OpenDAP loaded Dataset has lon/lats with type 'object'.

[akleeman]

ds = xray.open_dataset('http://motherlode.ucar.edu/thredds/dodsC/grib/NCEP/GFS/Global_0p5deg/files/GFS_Global_0p5deg_20140303_0000.grib2', decode_cf=False)
In [4]: ds['lat'].dtype
Out[4]: dtype('O')

This makes serialization fail.

[shoyer]

This is because coordinates are loaded as pandas.Index objects... which don't always faithfully preserve the type of the underlying object (see pandas-dev/pandas#6471).

I believe serialization should still work though thanks to a work around I added for dtype=object. Do let me know if this is not the case. One solution to make this less awkward would be to wrap pandas.Index in something that keeps track of the dtype of the original arguments for use in mathematical expression.

[ebrevdo]

Indices also have an .inferred_type getter. unfortunately it doesn't seem
to return true type names...

In [13]: pandas.Index([1,2,3]).inferred_type
Out[13]: 'integer'

In [14]: pandas.Index([1,2,3.5]).inferred_type
Out[14]: 'mixed-integer-float'

In [15]: pandas.Index(["ab","cd"]).inferred_type
Out[15]: 'string'

In [16]: pandas.Index(["ab","cd",3]).inferred_type
Out[16]: 'mixed-integer'

On Sun, Mar 2, 2014 at 10:14 PM, Stephan Hoyer notifications@github.comwrote:

This is because coordinates are loaded as pandas.Index objects... which
don't always faithfully preserve the type of the underlying object (see
pandas-dev/pandas#6471 pandas-dev/pandas#6471).

I believe serialization should still work though thanks to a work around I
added for dtype=object. Do let me know if this is not the case. One
solution to make this less awkward would be to wrap pandas.Index in
something that keeps track of the dtype of the original arguments for use
in mathematical expression.

Reply to this email directly or view it on GitHubhttps://github.com/akleeman/xray/issues/39#issuecomment-36484122
.

[akleeman]

@shoyer You're right I can serialize the latitude object directly from that opendap url ... but after some manipulation I run into this:

ipdb> print fcst
dimensions:
    latitude = 31
    longitude = 46
    time = 7
variables:
    object latitude(latitude)
        units:degrees_north
        _CoordinateAxisType:Lat
    object longitude(longitude)
        units:degrees_east
        _CoordinateAxisType:Lon
    datet... time(time)
        standard_name:time
        _CoordinateAxisType:Time
        units:hours since 2014-03-03 00:0...
ipdb> fcst.dump('./test.nc')
*** TypeError: illegal primitive data type, must be one of ['i8', 'f4', 'u8', 'i1', 'U1', 'S1', 'i2', 'u1', 'i4', 'u2', 'f8', 'u4'], got object

Currently tracking down exactly whats going on here.

[shoyer]

I believe this was fixed by #54.