Skip to content

syskern fails Rcmd check on Windows, gives incorrect information (PR#781)

2 messages · Brian Ripley, Paul Gilbert

#
syskern (`Functions for writing code that is OS and R/S independent', an
amazing description given the reality) fails Rcmd check on Windows, and
also gives incorrect information where it does work.

There are already correct OS-independent functions for these purposes in
R.  It appears not to be needed to run the other three packages in the dse
bundle, but to be the only R package that deliberately does not work on
one of the major platforms.  I doubt if it works on Macs, either.  It
should not be in the main section of CRAN unless it does what it claims.
glbl.s> date.parsed()
Error in eval.with.vis(expr, envir, enclos) : 
        couldn't find function "date.parsed"

It has not been implemented!
systm.> system.call("ls")
Error in system.call("ls") : system calls must be implemented for this operating system.

But they are:
#R.bug.report#       cvs-co               test1
DF.R                 jpegsrc.v6b.tar.gz   test2.emf
GSD2.EMF             libpng-1.0.8.tar.gz  test3.emf
R-1.2.0.tgz          library              test4.emf
R.bug.report         mri.dat              test5.emf
Rconsole             mydata               texinfo-4.0.tar.bz2
Rdist                packages             tools
TB                   rw1020               tools.zip
VR3sc                rw1021               winold.zip
addlibs              rw1030               zlib113.zip
[1] "#R.bug.report#"      "DF.R"                "GSD2.EMF"           
 [4] "R-1.2.0.tgz"         "R.bug.report"        "Rconsole"           
 [7] "Rdist"               "TB"                  "VR3sc"              
[10] "addlibs"             "cvs-co"              "jpegsrc.v6b.tar.gz" 
[13] "libpng-1.0.8.tar.gz" "library"             "mri.dat"            
[16] "mydata"              "packages"            "rw1020"             
[19] "rw1021"              "rw1030"              "test1"              
[22] "test2.emf"           "test3.emf"           "test4.emf"          
[25] "test5.emf"           "texinfo-4.0.tar.bz2" "tools"              
[28] "tools.zip"           "winold.zip"          "zlib113.zip"        

Of course, `ls' is OS-dependent, so should not be used as an example
in a purportedly OS-independent package.
systm.> system.info()
          _              
platform  i386-pc-mingw32
arch      x86            
os        Win32          
system    x86, Win32     
status                   
major     1              
minor     2              
year      2000           
month     12             
day       15             
language  R              
OSversion MS Windows 95  
OStype    MS Windows     

That is incorrect: see what bug.report gives below.  The heading looks
strange too.
whoami> whoami()
Error in whoami() : system calls must be implemented for this operating system.

They are:
[1] "ripley"
user 
"ripley" 


--please do not edit the information below--

Version:
 platform = i386-pc-mingw32
 arch = x86
 os = Win32
 system = x86, Win32
 status = 
 major = 1
 minor = 2.0
 year = 2000
 month = 12
 day = 15
 language = R

Windows NT 4.0 (build 1381) Service Pack 4

Search Path:
 .GlobalEnv, package:syskern, package:ctest, Autoloads, package:base
#
Brian

I have a new version of syskern which takes advantage of the new
features you have added in R 1.2. I just sent details to Fritz. I'm
sorry that I am a bit slow getting it to CRAN, the release of R 1.2 came
at a very busy time for me. In the future, if you give me a bit more
warning when you are going to test packages with a brand new release of
R, I can probably save you some time by telling you what problems to
expect.

I believe most of syskern now uses standard features of R, so almost
everything should work on all platforms on which R 1.2 works. The main
outstanding facility is a mail interface, which you indicated previously
is problematic in Windows. In any case, it would be useful to
standardize R's side of this so that R programs, like mine and also
bug.report(), can use a program mail() which works on some platforms and
may eventually work on others.
I think you are mis-reading this. If you would like to suggest better
phrasing then that may be helpful. The intention was not to provide the
OS-independent functions as you have now done in R. The intention was to
provide an R-side definition of these functions so that I (and others)
could write R programs in a standardized way to use these facilities
when they became available in R. I kludged a Unix version of the
facilities and invited others to do Windows and Mac versions. Five years
ago when R was still young I needed these facilities to use in R
programs, and it has worked very well. It was very little work for me to
take advantage of the new features in R 1.2. I expect my versions will
become depreciated in R and their main use will be to provide S with
similar names for the same facilities, but there are still a few issues
so it is too early to depreciate them yet.

Seasons Greetings,
Paul


-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
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
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._