Error in eval() for tbl_dt

[nirski]

Boolean variables in tbl_dt are not evaluated when there are no other filters.

This gets filtered:

> data.frame(x = c(1, 1, 2), r = c(T, F, T)) %>% tbl_dt %>% filter(r, x == 1)
Source: local data table [1 x 2]

  x    r
1 1 TRUE

This one throws an error:

> data.frame(x = c(1, 1, 2), r = c(T, F, T)) %>% tbl_dt %>% filter(r)
Error in eval(expr, envir, enclos) : object 'r' not found

Both tbl_df and a base data.frame keep working just fine.

> data.frame(x = c(1, 1, 2), r = c(T, F, T)) %>% tbl_df %>% filter(r)
Source: local data frame [2 x 2]

  x    r
1 1 TRUE
2 2 TRUE

> data.frame(x = c(1, 1, 2), r = c(T, F, T)) %>% filter(r)
  x    r
1 1 TRUE
2 2 TRUE

> sessionInfo()
R version 3.1.1 Patched (2014-09-06 r66537)
Platform: x86_64-w64-mingw32/x64 (64-bit)

locale:
[1] LC_COLLATE=Polish_Poland.1250  LC_CTYPE=Polish_Poland.1250    LC_MONETARY=Polish_Poland.1250
[4] LC_NUMERIC=C                   LC_TIME=Polish_Poland.1250    

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] data.table_1.9.2 dplyr_0.2       

loaded via a namespace (and not attached):
[1] assertthat_0.1 magrittr_1.1.0 parallel_3.1.1 plyr_1.8.1     Rcpp_0.11.2    reshape2_1.4  
[7] stringr_0.6.2  tools_3.1.1   

[romainfrancois]

@hadley do you really want me to pick this up; This is data.table stuff for which I'm kind of clueless.

It looks like it happens in the eval call in filter.data.table

filter.data.table <- function(.data, ..., .env = parent.frame()) {
  expr <- and_expr(dots(...))
  call <- substitute(.data[expr, ])

  eval_env <- new.env(parent = .env)
  eval_env$.data <- .data

  eval(call, eval_env)
}

The generated call looks like : .data[r,] which does not look to work with data.table, as in:

> dt <- data.table( x = c(1, 1, 2), r = c(T, F, T))
> dt[r,]
Erreur dans eval(expr, envir, enclos) : objet 'r' introuvable

[hadley]

Ooops, no. I misread the message.

[hadley]

@arunsrinivasan is this a bug, or deliberate?

dt <- data.table( x = c(1, 1, 2), r = c(T, F, T))
dt[r, ]
# Error in eval(expr, envir, enclos) : object 'r' not found 
dt[r == T,]
#    x    r
# 1: 1 TRUE
# 2: 2 TRUE

Comparing a logical vector to TRUE is usually a redundant operation.

[arunsrinivasan]

Expected. When i is symbol DT[i] looks for i in calling scope (for data.tables). DT[(r)] would be the way to go - until #697 is taken care of.

[hadley]

Fixed as part of 225f01b