C Interfaces and Implementations: Techniques for Creating Reusable Software

Chapter 3. Atoms

An atom is a pointer to a unique, immutable sequence of zero or more arbitrary bytes. Most atoms are pointers to null-terminated strings, but a pointer to any sequence of bytes can be an atom. There is only a single occurrence of any atom, which is why it’s called an atom. Two atoms are identical if they point to the same location. Comparing two byte sequences for equality by simply comparing pointers is one of the advantages of atoms. Another advantage is that using atoms saves space because there’s only one occurrence of each sequence.

Atoms are often used as keys in data structures that are indexed by sequences of arbitrary bytes instead of by integers. The tables and sets described in Chapters 8 and 9 are examples.

Interface ...