We previously took a look at the explicit nature of changes to data and system state that defines imperative programming as opposed to declarative programming, where instead of manipulating data in a loop such functionality could be declared as mapping an operator to some data, thus spelling out the functionality, not the specific order of operations. But why should programming languages necessarily be a choice between imperative and declarative paradigms?

In fact, one of the main distinguishing features of C++ is its multi-paradigm nature making use of both imperative and declarative paradigms. With the inclusion of object-oriented, generic, and functional programming into C++ in addition to C's procedural programming, it would seem natural to assume that this would all have to come at a cost, whether in terms of higher CPU usage or more RAM and/or ROM consumed.

However, as we learned earlier in this chapter, C++ language features are ultimately built upon the C language, and as such there should in turn be little or no overhead relative to implementing a similar constructs in plain C. To resolve this conundrum and to investigate the validity of the low-overhead hypothesis, we'll now take a detailed look at a number of C++ language features, and how they are ultimately implemented, with their corresponding cost in binary and memory size.

Some of the examples that focus specifically on C++ as a low-level embedded language are taken with permission from Rud Merriam's Code Craft series, as published on Hackaday: https://hackaday.io/project/8238-embedding-c.