Type introspection?

1

The collection/group API has some very basic introspection via tiledb.object_type() (in Python). Is there a means to perform further introspection, in particular:

  • determine if an array is sparse or dense
  • access schema properties (eg, attribute names and types)
    etc?

Or does the current API require that the application code remember the schema of any given object?

2

You can introspect an array by reading its array schema. In TileDB-Py:
https://docs.tiledb.io/projects/tiledb-py/en/latest/python-api.html#tiledb.libtiledb.Array.schema
https://docs.tiledb.io/projects/tiledb-py/en/latest/python-api.html#array-schema

New docs are coming up soon, so hopefully it will be easier to find this information :slight_smile:.

3

A quick example usage of opening the array and accessing the schema and printing the domain:

with tiledb.SparseArray("s3://mybucket/myarray") as A:
  schema = A.schema
  domain = schema.domain
  print(domain)
4

Thank you!

If there is no a priori knowledge of sparse/dense format, does the Array class support schema access, ie, is this legal? (it seems to work, but the docs suggest it might not work)

with tiledb.Array('path') as A:
    my_array_is_sparse = A.schema.sparse
5

Yeah, that should work.