The R Project SVN R

Rev

Rev 28254 | Go to most recent revision | Blame | Compare with Previous | Last modification | View Log | Download | RSS feed 

<html lang="en">
<head>
<title>R Installation and Administration</title>
<meta http-equiv="Content-Type" content="text/html; charset=ISO-8859-1">
<meta name="description" content="R Installation and Administration">
<meta name="generator" content="makeinfo 4.6">
<meta http-equiv="Content-Style-Type" content="text/css">
<style type="text/css"><!--
  pre.display { font-family:inherit }
  pre.format  { font-family:inherit }
  pre.smalldisplay { font-family:inherit; font-size:smaller }
  pre.smallformat  { font-family:inherit; font-size:smaller }
  pre.smallexample { font-size:smaller }
  pre.smalllisp    { font-size:smaller }
--></style>
</head>
<body>
<h1 class="settitle">R Installation and Administration</h1>
<a name="Top"></a>

<h2 class="unnumbered">R Installation and Administration</h2>

<p>This is a guide to installation and administration for R.

   <p>The current version of this document is 1.9.0 (2004-02-04).

   ISBN 3-900051-02-X

<a name="Obtaining%20R"></a>

<h2 class="chapter">1 Obtaining R</h2>

<p>Sources, binaries and documentation for R can be obtained via
<small>CRAN</small>, the "Comprehensive R Archive Network".  See the file
<code>RESOURCES</code> in the R distribution for information on
<small>CRAN</small>.

<a name="Getting%20and%20unpacking%20the%20sources"></a>

<h3 class="section">1.1 Getting and unpacking the sources</h3>

<p>The simplest way is to download the most recent
<code>R-</code><var>x</var><code>.</code><var>y</var><code>.</code><var>z</var><code>.tgz</code> file, and unpack it with

<pre class="smallexample">     tar xvfz R-<var>x</var>.<var>y</var>.<var>z</var>.tgz
     </pre>

<p>on systems that have GNU <code>tar</code> installed.  On other systems you
need at least to have the <code>gzip</code> program installed.  Then you
can use

<pre class="smallexample">     gzip -dc R-<var>x</var>.<var>y</var>.<var>z</var>.tgz | tar xvf -
     </pre>

   <p>If you need to transport the sources on floppy disks, you can download
the <code>R-</code><var>x</var><code>.</code><var>y</var><code>.</code><var>z</var><code>.tgz-split.*</code> files and paste them
together at the destination with (Unix)

<pre class="smallexample">     cat R-<var>x</var>.<var>y</var>.<var>z</var>-split.* &gt; R-<var>x</var>.<var>y</var>.<var>z</var>.tgz
     </pre>

<p>and proceed as above.  If you want the build to be usable by a group of
users, set <code>umask</code> before unpacking so that the files will be
readable by the target group (e.g., <code>umask 022</code> to be usable by all
users).

<a name="Using%20rsync"></a>

<h3 class="section">1.2 Using rsync</h3>

<p>Sources are also available via anonymous rsync.  Use

<pre class="smallexample">     rsync -rC rsync.r-project.org::<var>module</var> R
     </pre>

<p>to create a copy of the source tree specified by <var>module</var> in the
subdirectory <code>R</code> of the current directory, where <var>module</var>
specifies one of the four existing flavors of the R sources, and can be
one of <code>r-release</code> (current released version), <code>r-patched</code>
(patched released version), and <code>r-devel</code> (development version).
(These flavors are described in the R FAQ,
<a href="http://www.ci.tuwien.ac.at/~hornik/R/">http://www.ci.tuwien.ac.at/~hornik/R/</a>.)  The rsync trees are
created directly from the master CVS archive and are updated hourly.
The <code>-C</code> option in the <code>rsync</code> command is to cause it to
skip the CVS directories.  Further information on <code>rsync</code> is
available at <a href="http://rsync.samba.org/rsync/">http://rsync.samba.org/rsync/</a>.

