Ability to merge even if no common columns

[RacingTadpole]

Hi,
I would find it very helpful to be able to merge dataframes without common columns, (which is allowed in R), e.g.

x = DataFrame({'farm' : ['A','B','C']})
y = DataFrame({'crop':['wheat','millet']})

so that x.merge(y, how='outer') returns:

  farm    crop
0    A   wheat
1    A  millet
2    B   wheat
3    B  millet
4    C   wheat
5    C  millet

Currently pandas returns: MergeError: No common columns to perform merge on.

One work-around is to set the indices of x and y to zero, perform a join and the reset the index, as per this StackOverflow post. Another use case is here.

Alternatively, if there is a better solution, please let me know. Thanks!

[jreback]

looks like a dupe of: #5401 ?

[RacingTadpole]

Thanks, I didn't see that one. Yes, although my suggested solution is a little different, ie. generalizing how merge works with how='outer', rather than adding a new how='cross'.

[jreback]

going to close this as a dupe (but its linked); I think its better add a different how='cross' to avoid accidental merges when their are no overlapping columns (and it wasn't explicitly specified). but that's in the future when this is implemented.

of course if you are interested in doing a PR, would be great!

this isn't that difficult as I think the proposed soln is actually fine;