The samples page you cite is helpful. The other two affordances that caught my attention on that page are (optional) multiline captures, e.g. for XML entities

The samples page you cite is helpful. The other two affordances that caught my attention on that page are (optional) multiline captures, e.g. for XML entities that might span lines, and the --replace modifier, which can transform a captured bit of text, simplifying the grep var= FILE | sed 's/var=\(.*\);/\1/' usage pattern.