I have about 20,000 questions and non-implemented bug-reports regarding this, but I don't want to flood the list or write a novel, so let's start with a kind of announcement/heads-up:
at git://github.com/gmt/gmt-cygwin-overlay.git I have a WIP cygwin-gentoo-prefix overlay. It's very, very rough; indeed, I still am using out-of-tree scripts to handle the more gruesome rebase issues. However, I do have some ameliorants for rebase problems in the overlay and several ideas for how to fundamentally fix the problem going forward. Where I am now is just after emerge --sync in the standard prefix bootstrap process. Most of the overlay is based against the bootstrap tree, not the rsync tree, so, going forward, I will be trying to forward-port, playing catch-up with the rsync tree, as my top-ish priority. In the meanwhile, please be advised that I've taken an uber-agressive posture regarding what I'm willing to put in there -- any kludge, no matter how vomit-inducingly awful it may be from an elegance/portability perspective, has gone in, if I so much as suspected it might fix a problem. That's resulted in a good number of silly changes with low or negative intrinsic value. So, there is some noise and there is some signal in there. Much of it will eventually go into the bit-bucket. Please try not to think I'm a total freak, if you see some really horrible stuff in there. Please, tinker with it if you like, but I'd advise folks don't drive themselves crazy trying to actually use this thing. I do seem to have all of @system working reasonably well (pre-rsync) while bootstrapping this tree, but that doesn't mean there's any sane path leading to where I managed to get with it. I do eventually want to produce a reasonable bootstrap recipe, but I clearly have my work cut out for me. -gmt
