Hi Ken If you don't want GLR parsing (or attribute grammars) the Happy docs should be fine. GHC uses Happy, though an LR grammar, for its front end, so regular LR parsing always works. If you want GLR you might want to search the web to find out who authored the GLR extension. I'm afraid I've forgotten the person, but I seem to remember there was a thesis (possibly United Kingdom MSc ?) accompanying it. IIRC, the Happy GLR extension seemed to focus on natural language parsing rather than parsing ambiguous programming languages (Happy is generally used just to parse programming languages rather than natural language). The GLR parsing algorithm originally came from Natural Language Processing but because it handles ambiguity it was adopted by compiler developers. In practice, industrial strength GLR parsers for programming languages (e.g. SGLR and the commercial DMS from Semantic Designs) add a fair amount of kit so they can disambiguate what they parse and make good parse trees - I don't think Happy's GLR extension had this. On 14 June 2012 20:43, Ken Overton <koverton@lab49.com> wrote:
Thanks; is there a better, more current example that you know of where I might start instead?