R help - Oct 2008 - Rscript -e, Sweave and tempdir()

Hello R-Help

I have a question about the "behind the scenes" behaviour of the
Rscript -e command and particularly its interaction with Sweave and tempdir().

We are trying to deploy R as a web service to do water quality analyses and have
been writing software to call Sweave via Rscript eg:

cmd> Rscript -e Sweave('sweavefile.Rnw')

One problem we have been having is that Sweave tries to write temporary files in
a temporary directory given by the tempdir() command (as per an email from F.
Leitsch appended below).  This is where our problem seems to lie.

As a simplified example, if we call

cmd> Rscript -e tempdir()

from our web service we get a win32(Cannot read memory:0x00000037) error message
unless we grant Read+Write permissions to the root directory, which we are
trying to avoid doing for security reasons.  This suggests Sweave(?) is trying
to write to somewhere else OTHER than tempdir() perhaps.

However if we run:

Rscirpt "testdir.r"  (ie. "sourcing" a file without the -e
option)

where testdir.r contains:

outpath <- tempdir()
write.table(outpath,"D:\\Internet\\RWebService\\Data\\dir.txt")

there appears to be no problem with the call to the tempdir() command.  Also if
we put the call to Sweave within a call2sweave.r file and do

cmd>Rscript call2sweave.r (ie. run sweave via a file that is sourced, rather
than using the -e option)

then things go OK.

In short, something seems to be happening with the -e version of Rscript, the
tempdir() command and its invocation by Sweave.

Any ideas?

Thanks
Paul.


EARLIER EMAIL FROM F. LEISCH (AUTHOR OF SWEAVE)
___________________________________
It's not figures but text output which is collected via temporary files. The
corresponding code in Sweave is:

tmpcon <- file()
sink(file=tmpcon)
err <- NULL
if(options$eval) err <- evalFunc(ce, options)
cat("\n") # make sure final line is complete
sink()
output <- readLines(tmpcon)
close(tmpcon)


So you need to figure out where file() opens the anoymous temporary file on your
operating system. His is the same place as a call to
tempdir() shows, e.g. on my Linux box:

R> tempdir()
[1] "/tmp/RtmpIzi6Yv"

so it tries to write under /tmp.
_____________________________________________
Paul Rustomji
Rivers and Estuaries
CSIRO Land and Water
GPO Box 1666
Canberra ACT 2601

ph +61 2 6246 5810
mobile 0406 375 739

My first thought was that this is R < 2.7.2 and you have encountered
(from the CHANGES file)

    o   Rscript -e (and Rterm -e) failed on Vista because the MSVCRT
        function 'tmpfile' is broken on that platform.

But that couldn't possibly be the case, as the posting guide required you 
to update your R before posting (and tell us lots of information I don't 
see, including your OS).

On Thu, 16 Oct 2008, Paul.Rustomji at csiro.au wrote:
>
> Hello R-Help
>
> I have a question about the "behind the scenes" behaviour of the
Rscript
> -e command and particularly its interaction with Sweave and tempdir().
>
> We are trying to deploy R as a web service to do water quality analyses 
> and have been writing software to call Sweave via Rscript eg:
>
> cmd> Rscript -e Sweave('sweavefile.Rnw')
>
> One problem we have been having is that Sweave tries to write temporary 
> files in a temporary directory given by the tempdir() command (as per an 
> email from F. Leitsch appended below).  This is where our problem seems 
> to lie.
>
> As a simplified example, if we call
>
> cmd> Rscript -e tempdir()
>
> from our web service we get a win32(Cannot read memory:0x00000037) error 
> message unless we grant Read+Write permissions to the root directory, 
> which we are trying to avoid doing for security reasons.  This suggests 
> Sweave(?) is trying to write to somewhere else OTHER than tempdir() 
> perhaps.
>
> However if we run:
>
> Rscirpt "testdir.r"  (ie. "sourcing" a file without the
-e option)
>
> where testdir.r contains:
>
> outpath <- tempdir()
> write.table(outpath,"D:\\Internet\\RWebService\\Data\\dir.txt")
>
> there appears to be no problem with the call to the tempdir() command. 
> Also if we put the call to Sweave within a call2sweave.r file and do
>
> cmd>Rscript call2sweave.r (ie. run sweave via a file that is sourced, 
> rather than using the -e option)
>
> then things go OK.
>
> In short, something seems to be happening with the -e version of 
> Rscript, the tempdir() command and its invocation by Sweave.
>
> Any ideas?
>
> Thanks
> Paul.
>
>
> EARLIER EMAIL FROM F. LEISCH (AUTHOR OF SWEAVE)
> ___________________________________
> It's not figures but text output which is collected via temporary
files. The corresponding code in Sweave is:
>
> tmpcon <- file()
> sink(file=tmpcon)
> err <- NULL
> if(options$eval) err <- evalFunc(ce, options)
> cat("\n") # make sure final line is complete
> sink()
> output <- readLines(tmpcon)
> close(tmpcon)
>
>
> So you need to figure out where file() opens the anoymous temporary file on
your operating system. His is the same place as a call to
> tempdir() shows, e.g. on my Linux box:
>
> R> tempdir()
> [1] "/tmp/RtmpIzi6Yv"
>
> so it tries to write under /tmp.
> _____________________________________________
> Paul Rustomji
> Rivers and Estuaries
> CSIRO Land and Water
> GPO Box 1666
> Canberra ACT 2601
>
> ph +61 2 6246 5810
> mobile 0406 375 739
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>
-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595