R devel - Dec 1998 - [R & Unix ..] system("test",...) gives funny results...

[in an attempt to provide 

	  system.test <- 
		function(...) { system(paste("test", ...)) == 0 }
and then 
	  file.exists <- 
		function(file){sapply(file, function(f)system.test("-e", f))} 
 
  (which would provide S-plus [>= 4.x] compatibility)
]

Look at this -- ``the horror'' to me :
> (system("test -e /tmp"))
sh: test: argument expected
[1] 256
> system("type test")
test is a shell builtin
> system("echo $SHELL")
/usr/local/bin/tcsh
> (system("test -d /tmp")) # correct: it IS a directory
[1] 0
> (system("test -f /tmp")) # correct: it is NOT a plain file
[1] 256
> (system("/usr/local/bin/test -e /tmp"))  ## this is GNU test from
sh-utils
[1] 0
> (system("/usr/bin/test -e /tmp")) ##- Solaris 2.5.x 
[1] 0
> (system("test -e /tmp"))
sh: test: argument expected
[1] 256

However, there doesn't seem to be a problem 
in the shell [/usr/local/bin/tcsh] :

    sophie{maechler}637> test -e /tmp; echo "$?"
    0
    sophie{maechler}638> test -e /tp; echo "$?"
    1

-----------------

Similar funny behavior  (in tcsh; $version gives the tcsh version)

In the shell

   sophie{maechler}639> echo $version
   tcsh 6.04.00 (Cornell) 93/07/03 (sun4) options 8b,nls,dl,al

In R :

   > (system("echo $version"))

   [1] 0

   > {system("echo $version > /tmp/qq");
scan("/tmp/qq",what="")}
   Read 0 items
   character(0)
   >

----------
One way to get an R internal "test" 
	-- even working in the Win/Mac version of R ?! --- 
would be to incorporate GNU sh-utils "test"
(standalone) and provide something like

	test <- function(...) .Internal(test(...))

However, this probably should have to be done by someone who understands
more than me about this kind of C codes.

----------
Comments, suggestions ?

Martin Maechler <maechler@stat.math.ethz.ch>
http://stat.ethz.ch/~maechler/
Seminar fuer Statistik, ETH-Zentrum SOL G1;	Sonneggstr.33
ETH (Federal Inst. Technology)	8092 Zurich	SWITZERLAND
phone: x-41-1-632-3408		fax: ...-1086			<><
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

Martin Maechler <maechler@stat.math.ethz.ch> writes:
> [in an attempt to provide 
> 
> 	  system.test <- 
> 		function(...) { system(paste("test", ...)) == 0 }
> and then 
> 	  file.exists <- 
> 		function(file){sapply(file, function(f)system.test("-e", f))} 
>  
>   (which would provide S-plus [>= 4.x] compatibility)
> ]
> 
> Look at this -- ``the horror'' to me :
> 
> > (system("test -e /tmp"))
> sh: test: argument expected
> [1] 256
Happens to me on Solaris too, but *not* on Linux!
> Similar funny behavior  (in tcsh; $version gives the tcsh version)
> 
> In the shell
> 
>    sophie{maechler}639> echo $version
>    tcsh 6.04.00 (Cornell) 93/07/03 (sun4) options 8b,nls,dl,al
> 
> In R :
> 
>    > (system("echo $version"))
> 
>    [1] 0
> 
>    > {system("echo $version > /tmp/qq");
scan("/tmp/qq",what="")}
>    Read 0 items
>    character(0)
>    >
Isn't this just because system() uses /bin/sh?
>  {system("echo $USER > /tmp/qq");
scan("/tmp/qq",what="")}
Read 1 items
[1] "pd"

-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard@biostat.ku.dk)             FAX: (+45) 35327907
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

>One way to get an R internal "test"
>        -- even working in the Win/Mac version of R ?! ---
...
>However, this probably should have to be done by someone who understands
>more than me about this kind of C codes.
I've often wondered about this too. In my "compatibility library"
I do a few
similar things like copying files and checking file dates. I haven't
implemented
them other than in Unix and I've often thought some cross-platform issues
would
be resolved by doing these in C. Of course, this may just move the problem. To
what extent are these kinds of things standardized across platforms in C?

Paul Gilbert


-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

> Date: Thu, 10 Dec 1998 14:57:58 +0100
> From: Martin Maechler <maechler@stat.math.ethz.ch>
> [in an attempt to provide 
> 
> 	  system.test <- 
> 		function(...) { system(paste("test", ...)) == 0 }
> and then 
> 	  file.exists <- 
> 		function(file){sapply(file, function(f)system.test("-e", f))} 
>  
>   (which would provide S-plus [>= 4.x] compatibility)
> ]
> 
But surely that is not the way to do this (and how is it going to work on
Windows?)  S-PLUS uses a system call for file.exists, presumably to
access(), even on Unix.

If you want to use test, do remember that what you get is shell-specific.
If you use

system("/usr/bin/test -e /tmp")

you have some chance of working on different shells. (Even Linux, RH5.2,
has test at /usr/bin/test!)


[still interested to hear about the potential of an R builtin 'test',
 relying on GNU sh-utils].

Why not use POSIX system calls instead?
-- 
Brian D. Ripley,                  ripley@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 272860 (secr)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._