html_url,issue_url,id,node_id,user,created_at,updated_at,author_association,body,reactions,performed_via_github_app,issue
https://github.com/pydata/xarray/issues/6433#issuecomment-1085589487,https://api.github.com/repos/pydata/xarray/issues/6433,1085589487,IC_kwDOAMm_X85AtMfv,14808389,2022-04-01T08:21:36Z,2022-04-01T09:19:14Z,MEMBER,"`parallel=True` will *open* (read the metadata) and preprocess the datasets in parallel using `dask`, which will happen in sequence with `parallel=False`. The difference is [roughly](https://github.com/pydata/xarray/blob/305533d585389f7240ae2383a323337d4761d33a/xarray/backends/api.py#L896-L916):
```python
if parallel:
    datasets, = dask.compute([dask.delayed(xr.open_dataset)(path, *args, **kwargs) for path in paths])
else:
    datasets = [xr.open_dataset(path, *args, **kwargs) for path in paths]
```
So unlike `load_dataset` / the proposed `load_mfdataset` this will read metadata but not the actual data, and the result should be the same regardless of the `parallel` parameter (if it doesn't that would be a bug).

I agree that we could improve the parameter description, though.","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,1188965542