Mike,
You might consider incorporating gl.tter's LSE (Light Signal Engine) code into your custom version of MAME.
Alternatively, if you make your DLL conform to what gl.tter expects, you could just supply that and BuddaMAME will drive the board. You wouldn't need to maintain your own build.
Is this gl.tter thing available?
My DLL just talks to the LEDwiz. It doesn't and shouldn't do anything else. It is an API for using an LEDwiz. Applications should conform to the API. It exists purely as an abstraction of the hardware so users don't need to know how to communicate with the device.
edit: (warning... sort of a rant)
I should also mention that the real reason I started working on my own version of MAME was to add hardware acceleration of 3D graphics (using D3D/OpenGL or a custom driver if necessary) to the later 3D games. I get a lot of people suggesting I should make my stuff external to MAME so that I don't need to support a custom version of MAME. I get suggestions that I should do this or that. I get a lot of people just asking questions. This takes up time and frankly, the suggestions are all so someone else can get some benefit from my work. I never intended on releasing this stuff to the public and have been developing the setup that I wanted. The LEDWiz project has become a distraction from my original plans. If people can use what I have, I don't mind releasing it.
People have to understand though that writing large scale software takes a lot of time. It is a lot of work and it is work that I don't get paid for. For me it is doubly like work because I write software for a living. It also takes a lot of time to dive into the MAME code and figure out how things work. I enjoy this when I'm doing it for me at my own pace. When I suddenly feel like I'm on someone else schedule working on someone else's agenda, it isn't fun anymore.
If the MAME software community could come together and create a single custom version of MAME that we all put our modifications into, this would be something I could really get behind. This way, we don't have a bunch of different MAME's floating around, all of which have some nice feature but no one of them having all the features I want. I would be willing to listen to input and support the thing like it was a product, but everyone else who contributes to this version of MAME would have to make an equal commitment. I guess what I'm suggesting is a MAMEdev type of team for extra features. This would be beneficial too because when one person is too busy to add some feature, someone else on the team could do it with help/guidance from the owner of the feature. Everyone on the team would learn from each other. It would also reduce the amount of request we all would get to " add bob's feature, its cool". It would be beneficial because the team might be able to generate financial resources to purchase tools, source code control softare, hosting space, etc. These are all real expenses that are incurred by making a free product available to the public. My biggest hesitation on releasing something is my fear that I will become inundated with requests, suggestions, etc and that I will have to abandon this forum for sanity sake.
If people aren't interested in something like this, then people will have to just accept what what they get and when they get it. "If it doesn't do what you want, don't use it."