Inconsistent results when parsing element with missing end tag.

[grokys]

Consider the following test:

[Theory]
[InlineData("<X><a><</X>")]
[InlineData("<X><a></</X>")]
public void IncompleteEndTag(string xml)
{
    var document = T(xml);
    var element = Assert.IsType<XmlElementSyntax>(document.Root.Elements.Single());
    Assert.Equal("a", element.StartTag.Name);
    Assert.Equal("", element.EndTag.Name);
}

In the first case ("<X><a><</X>") the test passes and element.EndTag.Name == "" however in the second case ("<X><a></</X>") the test fails because element.EndTag.Name == null.

It feels like this inconsistency is accidental, should it be normalized?

[KirillOsenkov]

I think it's purely accidental, we should normalize. Seems like an empty string is better than null.