Construct an in-memory array source

A memory source holds its values in the descriptor, so its revision fingerprint cannot be derived from a file state. It is instead derived from the shape, dtype, chunk grid, and a per-object identity token assigned once at construction, plus the optional caller-supplied revision. Construction therefore never reads or hashes the payload, whatever its size.

Usage

memory_source(
  data,
  dtype = NULL,
  chunks = NULL,
  revision = NULL,
  identity = c("object", "content")
)

Arguments

data

A two-dimensional matrix or array.

dtype

Logical storage dtype. Numeric R matrices default to "float64".

chunks

Optional logical chunk shape.

revision

Optional single string naming the revision of data, such as a version label or an upstream checksum. It is never interpreted, but it replaces the per-object identity token, so two sources built independently under the same revision share a fingerprint: the caller is asserting they are the same source in the same revision.

identity

"object" assigns a fresh identity token (unless revision is supplied); "content" derives it from content_hash() of data.

Value

A serializable memory_source.

Details

Two independently constructed memory sources with equal values have different fingerprints by design, and fingerprint equality is never value equality. A serialization round trip preserves the token, so a descriptor and its copies agree. Values compared across objects require content_hash(). Set identity = "content" to opt into a content-derived token: the payload is hashed once at construction and equal-valued sources then share a fingerprint at O(n) cost.

See also

source_fingerprint() and content_hash() for the fingerprint policy.

Examples

src <- memory_source(matrix(seq_len(6), nrow = 2))
source_shape(src)
#> [1] 2 3
source_dtype(src)
#> [1] "float64"