<a name="Installing%20R%20under%20Unix"></a>

<h2 class="chapter">2 Installing R under Unix</h2>

<p>R will configure and build under a number of common Unix platforms
including i386-freebsd, <var>cpu</var>-linux-gnu for the i386, alpha, arm,
hppa, ia64, m68k, powerpc, and sparc CPUs (see e.g.
<a href="http://buildd.debian.org/build.php?&amp;pkg=r-base">http://buildd.debian.org/build.php?&amp;pkg=r-base</a>),
i386-sun-solaris, powerpc-apple-darwin, mips-sgi-irix, alpha-dec-osf4,
rs6000-ibm-aix, hppa-hp-hpux, and sparc-sun-solaris.

   <p>In addition, binary distributions are available for most common Linux
distributions, and for MacOS X (Darwin) with X11.  See the <small>FAQ</small>
for current details.  These are installed in platform-specific ways.  So
for the rest of this chapter we consider only building from the sources.

<a name="Simple%20compilation"></a>

<h3 class="section">2.1 Simple compilation</h3>

<p>First review the essential and useful tools and libraries in
<a href="#Essential%20and%20useful%20other%20programs">Essential and useful other programs</a>, and install those you want or
need.

   <p>Choose a place to install the R tree (R is not just a binary, but
has additional data sets, help files, font metrics etc).  Let us call
this place <var>R_HOME</var>.  Untar the source code.  This should create
directories <code>src</code>, <code>doc</code>, and several more.  Issue the
following commands:

<pre class="smallexample">     ./configure
     make
     </pre>

<p>(See <a href="#Using%20make">Using make</a> if your make is not called <code>make</code>.)

   <p>Then check the built system works correctly, by

<pre class="smallexample">     make check
     </pre>

<p>Failures are not necessarily problems as they might be caused by missing
functionality, but you should look carefully at any reported
discrepancies.  To re-run the tests you would need

<pre class="smallexample">     make check FORCE=FORCE
     </pre>

   <p>More comprehensive testing can be done by

<pre class="smallexample">     make check-devel
     </pre>

<p>or

<pre class="smallexample">     make check-all
     </pre>

<p>see <code>tests/README</code>.

   <p>If these commands execute successfully, the R binary will be copied
to the <code></code><var>R_HOME</var><code>/bin</code> directory.  In addition, a shell-script
front-end called <code>R</code> will be created and copied to the same
directory.  You can copy this script to a place where users can invoke
it, for example to <code>/usr/local/bin/R</code>.  You could also copy the man
page <code>R.1</code> to a place where your <code>man</code> reader finds it,
such as <code>/usr/local/man/man1</code>.  If you want to install the complete
R tree to, e.g., <code>/usr/local/lib/R</code>, see <a href="#Installation">Installation</a>.
Note: you do not <em>need</em> to install R: you can run it from where
it was built.

   <p>You do not necessarily have to build R in the top-level source
directory (say, <code></code><var>TOP_SRCDIR</var><code></code>).  To build in
<code></code><var>BUILDDIR</var><code></code>, run

<pre class="smallexample">     cd <var>BUILDDIR</var>
     <var>TOP_SRCDIR</var>/configure
     make
     </pre>

<p>and so on, as described further below.  This has the advantage of always
keeping your source tree "clean".  (You may need GNU <code>make</code> to
allow this.)

   <p>Make will also build plain text help pages as well as <small>HTML</small> and
LaTeX versions of the R object documentation (the three kinds can
also be generated separately using <code>make help</code>, <code>make html</code>
and <code>make latex</code>).  Note that you need Perl version 5: if this is
not available on your system, you can obtain PDF versions of the
documentation files via <small>CRAN</small>.

   <p>Now <code>rehash</code> if necessary, type <kbd>R</kbd>, and read the R manuals
and the R <small>FAQ</small> (files <code>FAQ</code> or
<code>doc/html/faq.html</code>, or
<a href="http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html">http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html</a> which always has
the latest version).

