DEPR: remove Int|Uint|Float64Index from conftest & _testing #49678
Conversation
| "int64", | ||
| "int32", |
There was a problem hiding this comment.
just for my understanding, why don't 'int16' and 'int8' need to be here as well? (likewise for other tests)
this isn't a request to change it, just trying to understand
There was a problem hiding this comment.
I thought it was a bit strange to test these things here, and suspect many of the tests here should be located in pandas.tests.indexes.common.Base and adapted in its subclasses as needed. At the same time I didn't really want to delve too deep into refactoring the code base, as it would distract from my main objective right now i.e. getting rid of the old numeric indexes. I probably shouldn't even have added the 32bit versions, but I was curious if it would still pass or some issue would show up.
How about I write an issue about this refactoring and fix it after I've removed Int64Index & friends?
There was a problem hiding this comment.
Thanks, I've written a issue, see #49679.
mroeschke
added
Testing
|
Thanks @topper-123 |
…ev#49678) * DEPR: remove Int|Uint|Float64Index from conftest + related * fix failures * fix flake8 Co-authored-by: Terji Petersen <terjipetersen@Terjis-Air.fritz.box>
…ev#49678) * DEPR: remove Int|Uint|Float64Index from conftest + related * fix failures * fix flake8 Co-authored-by: Terji Petersen <terjipetersen@Terjis-Air.fritz.box>
Progress towards #42717.
In particular cleans up conftest.py::indices_dict to not use old style indexes and cleans up the
tm.make*Indexfunctions to not return old style numeric indexes.I've kept
tm.MakeIntIndexetc. and just made them tighter versions oftm.makeNumericIndexthough if they should be deleted that is also fine for me.