Questions about HTMLTemplateElement.content getter

[AtkinsSJ]

What is the issue with the HTML Standard?

From the spec:

The content getter steps are to return template's template contents, if the template contents is not a ShadowRoot node; otherwise null.

Implementing this in Ladybird has no apparent effect on WPT test results, which raises a few questions to me:

1. How would one produce a template element with a ShadowRoot as the template contents?
2. Is it actually possible to do so?
3. The IDL defines it as readonly attribute DocumentFragment content; - this suggests it is not nullable and should always return a value, which conflicts with this algorithm.

[annevk]

@mfreed7 and I probably made this mistake while working on #5465. Finding the exact discussion is hard given that GitHub hides everything by default.

I think we can do what WebKit implements here which is assert that template contents is not a ShadowRoot node. Because whenever it is a ShadowRoot node we don't actually create a template element (it's only on the HTML parser stack) so there's nothing to run script against.

[annevk]

#11954 addresses this, thanks for reporting!

[mfreed7]

@mfreed7 and I probably made this mistake while working on #5465. Finding the exact discussion is hard given that GitHub hides everything by default.

+1, I think this is a leftover from when declarative shadow DOM didn't do streaming, so the <template> element could live in the tree for a bit while parsing. #11954 LGTM