Inconsistent parsing for tag classes and ids about shakespeare

[snoyberg]

Just to confirm: this is applying to Hamlet, not Lucius or Cassius, correct?

[pacak]

Yes. I actually didn't tried replicating this in Lucius/Cassius, so it's just Hamlet.

[snoyberg]

The fact is that the spaces are not intended to be optional between the class and ident tokens, as we start to run into syntax ambiguities (the hash is overloaded for "insert a variable" and "start an identifier"). So I'd actually be tempted to say that both of those are invalid Hamlet and should raise a warning. But I'm not sure how others would feel about such a change.

[pacak]

h1.foo#bar and h1#bar.foo looks pretty unambiguous to me and converting that into invalid hamlet sounds like a bad idea.

[gregwebs]

I don't see where there would be ambiguity. I think both should be supported. An id is more important than a class and I think should go first as a best practice, but I don't think Hamlet should enforce that.

[snoyberg]

The ambiguity comes from:

h1.foo#{bar}

We could do more checking: if a hash is followed by a left brace, then it's not an ID. But that's the reason I'm in favor of just making all of the spaces a requirement.

[gregwebs]

I constantly use h1#id.class
I never use .class#id, and am ok with saying that it isn't supported.

It is bad to currently pass # through as an invalid class name character though.

[snoyberg]

Agreed on that last point, I've added an error explaining the problem.