<a name="Making%20the%20manuals"></a>

<h3 class="section">2.2 Making the manuals</h3>

<p>There is a set of manuals that can be built from the sources,

     <dl>
<dt><code>refman</code>
     <dd>Printed versions of all the help pages.
<br><dt><code>R-FAQ</code>
     <dd>R <small>FAQ</small> (which is already built for you).
<br><dt><code>R-intro</code>
     <dd>"An Introduction to R".
<br><dt><code>R-data</code>
     <dd>"R Data Import/Export".
<br><dt><code>R-admin</code>
     <dd>"R Installation and Administration", this manual.
<br><dt><code>R-exts</code>
     <dd>"Writing R Extensions".
<br><dt><code>R-lang</code>
     <dd>"The R Language Definition".
</dl>

<p>To make these, use

<pre class="smallexample">     make dvi      to create DVI versions
     make pdf      to create PDF versions
     make info     to create info files (not <code>refman</code>).
     </pre>

   <p>You will not be able to build the info files unless you have
<code>makeinfo</code> version 4 or later installed (and some Linux
distributions have 3.12).

   <p>The DVI versions can be previewed and printed using standard programs
such as <code>xdvi</code> and <code>dvips</code>.  The PDF versions can be
viewed using Acrobat Reader or (recent versions of) ghostscript: they
have hyperlinks that can be followed in Acrobat Reader.  The info files
are suitable for reading online with Emacs or the standalone GNU Info.

<a name="Installation"></a>

<h3 class="section">2.3 Installation</h3>

<p>After

<pre class="smallexample">     ./configure
     make
     make check
     </pre>

<p>have been completed successfully, you can install the complete R tree
to your system by typing

<pre class="smallexample">     make install
     </pre>

<p>This will install to the following directories:

     <dl>
<dt><code></code><var>prefix</var><code>/bin</code>
     <dd>the front-end shell script
<br><dt><code></code><var>prefix</var><code>/man/man1</code>
     <dd>the man page
<br><dt><code></code><var>prefix</var><code>/lib/R</code>
     <dd>all the rest (libraries, on-line help system, <small class="dots">...</small>)
</dl>

<p>where <var>prefix</var> is determined during configuration (typically
<code>/usr/local</code>) and can be set by running <code>configure</code> with
the option <code>--prefix</code>, as in

<pre class="smallexample">     ./configure --prefix=/where/you/want/R/to/go
     </pre>

<p>This causes <code>make install</code> to install the R executable to
<code>/where/you/want/R/to/go/bin</code>, and so on.  The prefix of the
installation directories can be seen in the status message that is
displayed at the end of <code>configure</code>.  You can install into
another directory by using

<pre class="smallexample">     make prefix=/path/to/here install
     </pre>

   <p>To install DVI, info and PDF versions of the manuals, use one or more of

<pre class="smallexample">     make install-dvi
     make install-info
     make install-pdf
     </pre>

<p>To ensure that the installed tree is usable by the right group of users,
set <code>umask</code> appropriately (perhaps to <code>022</code>) before unpacking
the sources and throughout the build process.

<a name="Installing%20R%20under%20Windows"></a>

<h2 class="chapter">3 Installing R under Windows</h2>

<p>The <code>bin/windows</code> directory of a <small>CRAN</small> site contains
binaries for a base distribution and a large number of add-on packages
from <small>CRAN</small> to run on Windows 95, 98, NT4, 2000, ME and XP (at
least) on Intel and clones (but not on other platforms).

   <p>You do need one of those Windows versions: Windows 3.11+win32s will not
work.

   <p>Your file system must allow long file names (as is likely except
perhaps for some network-mounted systems).

   <p>The simplest way is to use <code>rw1090dev.exe</code> or
<code>miniR.exe</code>.  Just double-click on the icon and follow the
instructions.  If you installed R this way you can uninstall it from the
Control Panel.

   <p>See the <a href="http://www.stats.ox.ac.uk/pub/R/rw-FAQ.html">R Windows <small>FAQ</small></a> for more details.

<a name="Building%20from%20source"></a>

<h3 class="section">3.1 Building from source</h3>

<p>If you want to build this port from the sources, see the file
<code>src/gnuwin32/INSTALL</code> in the source distribution.  You will need
to collect, install and test an extensive set of tools: see
<a href="http://www.stats.ox.ac.uk/pub/Rtools/">http://www.stats.ox.ac.uk/pub/Rtools/</a> for the current locations.

