Abstract Parsing #2

Sometimes you get an idea. You play with it for a while, but after some time some unforeseen gotchas come to your attention. You rework the idea accordingly and then repeat. This can go on for a week or more, until you arrive at something that works, or maybe you ditched the idea at some point.

At other times, you get idea that kicks ass – it is elegant, clean, easy to understand, works in theory, and when you implement it everything actually works the first time around. I had an idea for how to upgrade the abstract stack representation (mentioned in previous post) that I was planning to try out tomorrow, but my other plans got marooned so I gave it a go. As it turned out, it only took about 30 minutes of coding, and once it compiled everything actually worked right off the bat. The parse graph for the previous example is as accurate as it can possibly get now:

I could hardly stop laughing in disbelief when I saw that graph! If you don't think this graph if beautiful something must be wrong with you then you must be a normal person.