Fix for scrambled order of columns

[PallHaraldsson]

Fixes: #72

[PallHaraldsson]

Hi, @AStupidBear a fix here I think. I copied latter commit from another package, it was slightly different with: DataStructures = "0.17, 0.18"

[PallHaraldsson]

Just merge this? I'm pretty sure this works. My "but is there a better way?" RFC was about efficiency, while not too important here.

[tkf]

I think it's better to wait for @malmaud (although technically I can hit the merge button).

Until then, I think adding a test would help to smooth the review. Maybe test that the order of columns is preserved with the data frames with columns (say) a, b, c and c, b, a?

Also, since Pandas.jl needs PyCall.jl anyway, I guess we can use PyDict to avoid relying on OrderedCollections.jl here just for one line? Alternatively, it also should be possible to add a column to pandas.DataFrame one-by-one.

[PallHaraldsson]

If I get my way and Dict will be OrderedDict in Base, then this wouldn't be needed, as of next Julia LTS. But the dependency is light, and may even end up as a stdlib, so I wouldn't worry about that part.

About a test: OrderedDict means the order couldn't be different.

[malmaud]

Thanks! I agree with @tkf - this needs a test and I'd rather avoid adding a dependency.

[PallHaraldsson]

I did try to avoid this cheap dependency. I probably spent way more time on that (and also the rest) than I have time for now, but couldn't see a good way. It's at least very simple this way, and I'm already using it by using my own fork, since not merged here. My priority now, is well my job, and I already made the PR to JuliaLang to make OrderedDict the default, but it may need followup/convincing people, as I want it in Julia 1.6 but if wouldn't help here for users on older versions.

[codecov-commenter]

Codecov Report

❗ No coverage uploaded for pull request base (main@bf5ddfd). Click here to learn what that means.
The diff coverage is n/a.

@@           Coverage Diff           @@
##             main      #73   +/-   ##
=======================================
  Coverage        ?   66.00%           
=======================================
  Files           ?        5           
  Lines           ?      253           
  Branches        ?        0           
=======================================
  Hits            ?      167           
  Misses          ?       86           
  Partials        ?        0           

Flag	Coverage Δ
unittests	66.00% <0.00%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update bf5ddfd...6910720. Read the comment docs.

[malmaud]

Thanks Pall, I just added a test and will merge this now.