   <p>You may need to compile under a case-honouring file system: we found
that a <code>samba</code>-mounted file system (which maps all file names to
lower case) did not work.  Open a commands window at a directory <em>whose
path does not contain spaces</em>, and run something like

<pre class="smallexample">     tar zxvf R-1.9.0.tgz
     cd R-1.9.0\src\gnuwin32
     make
     </pre>

<p>sit back and wait (for about 5 minutes on 1GHz PIII with a fast local
disc).

   <p>For further details, including how to make the documentation and how to
cross-compile, see <code>src/gnuwin32/INSTALL</code>.

<a name="Add-on%20packages"></a>

<h2 class="chapter">4 Add-on packages</h2>

<p>This chapter applies to Unix-like and Windows versions of R.

   <p>It is helpful to use the correct terminology.  A <em>package</em> is
loaded from a <em>library</em> by the function <code>library()</code>.  Thus a
library is a directory containing installed packages; the main library
is <code></code><var>R_HOME</var><code>/library</code>, but others can be used, for example by
setting the environment variable <code>R_LIBS</code> or using the R function
<code>.libPaths()</code>.

<a name="Installing%20packages"></a>

<h3 class="section">4.1 Installing packages</h3>

<p>Packages may be distributed in source form or compiled binary form.
Installing source packages requires that compilers and tools
(including Perl 5.004 or later) be installed.  Binary packages
are platform specific and generally need
no special tools to install, but see the documentation for
your platform for details.

   <p>Note that you need to specify implicitly or explicitly the library to
which the package is to be installed.  This is only an issue if you have
more than one library, of course.

   <p>To install packages from source on Unix use

<pre class="smallexample">     R CMD INSTALL -l /path/to/library <var>pkg1</var> <var>pkg2</var> ...
     </pre>

<p>The part <code>-l /path/to/library</code> can be omitted, in which case the
first library in <code>R_LIBS</code> is used if set, otherwise the main
library <code></code><var>R_HOME</var><code>/library</code> is used.  (<code>R_LIBS</code> is looked
for in the environment: <code>.Renviron</code> is not read by <code>R CMD</code>.)

   <p>The Windows equivalent is<a rel="footnote" href="#fn-1"><sup>1</sup></a>

<pre class="smallexample">     R CMD INSTALL -l /path/to/library <var>pkg1</var> <var>pkg2</var> ...
     </pre>

   <p>Alternatively, packages can be downloaded and installed from within
R.  First set the option <code>CRAN</code> to your nearest <small>CRAN</small>
mirror, for example

<pre class="smallexample">     &gt; options(CRAN = "http://cran.us.r-project.org/")
     </pre>

<p>Then download and install package <strong>foo</strong> by

<pre class="smallexample">     &gt; install.packages("foo")
     </pre>

<p>Unless the library is specified (argument <code>lib</code>) the first library
in the library search path is used.

   <p>What this does is different on Unix and Windows.  On Unix it consults
the list of available source packages on <small>CRAN</small>, downloads the
latest version of the <strong>foo</strong> sources, and installs it (via <code>R
CMD INSTALL</code>).  On Windows it looks at the list of <em>binary</em>
versions of packages and downloads the latest version (if any).

   <p>On Windows <code>install.packages</code> can also install a binary package
from a local <code>zip</code> file by setting argument <code>CRAN</code> to
<code>NULL</code>.   <code>RGui.exe</code> has a menu <code>Packages</code> with a GUI
interface to <code>install.packages</code>, <code>update.packages</code> and
<code>library</code>.

<a name="Updating%20packages"></a>

<h3 class="section">4.2 Updating packages</h3>

<p>The command <code>update.packages()</code> is the simplest way to ensure that
all the packages on your system are up to date.  Set the <code>CRAN</code>
option as in the previous section.  The <code>update.packages()</code>
downloads the list of available packages and their current versions,
compares it with those installed and offers to fetch and install any
that have later versions on <small>CRAN</small>.

