Sean Garborg

I just made `vcat(dfs)` do basic container type promotion in [this commit](https://github.com/garborg/DataFrames.jl/commit/1529509573929c81e5e92a6967438a34d391a482). (Not something that could reuse `vcat(das)` because of the filling-in-missing-columns step.) Anyway, it felt like what we'd ideally...

To fit with how `NA` is treated conceptually throughout the package, when either argument has any `NA`s, setdiff pretty much has to throw an error. (`Set(DataVector[..., NA, ...])` is unknown,...

That's nicer, for sure, and defining the three methods for each function would allow for tuning. On the other hand, it sounds like the quantity of methods that need to...

I think either `true` or `false` would be useful when when `==` already rightly returns `NA`. The only time `true` stands out as inconvenient to me is when the array...

That sounds good to me, but I agree that it's tricky.

That seems in line with Simon's outline, and makes sense to me.

Is there agreement now that `isequal([PooledDataArray], [other type])` should return `false`? @nalimilan seemed to be leaning the other way in a comment on a DataFrames issue, but I feel pretty...

`1:2` and `[1, 2]` come from the same domain and code that accepts both tends to give the same answer for both. `isequal([1.0, 2.0], and [1, 2])` returns false --...

I think we all agree about `isequal`'s role. I am having trouble wrapping my head around what function PDAs with non-string content will serve if PDAs aren't factors, especially because...

Thanks for updating `Distance.jl`! This can be closed if re-exporting happens or the README and docs are updated.