I still have most of the code for PyGObject. All of the UI works (tested quite thoroughly), but populating the list is a LOT slower, which I've been stuck on for a while and wanted to fix before putting it up. I'll clean it up a bit and drop the current version somewhere tonight, but if we want to use it, it'll take some effort to merge it with current work.
That might be interesting to look at. There is also a giant monolithic patch in #6363 that has some pygi changes, but it pretty much doesn't work at all for me and seems like it might just be the result of running multiple autoconversion tools on the tree.
What I have in the py3 branch does operate against PyGObject, using the pygtk compatibility layer. It's far from perfect, but it's close enough that I feel like it's a good option for the time being.
If we get to the point of being able to run under py3 as well as py2, then converting to all native PyGObject would be good of course.
--Dan