   <p>An alternative way of keeping packages up-to-date is provided by the
command <code>packageStatus()</code>, which returns an object with information
on all installed packages and packages available at multiple
repositories (CRAN, local archive, <small class="dots">...</small>). The <code>print</code> and
<code>summary</code> methods give an overview of installed and available
packages, the <code>upgrade</code> method offers to fetch and install outdated
packages.  This allows R to fetch packages from several repositories and
keep in sync with all of them, instead of only one CRAN mirror, and is
intended to become the default package manager for future versions of R.

<a name="Removing%20packages"></a>

<h3 class="section">4.3 Removing packages</h3>

<p>Packages can be removed in a number of ways.  From a command prompt they
can be removed by

<pre class="smallexample">     R CMD REMOVE -l /path/to/library <var>pkg1</var> <var>pkg2</var> ...
     </pre>

<p>(Unix) or

<pre class="smallexample">     R CMD REMOVE -l /path/to/library <var>pkg1</var> <var>pkg2</var> ...
     </pre>

<p>(Windows).

   <p>From a running R process they can be removed by

<pre class="smallexample">     &gt; remove.packages(c("pkg1", "pkg2"),
                       lib = file.path("path", "to", "library"))
     </pre>

   <p>Finally, in most installations one can just remove the package directory
from the library.

   <p><strong>Note:</strong> whereas it is currently possible to install package
<em>bundles</em>, one cannot remove these as such--the packages contained
in the bundle have to removed individually.

<a name="Essential%20and%20useful%20other%20programs"></a>

<h2 class="appendix">Appendix A Essential and useful other programs</h2>

<p>This appendix gives details of programs you will need to build R on
Unix-like platforms, or which will be used by R if found by
<code>configure</code>.

   <p>Remember that some package management systems (such as <small>RPM</small> and
deb) make a distinction between the user version of a package and the
development version.  The latter usually has the same name but with the
extension <code>-devel</code> or <code>-dev</code>: you need both versions
installed.

<a name="Essential%20programs"></a>

<h3 class="section">A.1 Essential programs</h3>

<p>You need a means of compiling C and FORTRAN 77 (see <a href="#Using%20FORTRAN">Using FORTRAN</a>).  Some add-on packages also need a C++ compiler.

   <p>Unless you do not want to view graphs on-screen you need <code>X11</code>
installed, including its headers and client libraries.  (On RedHat Linux
this means the <code>XFree86-devel</code> and <code>XFree86-libs</code> RPMs, for
example.)

   <p>You will need Perl version 5.004 or later, available via
<a href="http://www.perl.com/CPAN/">http://www.perl.com/CPAN/</a>, to build any of the on-line
documentation.

   <p>You will not be able to build the info files unless you have
<code>makeinfo</code> version 4 or later installed.

   <p>The typeset documentation needs <code>tex</code> and <code>latex</code>, or
<code>pdftex</code> and <code>pdflatex</code>.

<a name="Useful%20libraries%20and%20programs"></a>

<h3 class="section">A.2 Useful libraries and programs</h3>

<p>The command-line editing depends on the <code>readline</code> library
available from any GNU mirror: you will need a fairly recent version.

   <p>The bitmapped graphics devices <code>jpeg()</code> and <code>png()</code> need the
appropriate headers and libraries installed: <code>jpeg</code> (version 6b
or later) or <code>libpng</code> (version 1.2.3 or later) and <code>zlib</code>
(version 1.1.3 or later) respectively.

   <p>The <code>bitmap</code> and <code>dev2bitmap</code> devices make use of ghostscript
(<a href="http://www.cs.wisc.edu/~ghost">http://www.cs.wisc.edu/~ghost</a>).

   <p>If you have them installed (including the appropriate headers and of
late enough versions), <code>zlib</code>, <code>libbz2</code> and PCRE will be used
if specified by <code>--with-zlib</code>, <code>--with-bzlib</code> or
<code>--with-pcre</code>: otherwise versions in the R sources will be
compiled in.  As the latter suffice and are tested with R you should
not need to change this.

