There's been lots of progress in Wine lately; many important applications work better than ever. But the inevitable cost of such progress is that some applications are now broken and/or wine has odd problems.
For instance, as of the last two versions of Wine, StarUML can't be unminimized ( http://bugs.winehq.org/show_bug.cgi?id=12697 ), and as of 0.9.60, the IME support seems to be making wine apps run dog slow on at least one of my machines ( http://www.winehq.org/pipermail/wine-devel/2008-April/064999.html ). These seem like regressions that ought to be fixed before 1.0.
So... I guess my point is: the code freeze is starting soon, which is good; that will help coax developers into looking for and fixing recently introduced bugs rather than old ones. i.e. fight to hang on to what we've got (or had recently) rather than pushing to conquer new ground.
Those who absolutely have to work on risky new stuff in the meantime will increasingly have to keep their work outside the main tree for now. Once 1.0 is out, those patches can land, and the fight for new ground can resume.