empty_like for object array raises warning

[jeromekelleher]

    import zarr
    z = zarr.empty(shape=(2,), dtype="array:u1")
    z[0] = [0, 1, 2]
    z[1] = [0, 0]
    print(z, z[:])
    y = zarr.empty_like(z) # Emits warning
    y[0] = [0, 1, 2]
    y[1] = [0, 0]
    print(y, y[:])

Gives:

<zarr.core.Array (2,) object> [array([0, 1, 2], dtype=uint8) array([0, 0], dtype=uint8)]
/home/jk/.local/lib/python3.5/site-packages/zarr/storage.py:372: FutureWarning: missing object_codec for object array; this will raise a ValueError in version 3.0
  'ValueError in version 3.0', FutureWarning)
<zarr.core.Array (2,) object> [array([0, 1, 2], dtype=uint8) array([0, 0], dtype=uint8)]

Problem description

It seems that empty_like is somehow not copying the codec? However, the array seems to behave perfectly well as a object array, so I'm not sure what's happening.

Version and installation information

Please provide the following:

• Value of zarr.__version__: 2.2.0
• Value of numcodecs.__version__: 0.5.3
• Version of Python interpreter: 3.5.2
• Operating system (Linux/Windows/Mac): Linux
• How Zarr was installed (e.g., "using pip into virtual environment", or "using conda"):pip

[alimanfoo]

Thanks Jerome, this looks like just an API inconsistency, I didn't properly consider the ..._like creation functions when writing the logic for how the object_codec keyword argument is handled. We should fix this so that no warning is emitted. For now you can ignore the warning, the array should be fine.

[hyanwong]

@alimanfoo FWIW I also see this when running copy_all (zarr-2.3.2), Max OS X

  File "/Users/yan/Documents/Research/BDI/tsinfer-hyanwong/tsinfer/formats.py", line 358, in copy
    zarr.copy_all(source=self.data, dest=other.data)
  File "/usr/local/lib/python3.7/site-packages/zarr/convenience.py", line 1069, in copy_all
    dry_run=dry_run, **create_kws)
  File "/usr/local/lib/python3.7/site-packages/zarr/convenience.py", line 962, in _copy
    dry_run=dry_run, **create_kws)
  File "/usr/local/lib/python3.7/site-packages/zarr/convenience.py", line 891, in _copy
    dtype=source.dtype, **kws)
  File "/usr/local/lib/python3.7/site-packages/zarr/hierarchy.py", line 770, in create_dataset
    return self._write_op(self._create_dataset_nosync, name, **kwargs)
  File "/usr/local/lib/python3.7/site-packages/zarr/hierarchy.py", line 628, in _write_op
    return f(*args, **kwargs)
  File "/usr/local/lib/python3.7/site-packages/zarr/hierarchy.py", line 783, in _create_dataset_nosync
    **kwargs)
  File "/usr/local/lib/python3.7/site-packages/zarr/creation.py", line 120, in create
    chunk_store=chunk_store, filters=filters, object_codec=object_codec)
  File "/usr/local/lib/python3.7/site-packages/zarr/storage.py", line 323, in init_array
    object_codec=object_codec)
  File "/usr/local/lib/python3.7/site-packages/zarr/storage.py", line 383, in _init_array_metadata
    'ValueError in version 3.0', FutureWarning)
FutureWarning: missing object_codec for object array; this will raise a ValueError in version 3.0

[alimanfoo]

Thanks @hyanwong, very useful to know. I think you can ignore the warning but we should fix this so it doesn't get emitted.

[hyanwong]

Yep thanks. Was ignoring it anyway :)

[hyanwong]

It would be great to fix this if possible, @alimanfoo ! But I guess you are pretty busy at the moment?

[jakirkham]

Contributions welcome :)

[hyanwong]

Happy to have a go, but not sure if I should just hack around it (i.e. omit this FutureWarning in this case), or if it's more complex than that. I imagine it's the latter!

[hyanwong]

Here's the warning (turned into an error for debugging) that I'm getting with the current master GitHub branch:

Any clues as to where/how to fix this welcome, and I can try to implement it @jakirkham

  File "/usr/local/anaconda3/lib/python3.8/site-packages/tsinfer/formats.py", line 497, in copy
    zarr.copy_all(source=self.data, dest=other.data)
  File "/usr/local/anaconda3/lib/python3.8/site-packages/zarr/convenience.py", line 1058, in copy_all
    c, s, b = _copy(
  File "/usr/local/anaconda3/lib/python3.8/site-packages/zarr/convenience.py", line 951, in _copy
    c, s, b = _copy(
  File "/usr/local/anaconda3/lib/python3.8/site-packages/zarr/convenience.py", line 882, in _copy
    ds = dest.create_dataset(name, shape=source.shape,
  File "/usr/local/anaconda3/lib/python3.8/site-packages/zarr/hierarchy.py", line 806, in create_dataset
    return self._write_op(self._create_dataset_nosync, name, **kwargs)
  File "/usr/local/anaconda3/lib/python3.8/site-packages/zarr/hierarchy.py", line 661, in _write_op
    return f(*args, **kwargs)
  File "/usr/local/anaconda3/lib/python3.8/site-packages/zarr/hierarchy.py", line 818, in _create_dataset_nosync
    a = create(store=self._store, path=path, chunk_store=self._chunk_store,
  File "/usr/local/anaconda3/lib/python3.8/site-packages/zarr/creation.py", line 121, in create
    init_array(store, shape=shape, chunks=chunks, dtype=dtype, compressor=compressor,
  File "/usr/local/anaconda3/lib/python3.8/site-packages/zarr/storage.py", line 344, in init_array
    _init_array_metadata(store, shape=shape, chunks=chunks, dtype=dtype,
  File "/usr/local/anaconda3/lib/python3.8/site-packages/zarr/storage.py", line 418, in _init_array_metadata
    warnings.warn('missing object_codec for object array; this will raise a '
FutureWarning: missing object_codec for object array; this will raise a ValueError in version 3.0