How it works...
In our recipe, we initially set DEBUG to *, which means enable all logs. Then we wanted to zoom in explicitly on express related log messages. So we set DEBUG to express:*, which means enable all logs that begin with express:. By convention, modules and frameworks delimit sub-namespaces with a : (colon).
At an internal level, the debug module reads from the process.env.DEBUG, splits the string by whitespace or commas, and then converts each item into a regular expression.
When a module uses the debug module, it will require debug and call it with a namespace representing that module to create a logging function that it then uses to output messages when debug logs are enabled for that namespace.
For more on using the debug module in our own code, see Instrumenting code with debug in the There's more... section.
Each time a module registers itself with the debug module the list of regular expressions (as generated from the DEBUG environment variable) are tested against the namespace provided by the registering module.
If there's no match the resulting logger function is a no-op (that is, an empty function). So the cost of the debug logs in production is minimal.
If there is a match, the returned logging function will accept input, decorate it with ANSI codes (for terminal coloring), and create a time stamp on each call to the logger.