C has its limits. If you know where to look.
In the early days of “K&R C”, things were quite a bit different. C was not nearly as portable as it is today. While the ANSI-C standard helped quite a bit, once it became a standard, there were still issues when moving C code from machines of different architectures — for example:
What is x? According to the C standard, and “int” is “at least 16 bits.” On my Radio Shack Color Computer, and int was 16-bits (0-65535). I expect on my friend’s Commodore Amiga, the int was 32-bits, though I really don’t know. And even when you “know”, assuming that to be the case is a “bad thing.”
I used a K&R C compiler on my CoCo, and later on my 68000-based MM/1 computer. That is when I became aware that an “int” was different. Code that worked on my CoCo would port fine to the MM/1, since it was written assuming an int was 16-bits. But trying to port anything from the MM/1 to the CoCo was problematic if the code had assumed an int was 32-bits.
When I got a job at Microware in 1995, I saw my first ANSI-C compiler: Ultra C. To deal with “what size is an int” issues, Microware created their own header file, types.h, which included their definitions for variables of specific sizes:
