id,node_id,number,state,locked,title,user,body,created_at,updated_at,closed_at,merged_at,merge_commit_sha,assignee,milestone,draft,head,base,author_association,auto_merge,repo,url,merged_by
21565728,MDExOlB1bGxSZXF1ZXN0MjE1NjU3Mjg=,236,closed,0,WIP: convert to/from cdms2 variables,1217238,"Fixes #133

@DamienIrving am I missing anything obvious here?
",2014-09-22T08:48:52Z,2014-12-19T09:11:42Z,2014-12-19T09:11:39Z,2014-12-19T09:11:39Z,d1099053c9ebbb1169f72fb301913899453fd5ef,,836999,0,784e24435c2cfddc3cdaab1866d2d8d7ee8eadd5,731a0e2b20fac802a6ce7899b1bb3ba14eb3cad0,MEMBER,,13221727,https://github.com/pydata/xarray/pull/236,
23436973,MDExOlB1bGxSZXF1ZXN0MjM0MzY5NzM=,269,closed,0,Fixes for dtype casting,1217238,"Fixes #264 
Fixes #259 (casting of masked arrays)
",2014-10-28T06:37:55Z,2014-10-28T06:47:50Z,2014-10-28T06:47:38Z,2014-10-28T06:47:38Z,9d4bef67bfa3108f605912a5e3a01b821f6a2ab4,,836999,0,6eacd47b9575fb44792f225494834497e1c7978c,ee1369f72f20dad6331aa96710cc9f3237b195ec,MEMBER,,13221727,https://github.com/pydata/xarray/pull/269,
23559561,MDExOlB1bGxSZXF1ZXN0MjM1NTk1NjE=,270,closed,0,Fix guess_time_units if NaT is present,1217238,,2014-10-29T19:54:05Z,2014-10-29T20:09:18Z,2014-10-29T20:07:02Z,2014-10-29T20:07:02Z,6a2966925b4cfde53c8d9182156fab1f0c97070b,,836999,0,5c8a89ff4e781e22c3a1ef3f97564d71819e912f,4e06b33fd9e2848caae50ca48c126a749bf2d9a5,MEMBER,,13221727,https://github.com/pydata/xarray/pull/270,
23575223,MDExOlB1bGxSZXF1ZXN0MjM1NzUyMjM=,271,closed,0,Add backends integration test for roundtripping datetime arrays,1217238,"Not sure how we missed this before.

Also includes a fix for writing datetime arrays with missing values.
",2014-10-30T00:19:58Z,2014-10-30T02:32:23Z,2014-10-30T02:32:21Z,2014-10-30T02:32:21Z,faa1de8ebcd6a20348c86df717dfd0bda2f019c7,,836999,0,4e31557f9a55af9926c0bd14b9426728c8ae25d9,fdb4d455937636bb3d1722b78009b23341b77f78,MEMBER,,13221727,https://github.com/pydata/xarray/pull/271,
25319567,MDExOlB1bGxSZXF1ZXN0MjUzMTk1Njc=,282,closed,0,Serialize non-index coordinates to netCDF files,1217238,"Fixes #231 

This PR ensures non-index coordinates can be roundtripped to netCDF and back by using the CF convention of the 'coordinates' attribute on a ""data variable"".

In cases where there are no data variables available, we fall back to the (extended, non-CF) convention of saving 'coordinates' as a global attribute.

I also added the `xray.decode_cf` function to the API. This function can decode either datasets or datastores into decoded datasets, which is handy, for example, if a dataset is serialized according to CF conventions but already in memory (e.g., saved in JSON).

