Subassignment with a symbol doesn't work

[DavisVaughan]

It does work for data frames though, so I figure we should be consistent.

Found here:
https://community.rstudio.com/t/unquoting-issue-in-purrr-map-df-function/106676/2

library(tibble)
library(rlang)

df <- data.frame(x = 1)
tbl <- tibble(x = 1)

x_sym <- sym("x")
x_chr <- "x"

df[[x_chr]] <- 2
tbl[[x_chr]] <- 2

# Can subset-assign with a symbol for data frames
df[[x_sym]] <- 2

# Can't subset-assign with a symbol for tibbles
tbl[[x_sym]] <- 2
#> Error: Can't use character names to index an unnamed vector.

[DavisVaughan]

Hmm, I guess this is just passing through to vec_as_location2(), so maybe it is a vctrs issue @lionel-?

tibble/R/subsetting.R

Line 579 in 0a50420

subclass_col_index_errors(vec_as_location2(j, n, names), j_arg = j_arg, assign = assign)

[lionel-]

I think so. We're now generally supporting symbol indexing so this should be fixed in vctrs.

[DavisVaughan]

Oh wait, this does already work in vctrs

> vctrs::vec_as_location2(rlang::sym("x"), n = 1L, names = "x")
[1] 1

My analysis was slightly off. This is.vector(j) call returns FALSE for symbols, but TRUE for character vectors.

tibble/R/subsetting.R

Line 171 in e78b695

if (!is.vector(j) || length(j) != 1L || is.na(j) || (is.numeric(j) && j < 0) || is.logical(j)) {

I think this is actually a tibble bug

[krlmlr]

It's even funnier: we do support symbols for one-bracket assignment, data frames don't.

---

author: kirill
date: 2021-07-20
output: "reprex::reprex\_document"
title: irate-booby_reprex.R

library(tibble)
library(rlang)

df <- data.frame(x = 1)
tbl <- tibble(x = 1)

x_sym <- sym("x")
x_chr <- "x"

df[[x_chr]] <- 2
df[x_chr] <- 2
tbl[[x_chr]] <- 2

df[[x_sym]] <- 1
df[x_sym] <- 1
#> Error in `[<-.data.frame`(`*tmp*`, x_sym, value = 1): anyNA() applied to non-(list or vector) of type 'symbol'

tbl[[x_sym]] <- 2
#> Error: Can't use character names to index an unnamed vector.
tbl[x_sym] <- 2

^{Created on 2021-07-20 by the reprex package (v2.0.0)}

[github-actions]

This old thread has been automatically locked. If you think you have found something related to this, please open a new issue and link to this old issue if necessary.