Jim Idle wrote:

I was dreaming about this last night - I know, it's sad - but when I woke up I thought of something that shuodl be easy:
 
1. Add integer ruleDepth; to the state
2. Reset it to zero in nextToken
3. Each lexer rule increments and decrements it
 
Then, depending on how much you want people to have to change grammar for 3.1, either:
 
Just add $ruledepth and a template state.ruleDepth and tell people to use:
 
$ruledepth == 1 $type = TTTT : null;

boy this sure sounds familiar! i thought i did a ruleLevel thing once to set token type...let's keep this on radar to fix.

 
Or, add dummy locals (or dummy int and String in state) so that the template for say $type is:
 
(state.ruledepth == 1 state.type ? _type)
 
And then have:
 
$super::type
$super::channel
 
And so on, which don't check the state.ruleDepth as they just set it and the next thing that does emit() or returns to nextToken picks it up correctly.
 
Trivial change for uses either way around. I kind of like super:: as it might be useful for other things later, but your call of course.