Related to #92, there may be cases where the data is small enough to load into memory. If this is done within .to_xarray() before the object is constructed, then there's enough FST information to load the data in the natural order on disk, speeding up the reading process.
So, if someone is intending to pre-load the data anyway, e.g.
ds = Buffer("file.fst").to_xarray().load()Then it may be more efficient to preemptively do this via something like:
ds = Buffer("file.fst").to_xarray(type="numpy")Or from open_dataset:
ds = xr.open_dataset("file.fst",engine="fstd",type="numpy")
Related to #92, there may be cases where the data is small enough to load into memory. If this is done within
.to_xarray()before the object is constructed, then there's enough FST information to load the data in the natural order on disk, speeding up the reading process.So, if someone is intending to pre-load the data anyway, e.g.
Then it may be more efficient to preemptively do this via something like:
Or from open_dataset: