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/727#issuecomment-244229155,https://api.github.com/repos/pydata/xarray/issues/727,244229155,MDEyOklzc3VlQ29tbWVudDI0NDIyOTE1NQ==,1217238,2016-09-01T22:14:31Z,2016-09-01T22:14:31Z,MEMBER,"@choldgraf no worries!
","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,128903340
https://github.com/pydata/xarray/issues/727#issuecomment-175756338,https://api.github.com/repos/pydata/xarray/issues/727,175756338,MDEyOklzc3VlQ29tbWVudDE3NTc1NjMzOA==,1217238,2016-01-27T17:26:45Z,2016-01-27T17:26:45Z,MEMBER,"@choldgraf Yes, that looks right to me -- this code path in particular should trigger a DeprecationWarning:
```
# try to infer dimensions from coords
if utils.is_dict_like(coords):
dims = list(coords.keys())
```
","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,128903340
https://github.com/pydata/xarray/issues/727#issuecomment-175415408,https://api.github.com/repos/pydata/xarray/issues/727,175415408,MDEyOklzc3VlQ29tbWVudDE3NTQxNTQwOA==,1217238,2016-01-27T05:51:06Z,2016-01-27T05:51:06Z,MEMBER,"Any interest in putting together a PR? :)
On Tue, Jan 26, 2016 at 9:46 PM, Chris Holdgraf notifications@github.com
wrote:
> That sounds like a fair plan to me - throwing errors etc definitely
> complicates things more than just deprecating the option :)
>
> On Tue, Jan 26, 2016 at 8:58 PM, Joe Hamman notifications@github.com
> wrote:
>
> > So I would support deprecating supplying dict-like coordinates without
> > explicitly provided dims. It's more magic than we really need. This line
> > is
> > in the relevant code path.
> >
> > I've also run into this and I'm [image: :+1:] on this solution.
> >
> > —
> > Reply to this email directly or view it on GitHub
> > https://github.com/pydata/xarray/issues/727#issuecomment-175399404.
>
> —
> Reply to this email directly or view it on GitHub
> https://github.com/pydata/xarray/issues/727#issuecomment-175414737.
","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,128903340
https://github.com/pydata/xarray/issues/727#issuecomment-175390102,https://api.github.com/repos/pydata/xarray/issues/727,175390102,MDEyOklzc3VlQ29tbWVudDE3NTM5MDEwMg==,1217238,2016-01-27T04:38:44Z,2016-01-27T05:13:44Z,MEMBER,"I agree, this can be ambiguous. It makes sense for ordered mappings, but not for normal dicts. And unfortunately there is no Pythonic way to tell whether a mapping is ordered or not.
I originally included it because I wanted to make sure operations like `xr.DataArray(2 * x.data, x.coords)` preserve all original dimensions in order. But, given that we define `DataArray.coords.dims` and use that to fill in an empty `dims` argument, we don't really need this anymore. If someone really likes supplying coordinates as an OrderedDict, it's simple enough for them to write `coords.items()` instead.
So I would support deprecating supplying dict-like coordinates without explicitly provided `dims`. It's more magic than we really need. [This line](https://github.com/pydata/xarray/blob/66057e60201ac5a2e4cd37913220f2ea42e6f06e/xarray/core/dataarray.py#L39) is in the relevant code path.
I don't want to simply error when coordinates have ambiguous lengths compared to the size of the data, because adding errors in special cases makes things harder to predict.
","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,128903340