Chris Barth

If you look at `selectWithResolver()`, which all these eventually call, it appears that any given `XPath` can return a `string`, `number`, `boolean`, or `result.nodes` in the case of `false`, or...

Thank you for that thorough explanation. It seems that `SelectSingleReturnType` should then be typed to return `undefined`. After you push that change, I'll check this against `xml-crypto` and let you...

I also wonder if we should more consistently name these checking function: e.g. should `isDocumentFragment()` be `isDocumentFragmentNode()` like we have `isDocumentTypeNode()`? That would keep it consistent and I think I'm...

> make a PR with that targeting the `types` branch of this repo See #125

It appears from #127 that people are using these new support functions. I've created #128 to continue to validate these functions as I work on the next release of `xml-crypto`...

@forty , I can certainly make a change in my other branch to accommodate that suggestion. In fact, it does make sense to use `unknown`.

I've moved all this type checking code to another repo so that we don't muddy the purpose of `xpath` with them See https://github.com/xmldom/is-dom-node I'll still gladly do what it takes...

I tried figuring out a clean way to do this, but was unable. Instead I had to change the way my code is structured to not initialize the `winston` loggers...

I would love to see your CI testing drop support for very old, unsupported, versions of Node.

@theasta Are you accepting PRs for this? If so, do you have any guidance on how you'd like it implemented? I was thinking a json object in a file that...