For those of you who don’t care how the DOM is implemented, only that it generally works: this is definitely a “tl;dr” blog post.

For those who do care: I’ve been thinking long and hard about how XBL’s shadow content, and how undo and redo operations, work in Gecko.  Ultimately, I find the current implementation just doesn’t support what I need for XML templates.  So I am working on a very different model, one where I base Node objects on “DOM relations” objects.  These latter objects control the relationships between nodes, instead of the nodes themselves doing it.  They also make undo/redo and shadow content much easier to maintain in my opinion (at the cost of additional memory, which in JavaScript land I don’t have to worry about… too much).

I’m not suggesting Gecko DOM implement this (though that would be very nice of them).  I’m going to prototype it for my own needs, and see how well it works.  I hope this explanation is in clear enough language for the average developer to understand.