Make a DataFrame not iterable

[pdeffebach]

I recently ran into the following problem doing a bunch of regressions on a DataFrame

regress.(Ref(df), [:y, :y1, :y2, :y3], :treat)

I needed the Ref(df) because otherwise Julia would try to iterate over a DataFrame. Since I don't think there is a way to just iterate through a DataFrame (we restrict iteration to eachcol and eachrow), we should maybe add a method for iterate(df::DataFrame) that makes it work like Ref.

I forget what the exact solution here is to make Julia always treat a DataFrame like a scalar.

[pdeffebach]

We need to add the code

Base.broadcastable(df::AbstractDataFrame) = Ref(df)

somewhere.

[nalimilan]

Makes sense. Though we may want to wait for a bit in case we decided to make data frames iterables over rows or columns (#1514).

[nalimilan]

#1840 takes a different approach, which has quite strong advantages.

[bkamins]

Yes, and with #1840 you still can do Ref(df) if you want to avoid broadcasting over it.

[bkamins]

Closing it as AbstractDataFrame is considered 2-dimensional in broadcasting. Use Ref to protect it against being broadcasted over.

It is not related to being iterable - AbstractDataFrame is not iterable.