TABLE OF CONTENT

Hi,

ReSharper is doing great job when it comes to semantics of your code and control flow graph analysis. The special edge case I want to talk about is nullness analysis.

Unfortunately sometimes it’s hard to predict whether the method returns null or it doesn’t.

To solve this problem R# provides an option to annotate your code.

Let’s look at this snippet:

public Bar Foo()
{
    return Random.NextDouble() < 0.1 
                    ? null
                    : new Bar();
}

TheFoo()returnsnullwith a probability of 10%. However ReSharper doesn’t warn us when we forget the null-check.

screenshot
without annotation

Developer can mark the code with[CanBeNull]attribute in order to give a hint to ReSharper.

screenshot
with annotation

Unfortunately this trick doesn’t work with asynchronous code.

What if we have a code like this:

[CanBeNull]
public async Task<Bar> FooAsync()
{
    await DoAsync();
    return Random.NextDouble() < 0.1
                    ? null
                    : new Bar();
}

Despite the fact that we returnnullhere the actual return type isTask<Bar>. Which is not null by the way. So the annotation is wrong as well as the warning given to us by R#.

screenshot
attribute applied to Task, but no to Bar

I was always missing the ability to express that it’s not theTask<T>which can benull, but theTask<T>.Result. There was even an issue created back in 2013 and finally in ReSharper 9.2 we have a support of brand new[ItemCanBeNull]and[ItemNotNull]attributes.

ItemCanBeNull in action
ItemCanBeNull in action

Both attributes can be applied toIEnumerable<T>,Lazy<T>andTask<T>

ItemCanBeNull applied to Lazy
ItemCanBeNull applied to Lazy

Happy annotating!