Data.update() fails with 'TypeError: 'list' object is not callable'

[j-mesnil]

🐛 Describe the bug

The data.Data.update() method fails in the following minimal example.

import torch
from torch_geometric.data import Data

x1 = torch.Tensor([[1.0, 0.0]])
x2 = torch.Tensor([[0.0, 1.0]])
d1 = Data(x=x1)
d2 = Data(x=x2)
d2.update(d1)

The following exception is thrown:

Traceback (most recent call last):
  File "(/path/to/minimal_example.py)", line 8, in <module>
    d2.update(d1)
  File "/usr/lib/python3.8/_collections_abc.py", line 834, in update
    for key in other.keys():
TypeError: 'list' object is not callable

The update method in _collections_abc.py will run

elif hasattr(other, "keys"):
    for key in other.keys():
        self[key] = other[key]

The Data class from PyG has a list of keys keys, so the call to other.keys() fails here.

Environment

• PyG version: 2.2.0
• PyTorch version: 1.12.1+cu113
• OS: Linux
• Python version: 3.8
• CUDA/cuDNN version: 11.1
• How you installed PyTorch and PyG (conda, pip, source): pip
• Any other relevant information (e.g., version of torch-scatter):

[rusty1s]

We just recently added support for Data.update(), see #6313. Should be fixed in PyG master and upcoming releases.

[j-mesnil]

I see, thanks!