Coverity fixes

[natoscott]

No description provided.

[Explorer09]

I remember you argued about "non-null-pointer-assert" being just "line noise in the code".

[natoscott]

Yes, that's correct. That's not what this change is doing.

By NULL-pointer-assert we refer to the following ...

assert(ptr != NULL);
if (ptr->x ...) {
   ...
}

In this scenario, the assert line is not useful (IMO) because the very next line of code performs the same function, for both debug and non-debug builds.

[Explorer09]

@natoscott Well, I mean while some assert lines are not useful by themselves, they might convey the intent of the programmers. This way I can see when the lack of null pointer check is intentional, and not a developer's oversight.

I know the proper way to mark the pointer as being non-null is to use __attribute__((nonnull)). The assert statement is an old way to mark the same intent. It's not just useless code noise and that's my point.

[natoscott]

Sure, I understand your POV. A comment can often be a better way to go in that situation though.

[BenBE]

LGTM. Some minor changes still pending.