Perspective layering — see Sanity Content Lake docs. The engine
threads this into reads that should reflect a release-stacked view
of the dataset (field-entry query resolution, spawn forEach discovery).
Engine-internal reads of workflow.instance / workflow.definition
stay raw regardless — those docs are unversioned metadata.
Either both the real @sanity/client and the fake test client
support the same shape. string[] is an ordered release-name stack;
"drafts" may appear in the array. "published" is the implicit
tail.
Perspective layering — see Sanity Content Lake docs. The engine threads this into reads that should reflect a release-stacked view of the dataset (field-entry query resolution, spawn forEach discovery). Engine-internal reads of
workflow.instance/workflow.definitionstay raw regardless — those docs are unversioned metadata.Either both the real
@sanity/clientand the fake test client support the same shape.string[]is an ordered release-name stack;"drafts"may appear in the array."published"is the implicit tail.