Driver Options

TileDB Config Options

Option
Required
Description
uri
Yes
URI of a TileDB sparse or dense array (required)
tiledb.*
No
Set a TileDB configuration parameter, e.g., option("tiledb.vfs.num_threads", 4).

Writing Options

Option
Required
Description
write_buffer_size
No
Set the TileDB read buffer size in bytes per attribute/coordinates. Defaults to 10MB
schema.dim.<D>.name
Yes
Specify which of the Spark dataframe columns will be dimension D.
schema.dim.<D>.min
No
Specify the lower bound for the TileDB domain on dimension D.
schema.dim.<D>.max
No
Specify the upper bound for the TileDB domain on dimension D.
schema.dim.<D>.extent
No
Specify the tile extent on dimension D.
schema.attr.<A>.filter_list
No
Specify a filter list for attribute A. The filter list is a list of tuples of the form (name, option), ex: "(byteshuffle, -1), (gzip, 9)".
schema.capacity
No
Specify the tile capacity for sparse fragments.
schema.tile_order
No
Specify the tile order.
schema.cell_order
No
Specify the cell order.
schema.coords_filter_list
No
Specify the coordinates filter list. The filter list is a list of tuples of the form (name, option), ex: "(byteshuffle, -1), (gzip, 9)".
schema.offsets_filter_list
No
Specify the offsets filter list. The filter list is a list of tuples of the form (name, option), ex: "(byteshuffle, -1), (gzip, 9)".

Reading Options

Option
Required
Description
order
No
Result layout order. It can be "row-major"/ "TILEDB_ROW_MAJOR", "col-major" / "TILEDB_COL_MAJOR", "global-order" / "TILEDB_GLOBAL_ORDER", or "unordered"/ "TILEDB_UNORDERED" (default "unordered").
read_buffer_size
No
Set the TileDB read buffer size in bytes per attribute/coordinates. Defaults to 10MB
allow_read_buffer_realloc
No
If the read buffer size is too small allow reallocation. Default: True
partition_count
No
Number of partitions.