@akleeman: I could use some guidance from you on want to do about `test_roundtrip_coordinates` on `TestCFEncodedDataStore`. I am tentatively skipping it, but it would be nice to get it working. To do so will require some tweaks to the API for the `Datastore.load` method.
",2014-12-02T05:11:23Z,2014-12-07T02:44:14Z,2014-12-07T02:44:11Z,2014-12-07T02:44:11Z,8b25628a5cad01444dc957c44399bcbcc577af18,,836999,0,e1000d680cc9fee011ec0eff29b3ae2f9017d7f9,a5837164f80a4bffcfda52a3edcc4f40c947d336,MEMBER,,13221727,https://github.com/pydata/xarray/pull/282,
25547586,MDExOlB1bGxSZXF1ZXN0MjU1NDc1ODY=,285,closed,0,Fix to_netcdf with 0d sting variables,1217238,"Fixes #284
",2014-12-05T03:29:54Z,2014-12-05T03:57:18Z,2014-12-05T03:57:16Z,2014-12-05T03:57:16Z,353d1bf952659ef188568a7a9d490c29882b34d7,,836999,0,f8022ac535c4854c9d123ef39f1dab74f1c8edba,a5837164f80a4bffcfda52a3edcc4f40c947d336,MEMBER,,13221727,https://github.com/pydata/xarray/pull/285,
25627557,MDExOlB1bGxSZXF1ZXN0MjU2Mjc1NTc=,287,closed,0,Fixes for Dataset.to_dataframe(),1217238,"Fixes #278
",2014-12-07T03:24:55Z,2014-12-07T06:22:51Z,2014-12-07T06:22:47Z,2014-12-07T06:22:47Z,94033ffe16a48f7ad15ed91b66735a05bc5be5e8,,836999,0,69ea1a09b90614242adca4314f827f437bbefb4b,b9b9b9c984c2ea0d796f6d6d45e15a63d32b5926,MEMBER,,13221727,https://github.com/pydata/xarray/pull/287,
25630036,MDExOlB1bGxSZXF1ZXN0MjU2MzAwMzY=,288,closed,0,Add optional dependency on cyordereddict,1217238,"This makes indexing a dataset (`ds['foo']`) about 30% faster in cases where `cyordereddict` is available.
",2014-12-07T09:15:35Z,2014-12-07T09:53:51Z,2014-12-07T09:53:49Z,2014-12-07T09:53:49Z,8c2715459fb7fa991a2c3fa167c6d9ae4d073b73,,836999,0,07823e96533e2dc6c870d3a372e2ee283f7312d5,2a8f1860d93ea9e2635044ad44dc859c67a6ac90,MEMBER,,13221727,https://github.com/pydata/xarray/pull/288,
25798861,MDExOlB1bGxSZXF1ZXN0MjU3OTg4NjE=,289,closed,0,Faster path for Dataset/DataArray indexing,1217238,"My benchmark shows this commit makes things about 35% faster
for positional indexing.
",2014-12-10T04:48:31Z,2014-12-10T04:53:17Z,2014-12-10T04:53:15Z,2014-12-10T04:53:15Z,fe1ce8fbc4aa366e37ab6d72b04914359f581528,,836999,0,a5528b44a95f4973ebe6546b1f143da1c8c19886,8e56736a87e53322c5bd8e3e20cf92ec352548f2,MEMBER,,13221727,https://github.com/pydata/xarray/pull/289,
25802877,MDExOlB1bGxSZXF1ZXN0MjU4MDI4Nzc=,291,closed,0,Support using dictionaries for labeled indexing,1217238,"Fixes #187

As of this change, you can use dictionaries for indexing along a dimension, e.g., in the form `array.loc[dict(x='2000-01-01')]` or `array[dict(x=0)]`.

In particular, you can use this for _assignment_, not just for selecting values (e.g., `array[dict(x=0)] = 0`). It was previously difficult to do assignment with labeled dimensions in xray.
",2014-12-10T07:14:47Z,2014-12-11T22:28:26Z,2014-12-11T22:28:19Z,2014-12-11T22:28:19Z,4564915e1fdf08baed1def669d8f453393aae66a,,836999,0,967809a1217694a31fd9b3c918ecb2bea20b5411,412215858a6c14b1354bf1a7c91c0177189a06c2,MEMBER,,13221727,https://github.com/pydata/xarray/pull/291,
25958168,MDExOlB1bGxSZXF1ZXN0MjU5NTgxNjg=,292,closed,0,Allow for pd.TimedeltaIndex and serialize it to netCDF,1217238,"Fixes #55
",2014-12-12T08:19:32Z,2014-12-12T09:41:03Z,2014-12-12T09:41:01Z,2014-12-12T09:41:01Z,81bbf9ef2643303d3e595a1d090c4610e7aa3416,,836999,0,2eee7df09dd0888569497ac227b969c8ba1faa21,a31e0e5f85336aa9da87dacf93e765007b3a60cb,MEMBER,,13221727,https://github.com/pydata/xarray/pull/292,
26349640,MDExOlB1bGxSZXF1ZXN0MjYzNDk2NDA=,294,closed,0,Coerce arrays of datetime.datetime objects into datetime64,1217238,"Fixes #136

