After some thoughts, I think this should be pushed to testing faster. Because the features I enabled which are currently in testing are just plain buggy, there are multiple security bugs in virtio-3d display (which is being disabled in sid), and because with just 2 weeks of this being in testing, it will come to a big surprize for users if it hits stretch and there will be as much complaints about new dependencies as I'm receiving now.
In short, we need 1:2.8+dfsg-2 in stretch, and the faster it will be there, the better, because we'll at least have some chances (including bpo version) to see what else could be missing/broken and fix that. I already prepared bpo version which is waiting this version to hit testing. Also there are a few more security fixes being accumulated, and also a few non-security probs being worked on, which should be in stretch, -- especially probs with migration from older qemu. Maintaining 1:2.8+dfsg-1 (with gtk and 3d) in stretch will be difficult. I understand this is my fault to enable large features this close to the release, hence I'm trying to fix that before it is too late. I waited for upstream 2.8 version in a hope there will be modular display support so it'll be possible to make graphics packages/dependencies optional, but these patches weren't accepted. Please consider shortening the waiting period for this version. Thank you! /mjt
