Hi!

I noticed that the chosen behaviour for boolean transform is

which is:

it's true if value of attribute starts with t, y or a non-0 number.

and it's tested:

On the other hand HTML boolean (https://developer.mozilla.org/en-US/docs/Glossary/Boolean/HTML) attribute is considered true just because it's present:

If an HTML tag contains a boolean attribute - no matter the value of that attribute - the attribute is set to true on that element. If an HTML tag does not contain the attribute, the attribute is set to false.

That leads to interesting behaviour - if attribute is present, but not assigned a value (i.e. it's HTML Boolean), it's cast to false (not just undefined but actually the polar opposite)

I detected it wile using with Lit booleans (https://lit.dev/docs/templates/expressions/#boolean-attribute-expressions) - Lit implements HTML boolean.

I think it can be resolved in a backwards compatible way.

I see that transforms are given the attribute name as 2nd argument

• in constructor
  

  react-to-web-component/packages/core/src/core.ts

  Line 100 in 002a007

  this[propsSymbol][prop] = transform.parse(value, attribute, this)

• in attributeChangedCallback
  

  react-to-web-component/packages/core/src/core.ts

  Line 130 in 002a007

  this[propsSymbol][prop] = transform.parse(value, attribute, this)

So parse could be made into

  parse: (value, attribute) => 
    // Check if it's HTML boolean
    value === undefined
    || value === ''
    || value === attribute
    /// Check if it's boolean by value
    || /^[ty1-9]/i.test(value), 

I don't have an idea for stringify, though. It would have to be a choice of the author (i.e. differently named transform) I guess