Well if you ever do multiple predicates at the same level, you'd have to make it visually clear that you're doing something weird. Certainly the indentation would change. I don't know how though, because I haven't thought through any such examples.

As for difficulty in trying to grok how the library works, my goal it to make that much, much easier than it is now. If I succeed then it should be a fairly short learning curve to come up to speed on the library.

That said, I think there are some features that will turn out to be important that aren't there yet. One is that I'd like to have a "stack context" that you can keep adding to. If a problem happens, it would be nice for debugging purposes to figure out how this sequence of execution got *here*, but right now there isn't a stack backtrace that I know of. I'd like to see better facilities for exception handling, but that is likely to be hard to do. And I'm sure that there are some spit and polish issues that will become apparent once people try to use it.

But I'm interested enough in what's there now to look at it in more depth.