299300 2025-09-22 09:51:06 -0700 Calling attachShadow does not inherit host customElementRegistry as expected 2025-10-06 23:33:52 -0700 1 1 1 Unclassified WebKit DOM Safari 26 Unspecified Unspecified RESOLVED INVALID P2 Normal --- 1 steveorvell webkit-unassigned annevk justin rniwa oldest_to_newest 2144895 0 steveorvell 2025-09-22 09:51:06 -0700 When `attachShadow` is called without a registry, the shadowRoot does not inherit the host's customElementRegistry. See reproduction: https://codepen.io/Steve-Orvell/pen/xbZGxOW?editors=1000 This expectation is based on https://dom.spec.whatwg.org/#dom-element-attachshadow "The attachShadow(init) method steps are: Let registry be this’s custom element registry." 2144905 1 justin 2025-09-22 10:25:44 -0700 The spec behavior seems wrong. If the custom element isn't passing a registry to attachShadow(), then presumably it's expecting use the global registry. Using its parent scope's registry would very easily break it as that scope might be missing custom element definitions from self-registering element dependencies. 2145013 2 rniwa 2025-09-22 15:26:09 -0700 attachShadow isn't supposed to inherit host's customElementRegistry. 2145038 3 steveorvell 2025-09-22 16:26:19 -0700 @Ryosuke 1. Can you explain the linked spec text? 2. Can you explain the inconsistency between a shadowRoot on a div and on an undefined custom element candidate in the linked repro? 2148993 4 annevk 2025-10-06 23:33:52 -0700 1. See https://github.com/whatwg/dom/issues/1407. 2. The patch for bug 300027 did capture the test for the div element inside an "inert" shadow root itself getting a shadow root with a custom element registry. No fix for that yet though, but it seems wrong to track it as part of this bug which started out about something else.