+1 on the content. This looks great!

Can I ask you to rebase in the whoopsie commit? Or perhaps all into one for something of this size?

So this is where I confess I'm not too good with git.

I think what you are suggesting is that I do the following

git checkout master
git fetch origin master
git checkout pd_dataframe_chunksize
git rebase master

Is that right? I've screwed up git a lot.

Wait wait...

git checkout master
git fetch upstream master
git checkout pd_dataframe_chunksize
git rebase master

Close. I'm suggesting that you do the following:

git checkout master
git fetch upstream master
git checkout pd_dataframe_chunksize 
git rebase -i master

Then, change pick on the left to squash on the commits that should arguably have been written in the above commit. In your case it will probably look like the following:

pick a0b0a3c ENH: Allow dataframe.from_pandas to accept chunksize or npartitions
squash 0d9913b DOC: Wrote a little more documentation
squash 5c6dcb7 DOC: Whoopsie...

Then you save/exit and rewrite the docstring.

Finally you force push back up to your fork on github. As a warning, this is where you can lose information.

git push -f origin pd_dataframe_chunksize

Alternatively, I can do all of this for you. It's a decent thing to learn how to do though.

Yep I screwed it up.

Let me make a new pull request. I think I should have squashed the ones above this commit as well.