Question

If the Dictionary<TKey, TValue>.TryGetValue method returns true, the parameter value contains the value associated with the specified key.

I have read some examples of changing the value associated with the specified key: after invoking the TryGetValue method, the updating of the value associated with the key is performed by the indexer, resulting in a further access to the Dictionary (see the following code).

var d = new Dictionary<string, MyClass>();
...
MyClass obj;
if (d.TryGetValue(key, out obj))
{
    d[key].Update(...); // update the value
}

The obj object is a reference to the value associated with the key. Why not directly use the object to update the value, as in the code below?

MyClass obj;
if (d.TryGetValue(key, out obj))
{
    obj.Update(...); // update the value
}

Answer 1

Example code is not the same as real code. It demonstrates a given point, and nothing else. This can mean lots of examples are sub-optimal (can we all promise never to use ToList() in linq example code that doesn’t actually need it? I’m convinced that’s why it’s so often used when the sole effect is to make the code slower) and very many that are pointless.

This isn’t even a bad thing. Commenting in example code will describe the things that are being explained, which in real code we should expect people to understand. Meanwhile it won’t document why you’re doing some weird-looking thing because that’ll either be in the surrounding text or the answer is simply “because it demonstrates this feature, nothing more”. In real code, anything that is surprising or weird-looking should always be explained in comments (as well as the obvious reason, if you can’t write a good explanation of the weirdness, you aren’t as justified in that weirdness as you thought you were). While some junior developers comments may suggest they’re doing the former in real code, the fact is that we want comments in examples to explain what would be obvious to someone who knows the features in question – because we’re not faimilar with them, and that’s what we’re reading the example for,

With a mutable reference type, the second version is indeed more sensible, and frankly a better example, but the author was presumably just trying to think about yet another example of yet another feature which is a different mental pressure to be under than that we are under when writing real code. The opposite in some ways (we want to favour the well-known over the exotic, while a tutorial writer has to use every single feature in order to explain them).