on its own, this fix makes sense to me. How does it relate to the next comment, which says there are two possibilities -- jsxtext inside a <div> and non-jsxtext inside a type-argument-having <Component<T>? It seems like the second case might be handled entirely by the new code, but I'm not sure.

If it is, then the following check could be simplified.

I thought the same thing, but removing it breaks the test added in the original "don't do completions in JsxText" PR; I can't remember if I figured out why or not at this point, though, so can go double check.

Seems like it's the comment that needs to be updated; the second listed possibility is no longer accurate as location in this case isn't GreaterThanToken, it's the entire element. I'll change the comment to reflect that it's only handling the first case.