The Shape of Code

All the data+code from my book The New C Standard: An Economic and Cultural Commentary is now available on GitHub. For many years I have been meaning to create an easy way to map from a graph/table in the book to the file containing the data, which has blocked me adding the data to GitHub. I have unblocked by releasing this minimal viable product, i.e., it is essentially a copy of the usage subdirectory in the book’s directory.

While the five stage process to get from graph/table to data is tedious, at least there is a process that provides the data. The caption of the graphs in my Evidence-based Software Engineering book contain a link to the corresponding file on GitHub. This was not possible for the C book because GitHub was still 3-years in the future when the book was published (in 2005).

Work on the book started in late 1999 and measurements of C usage was an integral component. Publicly available source code was still a novelty and large Open source projects were rare (SourceForge was launched at the end of 1999). The large projects with C source available to measure were: Linux, Netscape, Gcc, PostgresSQL, OpenAFS, and OpenMotif. Several popular projects originally written in C had migrated to using C++, and were therefore not applicable.

As the book was completed in 2005, evidence-based software engineering restarted, 20-years after the fall of Rome. Or rather, I have nominated 2005 as the year this happened. Feel free to quibble plus/minus a few years.

Search engines were an essential tool for obtaining research papers, reports, and occasionally downloading data. In 2000 the search engine of choice was AltaVista, but a few years later Google had become the best.

While writing the book, I was a regular visitor to bricks and mortar buildings called libraries. Back then, university libraries contained tens of thousands of physical books, and researchers would photocopy papers of interest. Little did I know that this research practice would soon be dead.

In 2005, I had this to say about software evolution:

Measuring the characteristics of software that change over many
releases (software evolution) is a relatively new research topic.
Software evolution is discussed in a few sentences, and any
future major revision ought to cover this important topic in
substantially more detail.

How might C source code characteristics have changed in the last 20 years?

• The use of K&R style function definitions is probably very rare; it was well on the way out in 1999,
• big software systems have gotten bigger, i.e., more lines of code and more #includes,
• A lot more code using 32-bit integers and 64-bit pointers,
• More storage allocated (memory capacity has increased) because it’s faster to do everything in memory, and there is more data.