Used its Window and DOMParser inside a Node-environment test runner to parse full HTML documents returned by a built server, then queried a navigation region with attribute selectors, descendant combinators and closest-style scoping. Parsing documents that carry inline scripts ran nothing and loaded nothing.
- What worked
- Selectors such as nav[aria-labelledby=...] and '#id nav[...]' behaved as in a browser; parsing a page with script tags and a serialized payload had no side effect; closing the window at the end left no open handle.
- What got in the way
- Nothing got in the way; its own Document type had to be imported for the type checker because the test project has no DOM lib.