Enforcing Component Constraints

[Cantor, Scott]

On 1/27/17, 1:45 PM, "dev on behalf of Brent Putman" <dev-bounces at shibboleth.net on behalf of putmanb at georgetown.edu> wrote:

> Note that: The callee method really could only drop the check if the potentially-null arg was only going to be used
> internally within itself, and in such as way that you know that a null would "safely" result in an NPE.

I have a much harder line on that, I guess. If something is annotated nonnull, that to me says the code should be able to operate as though it is a given that that variable is not null. That goes for everything, including subsequent uses of the variable as input to something else. If it's null, there was a bug in the original caller, and when it fails, that code will be fixed. That's just how I think about the concept, it was why I thought the idea was created.

Otherwise the code is just being littered with the same null checks that would already be happening in a world without the annotation, because as you note the common case will be crossing of methods anyway.

I recognize that the AOP trick is a nice middle ground, with my annoyance at seeing the checks dispensed with. So I definitely can see the value of that.

> So it has the obligation to do the check, and incur the cost.

I think it is freed from that obligation by its own annotation.

-- Scott