You can use install.packages() to install this package, but to use it
in R, you will have to add the XML/libs directory to your path or copy
the files

  libxml2.dll zlib.dll iconv.dll

to a directory in your path.  Putting them into the directory which
contains Rgui.exe and Rterm.exe for this version of R is probably
best.  They will hopefully not interfere with other applications
that rely on alternate versions.


=====================================================================


COMPILING the PACKAGE
---------------------

To compile this package on a Windows machine is quite easy.  Since the
package depends on libxml (version 2), you will have to get the
development version of that package for compiling with MingGW.  To do
this, I have followed the links from www.xmlsloft.org to

   http://www.zlatkovic.com/libxml.en.html 

(The libraries are available directly at http://www.zlatkovic.com/pub/libxml/ )

From the "Getting the Binaries" section, download the zlib, iconv,
libxml2 (and openssl) zip files.

Having fetched these binaries, I unzip these and move the relevant
files to a common area.  Specifically, in my HOME directory on the
cygwin setup on Windows, I create the following directories:

  include/
  lib/
  bin/

Then I copy the files from each of these directories in the .zip files 
to these directories. 


Now, all we need to do is tell the GNU make utility where to find
these header and library files.  You can either edit the
src/Makevars.win file and specify a value for (the now misnamed)
LIB_XML variable to the directory containing the include/, lib/ and
bin/ directory created earlier. In my case, this is $HOME, but
specified as

  D:/cygwin/home/duncan

(rather than /home/duncan).
The remainder of the file should not need to be modified.

Alternatively, you can specify this as an environment variable, e.g.
  export LIBM_XML=D:/cygwin/home/duncan
 

To create a binary, I copy libxml2.dll, iconv.dll and zlib.dll to the
directory XML/inst/libs in the source and then run

  Rcmd build --binary XML