<a name="Tcl%2fTk"></a>

<h3 class="subsection">A.2.1 Tcl/Tk</h4>

<p>The <strong>tcltk</strong> package needs Tcl/Tk installed: the sources are
available at <a href="http://www.scriptics.com/">http://www.scriptics.com/</a>.  To specify the locations
of the Tcl/Tk files you may need the configuration options

     <dl>
<dt><code>--with-tcltk</code>
     <dd>use Tcl/Tk, or specify its library directory
<br><dt><code>--with-tcl-config=</code><var>TCL_CONFIG</var><code></code>
     <dd>specify location of <code>tclConfig.sh</code>
<br><dt><code>--with-tk-config=</code><var>TK_CONFIG</var><code></code>
     <dd>specify location of <code>tkConfig.sh</code>
</dl>

<p>or use the configure variables <code>TCLTK_LIBS</code> and
<code>TCLTK_CPPFLAGS</code> to specify the flags needed for linking against
the Tcl and Tk libraries and for finding the <code>tcl.h</code> and
<code>tk.h</code> headers, respectively.

   <p>Versions of Tcl/TK from 8.3 to 8.4.4 have been used successfully: 8.0 is
no longer supported.

<a name="Linear%20algebra"></a>

<h3 class="subsection">A.2.2 Linear algebra</h4>

<p>The linear algebra routines in R can make use of enhanced BLAS (Basic
Linear Algebra Subprograms, <a href="http://www.netlib.org/blas/faq.html">http://www.netlib.org/blas/faq.html</a>)
routines.  Some are compiler-system-specific (<code>libsunperf</code> on Sun
Sparc<a rel="footnote" href="#fn-2"><sup>2</sup></a>, <code>libessl</code> on IBM, <code>vecLib</code> on MacOS
X) but ATLAS (<a href="http://math-atlas.sourceforge.net/">http://math-atlas.sourceforge.net/</a>) is a "tuned"
BLAS that runs on a wide range of Unix-alike platforms.  If no more
specific library is found, a <code>libblas</code> library in the library path
will be used.  You can specify a specific BLAS library by the
configuration option <code>--with-blas</code> and not to use an external
BLAS library by <code>--without-blas</code>.

   <p>For systems with multiple processors it is possible to use a
multi-threaded version of ATLAS.  (Prior to R 1.8.0 this was not
supported since <code>SIGINT</code> signals sent to the process and handled by
the wrong thread could result in segfaults.)  A remaining issue is that
R profiling, which uses the <code>SIGPROF</code> signal, may cause
problems.  You may want to disable profiling if you use a multi-threaded
version of ATLAS.  You can use a multi-threaded ATLAS by specifying
<pre class="smallexample">     --with-blas="-lptf77blas -lpthread -latlas"
     </pre>

   <p>Another tuned BLAS which is available for some processors under Linux is
by Kazushige Goto, currently available at
<a href="http://www.cs.utexas.edu/users/flame/goto/">http://www.cs.utexas.edu/users/flame/goto/</a>.  Once this is is
installed, it can be used by one of
<pre class="smallexample">     --with-blas=goto
     --with-blas=-lgoto
     </pre>
   Multi-threaded versions of Goto's BLAS are available, so please note the
<em>caveat</em> in the previous paragraph.

   <p>For Intel processors under Linux, Intel's Math Kernel Library
<a href="http://www.intel.com/software/products/mkl/">http://www.intel.com/software/products/mkl/</a>
can be used by
<pre class="smallexample">     --with-blas="-lmkl -lguide -lpthread"
     </pre>
   with the same caveat on multi-threading.  (Thanks to Andy Liaw for the
information.)

   <p>Note that the BLAS library will be used for several add-on packages as
well as for R itself.  This means that it is better to use a shared
BLAS library, as most of a static library will be compiled into the R
executable and each BLAS-using package.  In any case, the BLAS library
must be usable with dynamically-loadable code: this can be a problem
with ATLAS that is not by default built with position-independent code.

