Using Rscript, is there a decent way to suppress the non-script output? [duplicate]

Question

Possible Duplicate:
R suppress startupMessages from dependency

I've read about using sink("NUL") / sink("/dev/null"), but neither of them has fixed the problem I'm having. Even if I wrap the library() commands in sink("NUL") and sink(), my call to Rscript outputs all manner of information I don't want to see:

Loading required package: Matrix
Loading required package: methods
Loading required package: lattice
Loaded glmnet 1.8

Loading required package: MASS
Loading required package: lme4

Attaching package: 'lme4'

The following object(s) are masked from 'package:stats':

    AIC, BIC

Loading required package: R2WinBUGS
Loading required package: coda

Attaching package: 'coda'

The following object(s) are masked from 'package:lme4':

    HPDinterval

Loading required package: abind
Loading required package: foreign

arm (Version 1.5-05, built: 2012-6-6)

Working directory is C:/Users/andrews/bootstraps/user/_branches/ER-PoC/Bootstraps/R


Attaching package: 'arm'

The following object(s) are masked from 'package:coda':

    traceplot

[1] "client=51"         "date='01-01-2011'"
[1] "01-01-2011"
[1] 51

The stuff at the end is the only output I actually want, and also the only output I seem able to suppress with sink() commands. It really seems like there should just be an argument to Rscript that suppresses this output (which doesn't even show up if I source my script in the console)... any input?

Answer 1

Andrew, I ran into the same thing and suppressMessages() didn't remove all the extra output, but using sink() in the form of capture.output() wrapped around the suppressMessages() works.

$ rscript --vanilla -e 'library(Rmpfr)'
Loading required package: methods
Loading required package: gmp
---->8----
Loading C code of R package 'Rmpfr': GMP using 32 bits per limb
---->8----


$ rscript --vanilla -e 'suppressMessages( library(Rmpfr) )'
Loading C code of R package 'Rmpfr': GMP using 32 bits per limb


$ rscript --vanilla -e 'msg.out <- capture.output( suppressMessages( library(Rmpfr) ) )'

What is going on when loading the Rmpfr package is several well behaved startup messages written using the message connection along with a not so nice message using the output connection. Sure, you could create and manipulate a sink() on your own, but that is what capture.output() is already setup to do.

Perhaps setting a verbose arg to get a little more control would be helpful::

$ cat sample.R
#!/c/opt/R/R-2.15.0/bin/rscript --vanilla

cmd_args <- commandArgs( TRUE );

if( length( cmd_args ) > 0 ) {
  eval( parse( text = cmd_args[1] ) )
}

if( exists( "verbose" ) ) {
  library( Rmpfr )
} else {
  msg.trap <- capture.output( suppressMessages( library( Rmpfr ) ) )
}

print("Hello")

Which yields::

$ ./sample.R
[1] "Hello"


$ ./sample.R "verbose=TRUE"
Loading required package: methods
Loading required package: gmp

Attaching package: 'gmp'
---->8----
[1] "Hello"

Lots of stuff you could play around with there, but at least you can see how to totally suppress the msg output.

Hope it helps. Have fun!