It seems that IntelliJ has a notion of what characters are used as word separators, and this has a impact on things like "Select Word at Caret", selecting text word-by-word with the keyboard, or double-clicking on a piece of text.
Say you have `foo-bar` in Java. It means `foo` minus `bar`. So `foo` is a word and `bar` is a word. If you double click on a letter of `foo` it selects just `foo`. But if you have `foo_bar`, the whole thing is a word. Double click on a letter of `foo` and it will select `foo_bar`. This is reasonable for Java.
But it isn't for XML: in XML `foo-bar` as a whole should be considered as word, just like `foo_bar` is in Java. Because in XML, `foo-bar` can be a valid id, or in XPath it points to the element `<foo-bar>`. As far as I know it never means `foo` minus `bar`. The problem is that in XML, you often go for a "lowercase with dashes" convention, which makes it harder to select those ids in IntelliJ.
Let me get to the point: is there a way to setup IntelliJ so, for XML files, `-` is not considered as a word separator, similarly to `_`?