   <p>You will need double-precision and double-complex versions of the BLAS,
but not single-precision nor complex routines.

   <p>Provision is made for using an external LAPACK library, principally to
cope with BLAS libraries which contain a copy of LAPACK (such as
<code>libsunperf</code> on Solaris and <code>vecLib</code> on MacOS 10.2.2).
However, the likely performance gains are thought to be small (and may
be negative), and the default is not to search for a suitable LAPACK
library.  You can specify a specific LAPACK library or a search for a
generic library by the configuration option <code>--with-lapack</code>.  The
default for <code>--with-lapack</code> is to check the BLAS library and then
look for an external library <code>-llapack</code>.  Sites searching for the
fastest possible linear algebra may want to build a LAPACK library using
the ATLAS-optimized subset of LAPACK.  To do so specify something like
<pre class="smallexample">     --with-lapack="-L/path/to/libs -llapack -lcblas"
     </pre>

<p>since the ATLAS subset of LAPACK depends on <code>libcblas</code>.

   <p>If you do use <code>--with-lapack</code>, be aware of potential problems
with bugs in the LAPACK 3.0 sources (or in the posted corrections to those
sources).  In particular, bugs in <code>DGEEV</code> and <code>DGESDD</code> have
resulted in error messages such as
<pre class="smallexample">     DGEBRD gave error code -10
     </pre>

<p>(from the Debian <code>-llapack</code> which was current in late 2002).  Other
potential problems are incomplete versions of the libraries: for example
<code>libsunperf</code> from Sun Forte 6.x was missing the entry point for
<code>DLANGE</code> and <code>vecLib</code> has omitted the BLAS routine
<code>LSAME</code>.

   <p>As with all libraries, you need to ensure that they and R were
compiled with compatible compilers and flags.  For example, this means
that on Sun Sparc using the native compilers the flag <code>-dalign</code>
is needed so <code>libsunperf</code> can be used.

   <p>An ATLAS `tuned' BLAS can also be used on Windows: see
<code>src/gnuwin32/INSTALL</code> for how to enable this when building from
source, and <a href="http://www.stats.ox.ac.uk/pub/R/rw-FAQ.html">R Windows <small>FAQ</small></a> for adding pre-compiled support to binary versions.
Goto's BLAS can also be used when building from source.

   <p>Note that under Unix (but not under Windows) if R is compiled
against a non-default BLAS, then all BLAS-using packages must also be.
So if R is re-built after ATLAS is installed, then packages such as
<strong>quantreg</strong> will need to be re-installed.

<a name="Configuration%20on%20Unix"></a>

<h2 class="appendix">Appendix B Configuration on Unix</h2>

<a name="Configuration%20options"></a>

<h3 class="section">B.1 Configuration options</h3>

<p><code>configure</code> has many options: running

<pre class="smallexample">     ./configure --help
     </pre>

<p>will give a list.  Probably the most important ones not covered
elsewhere are (defaults in brackets)

     <dl>
<dt><code>--with-x</code>
     <dd>use the X Window System
<br><dt><code>--x-includes=</code><var>DIR</var><code></code>
     <dd>X include files are in <var>DIR</var>
<br><dt><code>--x-libraries=</code><var>DIR</var><code></code>
     <dd>X library files are in <var>DIR</var>
<br><dt><code>--with-readline</code>
     <dd>use readline library (if available) [yes]
<br><dt><code>--enable-R-profiling</code>
     <dd>attempt to compile support for <code>Rprof()</code> [yes]
<br><dt><code>--enable-R-shlib</code>
     <dd>build R as a shared library [no]
</dl>

<p>You can use <code>--without-foo</code> or <code>--disable-foo</code> for the
negatives.

   <p>You will want to use <code>--disable-R-profiling</code> if you are building
a profiled executable of R (e.g. with <code>-pg)</code>.

   <p>Flag <code>--enable-R-shlib</code> causes the make process to build R as a
shared library, typically called <code>libR.so</code>, and to take
considerably longer, so you probably only want this if you will be using
an application which embeds R.

