This is a really neat project; thanks for your work on it. I think I've discovered a repeatable issue to do with ID assignment for choosers: the grammar 'test.rm':

https://gist.github.com/danielghurley/2bb2e7fda7b992e42289#file-test-rm
https://gist.github.com/danielghurley/d605305ef43c4373a20d#file-first_include-rm

has one rule with a choice defined in the main text, and two more rules with choices defined in 'first_include.rm'. When you run it:

rmutt text.rm

should expand to text which always finishes with either 'first main option' or 'second main option', but occasionally it returns nothing on the third line. From inspection of the transpiled code:

https://gist.github.com/danielghurley/d39e278b1b44c3f09332#file-test_transpiled-txt

It looks like 'main_rule' (l916) and 'third_include_rule' (l1908) have the same ID, and I think this is a problem. The behaviour above doesn't occur if third_include_rule has the same number of possible choices (2) as 'main_rule'.

From inspection of the pegjs code, I think maybe nextId is not being incremented correctly during transpilation, and that means occasionally the chooser will return an index which is out of bounds for the array of choices. I am happy to try and fix this and submit a PR but I don't know much CoffeeScript or pegjs right now.

Use case: To use the tool in a browser based environment, e.g. because setting up node is not feasible or because one wants to provide a live-editing version etc. it may be useful to have a file which can be used via in a <script src="…

I'm the original author of the C version of rmutt, could you credit me in the docs and link to the rmutt github repository at joefutrelle/rmutt? Thanks.