top_n doesn't play well with desc

[HarlanH]

It looks like top_n is a wrapper around filter(min_rank(desc())), which means that top_n(x, 1, desc(col)) causes a SQL syntax error, and of course there's no way to do top_n(x, 1, asc(col)).

Perhaps a bottom_n function that does the same thing, but without the hard-coded desc?

Workaround is trivial, of course, but the symmetric API would be handy!

Thanks!

[klmr]

I think it also doesn’t work on data.frames but I might misunderstand the semantics. I’ve got a table with a column of p-values, and doing x %>% top_n(100, desc(pval)) returns a different set of of rows than x %>% arrange(pval) %>% head(100).

[hadley]

I didn't check this with SQL, but I think using -n to indicate that you want the bottom n values is a reasonable compromise between concision and expression.