grouped_df changes type of vars argument

[kasterma]

I just did:

install.packages("devtools")
devtools::install_github("hadley/lazyeval")
devtools::install_github("hadley/dplyr")

So this concerns the latest version available on github.

Executing:

library(dplyr)
features <- list("feat1", "feat2", "feat3")
class(features[[1]])
grouped_df(data.frame(feat1=1, feat2=2, feat3=3), features)
class(features[[1]])

Shows that the class of features has changed from "character" to name. Here is a copy paste from an Rstudio repl:

> library(dplyr)

Attaching package: ‘dplyr’

The following object is masked from ‘package:stats’:

    filter

The following objects are masked from ‘package:base’:

    intersect, setdiff, setequal, union

> features <- list("feat1", "feat2", "feat3")
> class(features[[1]])
[1] "character"
> grouped_df(data.frame(feat1=1, feat2=2, feat3=3), features)
Source: local data frame [1 x 3]
Groups: feat1, feat2, feat3

  feat1 feat2 feat3
1     1     2     3
> class(features[[1]])
[1] "name"
> 

From the top of the repl:

R version 3.1.1 (2014-07-10) -- "Sock it to Me"
Copyright (C) 2014 The R Foundation for Statistical Computing
Platform: x86_64-apple-darwin13.1.0 (64-bit)

[romainfrancois]

As far as I'm concerned, this is a bug in Rcpp from which dplyr gets the ListOf template class used in group_by underlying code. RcppCore/Rcpp#188

[romainfrancois]

Looking at this some more, I think what we can do is change the code of grouped_df so that it makes a list of symbols before going down to the C++ layer, so that we are immune to the Rcpp bug and/or we don't have to wait for the next Rcpp release.

[hadley]

Or just throw an error - grouped_df is mostly for internal use.

[romainfrancois]

Fair enough. This piles up to what @kevinushey and I have been discussing on RcppCore/Rcpp#188

I'll add defensive code in dplyr but I think Rcpp should throw an error, following what Rcpp11 does.

[romainfrancois]

I've put some defensive code against RcppCore/Rcpp#188, now I get:

> features <- list("feat1", "feat2", "feat3")
> grouped_df(data.frame(feat1=1, feat2=2, feat3=3), features)
Erreur : Elements 1, 2, 3 of sapply(vars, is.name) are not true