Refactor for DataFrames v0.21 compatibility #63
Conversation
|
Does the switching from |
The deprecation warning is not present using DataFrames v0.20, in DataFrames v0.21 there is a warning but the operation still works (probably to be fully removed in a later version). |
|
@danielolsen So the main reason we would like to switch as this point is for open source concerns, i.e have latest dependencies in our package? |
Exactly. Nothing is broken, but we would like to be using the latest and greatest versions of the dependencies, if it's not too much work. |
* refactor: use new DataFrames groupby syntax * refactor: use innerjoin instead of join for DataFrame * chore: update DataFrames compatibility from 0.20 to 0.21 * chore: update Manifest.toml for DataFrames 0.21.4
Purpose
Similar to #62, we refactor to be compatible with the most up-to-date version of a dependency. The fact that these both upgrade from 0.20 to 0.21 via breaking changes is coincidence.
What is the code doing
DataFrames.join, they are now done withDataFrames.innerjoin(distinct from the also-new.leftjoin,.outerjoin, etc.). See add inner/left/right/outer/semi/anti/cross-join JuliaData/DataFrames.jl#2101DataFrame.bytoDataFrame.combine+DataFrame.groupbyis not a breaking change, butDataFrame.byraises a DeprecationWarning in v0.21 (helpfully, the deprecation warning also suggests the proper new syntax!). Possibly related to [BREAKING] new design of select, transform and combine JuliaData/DataFrames.jl#2214 although I'm not 100% sure.Validation
Validated on a fresh install.
Time to review
The code itself is four lines. More time if you want to dig into the details.