Trailing "/" makes file.exists() return FALSE (PR#1098)

I agree to some extend, but I believe a consistent behavior across platforms
is something that should be a goal. Originally I found this problem from a

package (by a group only running WinNT and Solaris) trying to do

   system.file("",package="base")

which should work and give a path according to the help;

Usage:

     system.file(..., package = "base", lib.loc = .lib.loc, pkg, lib)

Arguments:

     ...: character strings, specifying subdirectory and file(s) within
          some package.  The default, none or `""', returns the root of
          the package.  Wildcards are not supported.
...

Doing this on WinMe it gives "". I can't try this on WinNT, but I assume it
works since the group that made the package it using it. I tracked down the
problem to originate in either the .find.package() call or the file.exists()
call done within system.file(). Which one is a matter of taste. Also, I
noticed that system.file(package="base") without the "", gives the wanted
result.

However, I still argue that file.exists() should be robust enough to give
the same answer if the argument is "foo/" or "foo". I don't know about any
filesystem that can have a file and a directory with the same name in the
same directory, so that shouldn't be an issue. Looking at Java, its solution
(java.io) is very robust across platforms. I am always thinking about the
end-user and making their life easier.

Thanks for a great work

Henrik Bengtsson

Dept. of Mathematical Statistics @ Centre for Mathematical Sciences
Lund Institute of Technology/Lund University, Sweden (+2h UTC)
Office: P316, +46 46 222 9611 (phone), +46 46 222 4623 (fax)
hb@maths.lth.se, http://www.maths.lth.se/matstat/staff/hb/

-----Original Message-----
From: Prof Brian Ripley [mailto:ripley@stats.ox.ac.uk]
Sent: Monday, September 24, 2001 12:10 PM
To: hb@maths.lth.se
Cc: R-bugs@biostat.ku.dk
Subject: Re: [Rd] Trailing "/" makes file.exists() return FALSE
(PR#1098)


On Mon, 24 Sep 2001 hb@maths.lth.se wrote:

Full_Name: Henrik Bengtsson
Version: 1.3.1
OS: WinMe
Submission from: (NULL) (130.235.2.229)


I found the following bug on WinMe (I call it a bug, since the

behavior is

different on WinMe than for instance Sun Solaris):

When calling file.exists() with an existing directory name

ending with a "/"

(slash) [R] on WinMe will return FALSE, but when doing the same

without the

ending "/" it will give TRUE.

I made a quick work around, but this should be taken care of in

the native

code:

file.exists <- function(...) {
  # Remove trailing "/", but keep "C:/", "D:/"...
  files <- gsub("\:/$", "\://", c(...));
  files <- gsub("/$", "", files);
  .Internal(file.exists(files));
}

This is not a bug.  A file path ending with "/" is not a file on Windows,
but it is on Solaris.  Next you'll be reporting that file specifications
with drives fail on Solaris: it other words, what is a valid file path is
OS-specific.

You'll find quite a lot of Windows tools reject paths ending in /: for
example the ls in Rtools thinks the directory is non-existent.

Similarly, Solaris accepts // in paths, but Windows does not.

platform i386-pc-mingw32
arch     x86
os       Win32
system   x86, Win32
status
major    1
minor    3.1
year     2001
month    08
day      31
language R

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.
-.-.-.-.-.-.-

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

_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._.
_._

--
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