I do still exist! I am currently working on the first implementation. After the funcakes I had with Bison, you can fully expect that I've been working on the custom parser. My custom lexer is nice and high quality thanks to the fact that I ended up re-writing it about six billion times. At least when my parser throws an exception, I will have nice high quality error messages. That's my end goal, anyway.
Now, for a custom parser, I am obviously going to do the horrendously efficient thing and work without an explicit stack. I am going to instead simply increment/decrement the current element pointer through an existing vector. This should permit a ridiculously small performance improvement for a 1billion times increase in development time.
Seriously though, I just find it more natural.
No comments:
Post a Comment