Beautiful Code

Chapter 5. Correct, Beautiful, Fast (in That Order): Lessons from Designing XML Verifiers

Elliotte Rusty Harold

This is the story of two routines that perform input verification for XML, the first in JDOM, and the second in XOM. I was intimately involved in the development of both, and while the two code bases are completely separate and share no common code, the ideas from the first clearly trickled into the second. The code, in my opinion, gradually became more beautiful. It certainly became faster.

Speed was the driving factor in each successive refinement, but in this case the improvements in speed were accompanied by improvements in beauty as well. I hope to dispel the myth that fast code must be illegible, ugly code. On the contrary, I believe that more often than not, improvements in beauty lead to improvements in execution speed, especially taking into account the impact of modern optimizing compilers, just-in-time compilers, RISC (reduced instruction set computer) architectures, and multi-core CPUs.

The Role of XML Validation

XML achieves interoperability by rigorously enforcing certain rules about what may and may not appear in an XML document. With a few very small exceptions, a conforming processor can process any well-formed XML document and can identify (and not attempt to process) malformed documents. This ensures a high degree of interoperability between platforms, parsers, and programming languages. You don’t have to worry that your parser won’t read my document because ...