DataFrames 0.11 released

Specific Domains Data
22

That’s because you have other packages installed that aren’t compatible with the new version. It will arrive in time.

3 Likes
23

How can I find all packages that depend on DataFrames?
I tried rm and then add, but it pulled the old version (0.10.1). I would like to delete whatever package is relying on DataFrames…

24 
help?> Pkg.dependents
  dependents(pkg)

  List the packages that have pkg as a dependency.
1 Like
25

Thanks!
I did not find this here
I guess I simply need to compare this list with Pkg.installed()

EDIT: in my case it was RData,SQLite, CSV and RDatasets.

26

A small comment from my side.
I have updated a module of mine which relied on PooledDataArrays.
It was quite easy to change to PooledArray.
However, some of my code ran into an error, because the behaviour of the levels function has changed.

In the absence of missing entries it is now equal to unique(your data) (missings.jl row 389), whereas before it was referring to the pool of the PooledDataArray.

For my purposes I have thus defined this:

function Missings.levels(x::PooledArray)
    return deepcopy(x.pool)
end
27

Yes, better use CategoricalArray if you want to use levels. I’m not sure whether PooledArray wants to support it (maybe worth filing an issue).

28

In case anyone runs into the same problem: the only thing that prevented upgrading for me was RDatasets, and I can’t figure out why so I don’t know for which package I should open an issue (or check if it is already there).

29

There’s a PR pending for RDatasets:
https://github.com/johnmyleswhite/RDatasets.jl/pull/48

2 Likes