in the list. I figured 10.4 would be more useful than "Other".