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/4714#issuecomment-748498256,https://api.github.com/repos/pydata/xarray/issues/4714,748498256,MDEyOklzc3VlQ29tbWVudDc0ODQ5ODI1Ng==,14808389,2020-12-19T16:58:20Z,2020-12-19T16:58:20Z,MEMBER,"> I think `reindex` would need to be changed

that's true, I only tried the special case where the data that would be used to do the forward fill is included in the result.

> I guess this works but it's a bit cumbersome

yeah, `to_dataset` is probably not the right tool for pointwise indexing.

> it does not fail if one of the sensors in the query list is missing

if I understand correctly, you would like to index with arbitrary values for `time`, but would like an error for missing values of `sensor`. Unfortunately, I don't think that is possible using a single call to `sel`. Instead, you could set the `fill_value` parameter of `reindex` to some other value (for example, `-np.inf`) and then drop these values after the pointwise indexing.","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,771382653
https://github.com/pydata/xarray/issues/4714#issuecomment-748483357,https://api.github.com/repos/pydata/xarray/issues/4714,748483357,MDEyOklzc3VlQ29tbWVudDc0ODQ4MzM1Nw==,14808389,2020-12-19T14:41:47Z,2020-12-19T14:41:47Z,MEMBER,"`reindex` does not have to be changed since we can just call e.g. `ffill` with the `dim` parameter for this to work:
```python
arr.reindex(...).ffill(dim=""dim"")
```

This really depends on how you intend to use the result of the indexing. For example, if you don't really need the big matrix, you could just convert the `DataArray` to a `Dataset` where the `sensor` dimension is the names of the variables (using `to_dataset(dim=""sensor"")`, or construct it that way). If you do need the matrix, this might be slightly better (you still end up allocating a `T * (S + n)` array):
```python
arr.reindex(sensor=[""A"", ""B"", ""C""]).sel({""sensor"": ..., ""time"": ...}, method=""ffill"")
```
but if you really care about the memory allocated at once, you might be better off using `dask`:
```python
arr.chunk({""time"": 100}).reindex(...).sel(...)
```

If all of that is not an option, I guess we might be able add a `method_kwargs` parameter (not sure if there is a better option, though).","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,771382653