Comment 27 for bug 670016

> Diff looks good now, thanks. Still holding back, see comment 23. Should
> we reject the current maverick-proposed version if this is urgent?

Given that this has some regression potential I think it needs to live in -proposed for a bit to get adequately tested.

Regarding the current maverick-proposed version, the only known regression seems to be fairly modest (behavioral change in how maximization works), and the issue that it fixes (x server crashes) seems to be pretty severe, and is getting confirmed by a number of people. The behavior regression has only been reported by one person, who isn't running stock packages (although he says it happens there too, but solid evidence is missing).

I'd be tempted to allow the current proposal through. The benefit seems to far outweigh the risk in this case. However, if you're not comfortable with that, then yeah reject it out of proposed and put this one in instead. We can then sharpen our pencils on seeing if the alleged regression can be independently confirmed.