Roadmap¶
- Make
coords
andattrs
persistent across slicing operations (trydf['pop'].to_xarray()
from above example and watch it fail…) - Improve hooks between
pandas
andmetacsv
:- update
coord
names ondf.index.names
assignment - update
coords
on stack/unstack - update
coords
on
- update
- Handle attributes indexed by coord/variable names –> assign to
coord/variable-specific
attrs
- Let’s start an issue tracker and get rid of this section!
- Should we rethink “special attributes,” e.g. coords? Maybe these should
have some special prefix like
_coords
when included in yaml headers to avoid confusion with other generic attributes… - Allow special attributes (
coords
,variables
) inread_csv
call - Allow external file headers
- Write tests
- Write documentation
- Maybe steal xarray’s coordinate handling and save ourselves a whole lotta work?
Feature Requests¶
- Create syntax for
multi-csv
–>Panel
or combining using filename regex - Eventually? allow for on-disk manipulation of many/large files with dask/xarray
- Eventually? add xml, SQL, other structured syntax language conversions