Unknown elements

A document may contain elements the active schema doesn't describe. Rather than silently dropping them (and their content), JinnTap synthesizes a generic schema entry for each unknown element so it round-trips through the editor and stays editable.

This is handled by src/util/unknown-elements.js, which runs whenever content is loaded.

The document below contains a block-level unknown element (<ab> from a foreign namespace) and an inline one (<soCalled>). Both survive loading, stay editable, and serialize back unchanged — toggle the source view (</>) to check.

How the kind is inferred

The synthesized type is inferred from how the element is actually used in the document, classified to a fixpoint (because an element's nature can depend on its unknown parents/children):

Attributes seen on the unknown element are collected into the synthesized entry so they are preserved too.

Visual marking

Unknown elements are tagged in the reading pane so authors can see that they aren't part of the schema:

• block/inline unknowns get a jinn-tap-unknown class and a data-tag marker,
• empty unknowns get a jinn-tap-unknown-empty class.

These render-time hints never leak into the serialized XML.

Original-case names

Custom-element tag names are lower-cased in HTML, but the original (possibly camelCase) name — e.g. soCalled — is recovered from the source XML so the element serializes back with its original spelling.

Notifications

When unknown elements are encountered, a toast reports how many element types were preserved (see Events → toasts).

Why this matters

It means loading a document into JinnTap is non-destructive even if the schema is incomplete: unrecognised markup survives the edit/save cycle instead of being thrown away. To make an element first-class (with a toolbar button, shortcuts, specific content model), add it to your schema.