[datatable-help] (Presumably) unintended issue with data.table1.6.3

[Matthew Dowle]

> As a digression, is there a better way to drop a number of columns?

Yes :

A[,a:=NULL]

but only one column at a time, currently.   That should be 0.000 time, 
regardless of the size of the table.

On the errors, have you installed 1.6.3 from R-Forge before, by any chance? 
If so, I think you may be having 'update' issues, too.
Does test.data.table() work and how many tests does it run?


"Timothée Carayol" <timothee.carayol at gmail.com> wrote in message 
news:CAGam+C4Wed3C1mn2RXqxS8tmqL1r60wkM44U4rHUy6pi0Oqdxg at mail.gmail.com...
Also,

A$a <- NULL
A

at best shows an error:
> A$a <- NULL
> A
Error in data.table(lapply(x, function(x) { :
  'translateCharUTF8' must be called on a CHARSXP

and at worst crashes my R process (segfault).

I guess I'll just do "as.data.table(within(as.data.frame(A)" for now ;-)

t


2011/8/4 Timothée Carayol <timothee.carayol at gmail.com>:
> I am used to removing columns using
> A <- within(A, rm(column))
>
> However this stopped working:
>
>> A <- data.table(a=1:10, b=11:20)
>> within(A, rm(a))
> Error in within.data.table(A, rm(a)) :
> Can only assign one column at a time currently via within
> In addition: Warning message:
> In rm(a) : object 'a' not found
>
> transform(A, a=NULL) doesn't work either.
>
> As a digression, is there a better way to drop a number of columns?
> A[, -col] doesn't work; A[, -"col", with=F] either.
>
> I have the latest version from CRAN I believe.
>
> Thanks
> Timothee
>