strings with backticks are misrecognized as template literals

[jpathy]

function foo() {
  a = '`sadsd`';
}
function bar() {
  a = '`sadas`';
}

Above program fails to parse.

[rhendric]

My bug! I'm looking into it.

[jpathy]

For the above program the following rule goes wrong:

language-javascript/src/Language/JavaScript/Parser/Lexer.x

Line 260 in b2f4cfa

<reg,divide> "}" @TemplateCharacters "`" { adapt (mkString' TemplateTailToken) }

I looked into the spec a bit and it seems that we need more lexical start codes than just reg, divide, see : https://www.ecma-international.org/ecma-262/7.0/#sec-ecmascript-language-lexical-grammar .

[rhendric]

Yeah, I came to the same conclusion. Unfortunately, I think we also need more state shared between the parser and the lexer, because the lexer alone can't determine whether a given } is the delimiter between expression and template characters. I have a work in progress fix; I'm just in the process of polishing it up now.