Shadow DOM: Contentious Bits

Multiple Shadow Roots per Element

Current state: removal being discussed

Discussion: Aug 2011, Feb 2015

Pros:

• enables consistent story for adding shadow trees to builtins
• provides reasoning about subclassing DOM trees

Cons:

• complexity: (link to trac.webkit.org patch removing the code?)
• performance: may result in "submerged" trees that aren't rendered but still participate in style/layout

Cost/benefit of change:

• Disables the use case for general inheritance-based component composition (examples: http://blog.quickui.org/2013/11/08/filling-slots-in-shadow/)
• Makes implementing Shadow DOM easier

Alternative ideas:

• add imperative equivalent, the ability to transfer a shadow tree from one host to another.

The default value of "closed shadow tree" flag

Current state: in need of consensus, currently spec'd as open-by-default, with strong opposition to flip to closed-by-default.

Discussion: June 2011, Feb 2012, Oct 2012, Mar 2013, Feb 2014, Feb 2014, Feb 2015

Open-by-default Pros:

• offers enough of a doorstep to avoid unintentional wandering across composition boundaries.
• compatible with most of today's tooling/testing/introspection techniques in the wild

Open-by-default Cons:

• developers are likely to ignore the doorstep in cases where they "just need stuff done", and thus negatively affect component's composition qualities (robustness, maintainability, reuse).

Closed-by-default Pros:

• provides a more explicit barrier to prevent deliberate access (though not actively hostile attempts at gaining entry)
• ensures that the internal structure of the component is never part of the contract between component user and component developer

Closed-by-default Cons:

• forces component testing to use different code from what is shipping (no way to test the inside of the component instance)
• disables a class of introspection tooling (example: design-mode introspection in a drag/drop designer) and checkers (example: accessibility checker)
• breaks event delegation

The imperative distribution API

Current state: needs solid proposal