Also includes some speedups for Variable construction
",2014-12-19T03:39:35Z,2014-12-19T05:27:51Z,2014-12-19T05:27:50Z,2014-12-19T05:27:50Z,dbf5fd51845c0510422243fbcfb5a374c3243470,,836999,0,3925fda821688b8d824b277b0d82798e83dc40e6,a9f63ffbee4bc929c29d682c3a119db10d11c78e,MEMBER,,13221727,https://github.com/pydata/xarray/pull/294,
26355014,MDExOlB1bGxSZXF1ZXN0MjYzNTUwMTQ=,295,closed,0,Cleanup & bug fix,1217238,"Fixes #279.
",2014-12-19T07:17:35Z,2014-12-19T07:21:33Z,2014-12-19T07:21:31Z,2014-12-19T07:21:31Z,3bdb8b84ea79a23ca73162aa67880e0ed4fd8d27,,836999,0,889d80332d34b597070ee32211475266e2e86c9f,0c7e1b168e05052a0bfcabb16588dec6aba6d9bf,MEMBER,,13221727,https://github.com/pydata/xarray/pull/295,
26410800,MDExOlB1bGxSZXF1ZXN0MjY0MTA4MDA=,296,closed,0,BUG: fix string <-> char conversion for non-contiguous arrays,1217238,,2014-12-20T02:23:57Z,2014-12-20T02:44:03Z,2014-12-20T02:44:01Z,2014-12-20T02:44:01Z,b93a1de8621879bca62e4ea86c2634d6db63fc44,,836999,0,543fa95e22e89be45c5c712feaa3dd2a5cb7334b,ad6d6a1f0c9c2558e0388041a351914ce4f08bd2,MEMBER,,13221727,https://github.com/pydata/xarray/pull/296,
26443413,MDExOlB1bGxSZXF1ZXN0MjY0NDM0MTM=,297,closed,0,Better decoding of netCDF times,1217238,"This version avoids using the netCDF4 library unless strictly
necessary (e.g., for non-standard calendars).
",2014-12-22T08:03:02Z,2014-12-22T20:09:43Z,2014-12-22T20:09:41Z,2014-12-22T20:09:41Z,83aa2c13cfe17e0a5b64314aa85ee2cf4e9529f7,,836999,0,14466e68f1ccdbaf428833aec04d8067379365a4,5ac8205d4d6475b6244ebd569b7e7beae1ca7b66,MEMBER,,13221727,https://github.com/pydata/xarray/pull/297,
26443536,MDExOlB1bGxSZXF1ZXN0MjY0NDM1MzY=,298,closed,0,Fix DataArray.__init__ docstring,1217238,"xref: #277
",2014-12-22T08:07:10Z,2014-12-22T08:25:15Z,2014-12-22T08:25:15Z,2014-12-22T08:25:15Z,57814c8dd2b192c42350337e9be78e3af1df54b7,,836999,0,5797d2eb93dab82891b3e3aae607f7245dd3cfa2,5ac8205d4d6475b6244ebd569b7e7beae1ca7b66,MEMBER,,13221727,https://github.com/pydata/xarray/pull/298,
26477924,MDExOlB1bGxSZXF1ZXN0MjY0Nzc5MjQ=,299,closed,0,Update docs,1217238,,2014-12-22T20:09:54Z,2014-12-22T21:00:58Z,2014-12-22T21:00:56Z,2014-12-22T21:00:56Z,d167c84f0f696a87423acbb7897035f32f0fdcc1,,836999,0,20c66b3dbd1481f7e1e09cdd46603cd321c4717b,013bbabfc0b3ac7606d4102efe231996c6c65da8,MEMBER,,13221727,https://github.com/pydata/xarray/pull/299,
26546974,MDExOlB1bGxSZXF1ZXN0MjY1NDY5NzQ=,301,closed,0,Add attribute style access for variables and attributes,1217238,"Fixes #286
",2014-12-24T06:41:22Z,2014-12-24T07:07:25Z,2014-12-24T07:07:24Z,2014-12-24T07:07:24Z,646579f51bf579608a76658f7882fe14ecda7ae8,,836999,0,5092b3c51711ede11558f5524522b37972e1f2c3,b1c8cdd46df56f1048c3606ef0f45ae3312ca2dd,MEMBER,,13221727,https://github.com/pydata/xarray/pull/301,