The new project is something I'm quite excited about. Another GDNet member and I have decided to collaborate on a project and (try to) do it right. The main thrust of the project is to use as many third party libraries and 'engines' as possible in order to focus squarely on the game layer. Of course, you can't just enter something like this blindly so we've started out on some exploratory research into existing engines. I discussed previously that Irrlicht looks promising and indeed it still does, but after looking through the code and playing around with some code it is evident that it has a few annoyances (or flaws, whichever way you want to look at it).
These flaws highlighted why I think I've been so quick to jump into designing my own engine. It seems a lot more comfortable to start from base and do it 'your way' rather than to put up with flaws, annoyances and shortcomings of someone else's work. Although there's a lot more work in the hand-rolled method, it is probably an important motivating factor for people to start out on their own. Rather than having to learn a system and an API, it seems easier to just jump straight into coding your own.
Whilst looking at these other options, I find myself saying "well I don't like this, that, the other" but never "wow, I do like the way it does xyz" or "Wow, this will save me months of work". Funny how we think like that really.
I'm still determined in this quest. There's currently no goals set - just a bit of R&D and experiementation but I hope to see it start panning out and becoming more tangible in the coming months. I should be able to hit the accelerator in my efforts as soon as my current project is completed.