Cache-warming entry point for the L0 proxy: warmChunk.
A warm-up only wants the chunk decoded and resident in L0 (and its bytes in
L1/L2). Routing it through zarrita get() would allocate and fill a full
output selection only to throw it away, and — because get() calls the
proxy's getChunk — would read the owning loader's DEMAND abort signal and
record into its DEMAND residency probe. warmChunk is the proxy method that
does neither: it joins/starts the cache-wide decode with the caller's own
signal, no probe, and (by default) origin 'prefetch'.
Kept dependency-free (no DecompressedChunkCache import) so data-layer
helpers can call it without pulling the cache module into their import
graph; the proxy implements the method.
Cache-warming entry point for the L0 proxy:
warmChunk.A warm-up only wants the chunk decoded and resident in L0 (and its bytes in L1/L2). Routing it through zarrita
get()would allocate and fill a full output selection only to throw it away, and — becauseget()calls the proxy'sgetChunk— would read the owning loader's DEMAND abort signal and record into its DEMAND residency probe.warmChunkis the proxy method that does neither: it joins/starts the cache-wide decode with the caller's own signal, no probe, and (by default) origin'prefetch'.Kept dependency-free (no
DecompressedChunkCacheimport) so data-layer helpers can call it without pulling the cache module into their import graph; the proxy implements the method.