About Error ":= and `:=`(...) are defined for use in j, once only and in particular ways. See help(":=")"

Bigchao picture Bigchao · Feb 2, 2014 · Viewed 10.8k times · Source

I encountered a weird error message in data.table

I modified a data.table using := , and it is totally OK without any error. When I trying to put the code into a function, the following error message comes out.

Error in `:=`(date, as.Date(as.character(date), "%Y%m%d") - 1) : 
:= and `:=`(...) are defined for use in j, once only and in particular ways. See     help(":="). Check is.data.table(DT) is TRUE.

Here's reproducible example

testdat <- data.table(ID = c(1:10), date = c(20130101, 20130101, 20130101, 20130101, 20130101, 20130101, 20130101, 20130101, 20130101, 20130101), Number = rnorm(10))
# The single line command works fine. 
testdat[, date := as.Date(as.character(date),"%Y%m%d") - 1][, Number:= NULL]
# But if I wrote them into a function, it failed. 
# ( In this case, it worked as well.. So I got totally lost. ) 
test2 <- data.frame(ID = c(1:10), date = c(20130101, 20130101, 20130101, 20130101, 20130101, 20130101, 20130101, 20130101, 20130101, 20130101), Number = rnorm(10))
readdata <- function(fn){
      DT <- data.table(fn)
      DT[, date:= as.Date(as.character(date),"%Y%m%d") - 1][, Number:= NULL]
      return(DT)
}

To better description, I put parts of my original code here. So you may understand where goes wrong.

readdata <- function(fn){
   DT <- fread(fn, sep=",")
   # DT <- fread("1202.txt")
   setnames(DT, paste0("V",c(1:12)), column_names)
   # Modification on date
   setkey(DT,uid)
   DT[,date := as.Date(as.character(date),"%Y%m%d") - 1][, ignore:= NULL] #ignore is the name of one column
...}

I have a list of txt files, and I want to do the calculation for each of them. First step is using fread, and proceed one by one. Suppose now the I want to do the calculation based on "1202.txt" file. If I start from DT <- fread("1202.txt") and then proceeded. It will not come up this error.

If I want to use readdata("1202.txt") the error message comes out. Most weird is that, I used the readdata before without any errors.

So what's going on here? Any suggestions? Thanks.

> sessionInfo()
R version 3.0.2 (2013-09-25)
Platform: x86_64-w64-mingw32/x64 (64-bit)

locale:
[1] LC_COLLATE=English_United States.1252 
[2] LC_CTYPE=English_United States.1252   
[3] LC_MONETARY=English_United States.1252
[4] LC_NUMERIC=C                          
[5] LC_TIME=English_United States.1252    

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

other attached packages:
[1] data.table_1.8.11

loaded via a namespace (and not attached):
[1] tools_3.0.2

EDIT

After some trials, I found that if I modified code as the following, it worked

   readdata <- function(fn){
   DT <- fread(fn, sep=",")
   DT <- data.table(DT) ## Just add this line compared to the original one.
   # DT <- fread("1202.txt")
   setnames(DT, paste0("V",c(1:12)), column_names)
   # Modification on date
   setkey(DT,uid)
   DT[,date := as.Date(as.character(date),"%Y%m%d") - 1][, ignore:= NULL] #ignore is the name of one column
...}

So the error is due to the fread? After fread, it should be a data.table. Why I need to use data.table(DT) to convert it ?

EDIT

Thanks for attention. Here's an update on Feb 4th, 2014

I first uninstalled my 1.8.11, and followed the instructions of Matt. Install 1.8.10 from CRAN again, and then followed his code step by step. It turns out totally OK without any error.

Then I uninstalled 1.8.11, and then tried to install 1.8.11 again using the precomplied zip file.

As usual, there's a warning message:

> install.packages("~/Desktop/data.table_1.8.11.zip", repos = NULL)
Warning in install.packages :
package ~/Desktop/data.table_1.8.11.zip?is not available (for R version 3.0.2)
Installing package into C:/Users/James/R/win-library/3.0?(as lib?is unspecified)
package data.table?successfully unpacked and MD5 sums checked

> require(data.table)
Loading required package: data.table
data.table 1.8.11  For help type: help("data.table")

It seems that the warning message is wrong, it is totally OK when I loaded the package. And at this time, it is totally OK for the whole process. Thanks for the patience of Matt, and Arun, and all other warmhearted ones. I'm a beginner of data.table. And your kindness is really appreciated.

Here's one more thing, as I have already reported in this link, and still unsolved.

> ?melt.data.table
No documentation for 憁elt.data.table?in specified packages and libraries:
you could try ??melt.data.table?

It's really a pity. Any ideas?

I reported my sessionInfo in that link. And I used Win8.1 64bit

Answer

Bigchao picture Bigchao · Feb 4, 2014

After reinstalling the data.table v1.8.10 / v1.8.11 (I tried for both the two versions), and restarted a new R session. The problem solved.

It turns out my problem was caused by a 5 month old development version being installed.

The data.table homepage was slightly misleading :

Last recommended snapshot precompiled for Windows: v1.8.11 rev931 04 Sep 2013

The [homepage][1] has been improved and now reads :

install.packages("data.table", repos="http://R-Forge.R-project.org")
Or, if that fails, the last precompiled .zip for Windows copied to this homepage may suffice: v1.8.11 rev1110 04 Feb 2014

Thanks for all of you for valuable answers and comments.