Typical project memes

>we need to have a file for every little block of code
>..and a separate makefile per directory
>plus a dozen of config files and crap that get written in .gitignore
>pull dependencies straight from github
>author creates issues as to-do list
>Last updated 3 years ago, but 120 pull requests not answered
>400 forks with minor differences because programmers are naturally autistic and can't stand others code

>>12
Segfaults and overflows are the low hanging fruit really. The bigger problem is that you've often balls deep in a larger system that's outside your control (think of kernel drivers or utility libraries - without somebody else's application, or worse, somebody else's hardware, your code doesn't even run).

You can talk all day about mocking things for tests but the reality is this almost never happens. C is not like JavaScript or Java where most functions are dynamically dispatched and implementation data structures are hidden. You can't even sprinkle virtual over things like you can in C++. You basically have to add a whole new target to rebind things at build time for every test, or rewrite your code to support dynamically rebinding things in ways the production version will never do. Both are expensive in terms of development time and can impact performance adversely if you take the latter approach.