There are still some fairly hard problems in parsing. For example, doing minimal work to convert a series of text deltas into abstract syntax tree deltas, using caching to avoid throwing away too much. This is highly relevant to IDEs for providing code completion and other analysis, but it's usually solved with a mix of brute force - restarting the whole parse from the top - and trickery, such as skipping uninteresting function bodies, or parsing a much simplified version of the language that ignores many productions.
