Output ambiguity

[diervo]

@inikulin An interesting though:

Input:

    <input value={foo}/>
    <input value="{foo}/">

    <div title={x}></div>
    <div title="{x}"></div>

Output:

    <input value="{foo}/">
    <input value="{foo}/">

    <div title="{x}"></div>
    <div title="{x}"></div>

Although the transformation is correct from a spec perspective, it generates a little bit of ambiguity from a developer semantics perspective.

How do you feel about adding some metadata to the AttributeNode (which we collect already as we generate tokens) such as for this example valueState: single-quotes|double-quotes|unquoted

I think via location you could potentially figure out, but is not really straight forward tough.

[inikulin]

I'm not sure I understand what's ambiguous here. Can you elaborate, please?

[diervo]

In the first example:

 <input value={foo}/>
 <input value="{foo}/">

From an AST perspective what you get in both case is exactly the same (attributeValue="{foo}/"). It will be great to have some extra metadata in the AST Node, so we could build some extra semantics on top (quoted value vs. unquoted)

For example, imagine we want to error out in any attributeValue that is non quoted. At the AST level we won't know if you did quoted or unquoted (be lost that information already), therefore ambiguous for this particular case.

[inikulin]

Hmm, I guess adding / to the list of characters that trigger unexpected-character-in-unquoted-attribute-value error can resolve this.