randolf-scholz commented on PR #35210:
URL: https://github.com/apache/arrow/pull/35210#issuecomment-1546589412
@danepitkin I'd add `dictionary[string,int32]` since it is likely the most
common one. Maybe one should add `dictionary[int8,string]`,
`dictionary[int16,string]` and `dictionary[int64,string]` immediately as well?
How common are non-string dictionaries anyway?
If all possible combinations with non-string types are needed one could
write a custom lookup instead.
Regarding the representation one could do either:
- differentiate `DictionaryType.__repr__` and `DictionaryType.__str__`, so
that the latter returns the type-code
- add an additional `name` / `type_code` property to each dtype that returns
the abbreviated type code
- for example, numpy has `name` attribute
- `np.dtype("int32").name` ⇝ "int32"
- `np.dtype("datetime64[ns]").name` ⇝ "datetime64[ns]"
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
