Skip to content

Conversation

@cpcloud
Copy link
Member

@cpcloud cpcloud commented Jun 3, 2014

@cpcloud cpcloud self-assigned this Jun 3, 2014
@cpcloud cpcloud added this to the 0.14.1 milestone Jun 3, 2014
@jreback
Copy link
Contributor

jreback commented Jun 3, 2014

passes for me

@cpcloud
Copy link
Member Author

cpcloud commented Jun 3, 2014

Yeah but not passing on Travis :(

@cpcloud
Copy link
Member Author

cpcloud commented Jun 3, 2014

Some tuple edge case

@cpcloud
Copy link
Member Author

cpcloud commented Jun 4, 2014

@jreback can u take a look here? previously this was sorting (since tuples are sortable), then comparing against a MultiIndex. So this is a breaking API change, albeit for non MultiIndex tuple Indexes, which are rare.

@cpcloud
Copy link
Member Author

cpcloud commented Jun 4, 2014

the change is that sorting mixed dtype indexes doesn't work for any objects, including things that do have an ordering.

@cpcloud
Copy link
Member Author

cpcloud commented Jun 4, 2014

@jreback ok with this?

@jreback
Copy link
Contributor

jreback commented Jun 4, 2014

seems ok, I don't really understand what its supposed to do. doubt this is really an issue in practice (except the error of course!) so ok

@cpcloud
Copy link
Member Author

cpcloud commented Jun 4, 2014

tuples have an ordering which means they can be sorted inplace no problem, but i check for mixed and mixed-integer inferred_type which tuples satisfy. i thought it was better to not sort anything that's mixed, than to special case tuples

cpcloud added a commit that referenced this pull request Jun 4, 2014
BUG: already mixed indexes should not sort
@cpcloud cpcloud merged commit 92bc165 into pandas-dev:master Jun 4, 2014
@cpcloud cpcloud deleted the fix-unorderables-mixed-union branch June 4, 2014 20:53
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Bug Indexing Related to indexing on series/frames, not to indexes themselves

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants