Lookup Tables for Code Readability

It has been shown that lookup tables are often not a great optimization. In fact, they can slow down programs when used in place of direct value tests in if and switch. But after a recent experience I have started liking lookup tables more.

Over the last couple weeks I took it upon myself to implement a code highlighter—you know, the logic that makes code light up like a Christmas tree. This requires endless testing of bytes that must come before, after, or within keywords and other highlighted blocks.

With the initial implementation, I had a lot of direct byte tests. The code was pretty ugly—and had some bugs. I decided to rewrite the code with lookup tables—byte arrays of 256 elements that can replace direct value tests. And the end result? The code was much easier to read, and I felt better about its overall quality.

It is likely the lookup tables did not speed up the logic, but they improved the code. I think the lesson here is that performance is not everything—a more balanced perspective, where readability, correctness, and of course performance are considered is key.