LifeLab · nested transclusion
Transcluding a page does not auto-expand its children. You see that page's surface as authored: its blocks, plus its child page-refs in their stored display state (usually collapsed lines) and a one-line count of unplaced shelf children. From there you unfold deliberately, level by level — canonical subpage refs nest to any depth (the parent→child tree is acyclic by construction), while summoned refs are always depth 1: anything they transclude renders as a plain [[link]].
A region shows one page's own surface. Its children appear as collapsed lines / a shelf count — unfold on demand, per surface.
Subpage refs follow the parent→child tree — no cycles possible. Summons are arbitrary edges, so they stay shallow windows.
Each nesting level lifts onto a lighter sheet with a longer shadow. Two sheets = two pages deep. Stars stay the block hierarchy — separate systems.
Past ~3 sheets the region header nudges ⌥↵ jump over further unfolding — change buffers instead of building a pane telescope.
[[acl2026/mellms-workshop]] unfolded on the conference page. Inside: its own blocks, its two session subpages as collapsed lines (their stored state on the workshop page), and its unplaced shelf remainder as a single count line. Nothing below depth 1 has rendered yet.
Tab on [[session-1-morning-papers]]: a second, lighter sheet lifts inside the workshop's. Paper blocks at depth 2 are fully native — the add line homes new blocks on the session page. You're editing three pages on one surface, and the sheet under your cursor tells you which one at a glance.
[[reading-group/queue]] summoned onto the conference page. The queue page itself transcludes [[reading-group/archive]] — but through a summon window that inner region degrades to a plain [[link]] line. No unfold affordance, no cycles possible: A→B→A dead-ends as a link. Same rule if the summon sits nested inside a subpage region.