Skip to content

Conversation

@jreback
Copy link
Contributor

@jreback jreback commented Mar 16, 2013

  • replace on an IntBlock with a float will upcast, it may yield multiple blocks if the original IntBlock didn't fully replace
  • small fix in com.mask_missing to guarantee that a bool ndarray is returned even in the case of a string comparison with a numeric array, (which for some reason numpy returns a single, non-ndarray value)

allows #3064 to proceed (e.g. df.replace(0,0.5) will work on an integer dtype)

 e.g. a replace on an IntBlock with a float will upcast, it may yield multiple blocks if the original IntBlock didn't fully replace small fix in com.mask_missing to guarantee that a bool ndarray is returned even in the case of a string comparison with a numeric array (which for some reason numpy returns a single, non-ndarray value) BUG: py3 issue in com._maybe_promote for recognizing string types on number like dtypes!
jreback added a commit that referenced this pull request Mar 16, 2013
BUG/ENH: guarantee blocks will upcast as needed, and split as needed
@jreback jreback merged commit fbfd16a into pandas-dev:master Mar 16, 2013
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

1 participant