<a name="Configuration%20variables"></a>

<h3 class="section">B.2 Configuration variables</h3>

<p>If you need or want to set certain configure variables to something
other than their default, you can do that by either editing the file
<code>config.site</code> (which documents all the variables you might want to
set) or on the command line as

<pre class="smallexample">     ./configure VAR=<var>value</var>
     </pre>

<p>These variables are <em>precious</em>, implying that they do not have to
be exported to the environment, are kept in the cache even if not
specified on the command line and checked for consistency between two
configure runs (provided that caching is used), and are kept during
automatic reconfiguration as if having been passed as command line
arguments, even if no cache is used.

   <p>See the variable output section of <code>configure --help</code> for a list of
all these variables.

   <p>One common variable to change is <code>R_PAPERSIZE</code>, which defaults to
<code>a4</code>, not <code>letter</code>.  (Valid values are <code>a4</code>,
<code>letter</code>, <code>legal</code> and <code>executive</code>.)

   <p>If you have libraries and header files, e.g., for GNU readline, in
non-system directories, use the variables <code>LDFLAGS</code> (for libraries,
using <code>-L</code> flags to be passed to the linker) and <code>CPPFLAGS</code>
(for header files, using <code>-I</code> flags to be passed to the C/C++
preprocessors), respectively, to specify these locations.  These default
to <code>/usr/local/lib</code> and <code>/usr/local/include</code> to catch the most
common cases.  If libraries are still not found, then maybe your
compiler/linker does not support re-ordering of <code>-L</code> and
<code>-l</code> flags (this has been reported to be a problem on HP-UX with
the native <code>cc</code>).  In this case, use a different compiler (or a
front end shell script which does the re-ordering).

   <p>Another precious variable is <code>R_BROWSER</code>, the default browser, which
should take a value of an executable in the user's path or specify
a full path.

   <p>If you find you need to alter configure variables, it is worth noting
that some settings may be cached in the file <code>config.cache</code>, and it
is a good idea to remove that file (if it exists) before re-configuring.
Note that caching is turned <em>off</em> by default; use the command line
option <code>--config-cache</code> (or <code>-C</code>) to enable caching.

<a name="Using%20make"></a>

<h3 class="section">B.3 Using make</h3>

<p>To compile R, you will most likely find it easiest to use GNU
<code>make</code>.  On Solaris 2.6/7/8 in particular, you need a version of
GNU <code>make</code> different from 3.77; 3.79.1 works fine, as does the Sun
<code>make</code>.  The native <code>make</code> is reported to fail on SGI
Irix 6.5 and Alpha/OSF1 (aka Tru64).

   <p>To build in a separate directory you need a <code>make</code> that uses the
<code>VPATH</code> variable, for example GNU <code>make</code>, or Sun
<code>make</code> on Solaris 2.7/8 (but not earlier).

   <p>If you want to use a <code>make</code> by another name, for example if your
GNU <code>make</code> is called <code>gmake</code>, you need to set the variable
<code>MAKE</code> at configure time, for example

<pre class="smallexample">     ./configure MAKE=gmake
     </pre>

<a name="Using%20FORTRAN"></a>

<h3 class="section">B.4 Using FORTRAN</h3>

<p>To compile R, you need a FORTRAN compiler or <code>f2c</code>, the
FORTRAN-to-C converter (<a href="http://www.netlib.org/f2c">http://www.netlib.org/f2c</a>).  The default
is to search for <code>g77</code>, <code>f77</code>, <code>xlf</code>,
<code>frt</code>, <code>pgf77</code>, <code>fl32</code>, <code>af77</code>,
<code>fort77</code>, <code>f90</code>, <code>xlf90</code>, <code>pgf90</code>,
<code>epcf90</code>, <code>f95</code>, <code>fort</code>, <code>xlf95</code>,
<code>lf95</code>, <code>g95</code>, and <code>fc</code> (in that
order)<a rel="footnote" href="#fn-3"><sup>3</sup></a>, and
then for <code>f2c</code>, and use whichever is found first; if none is
found, R cannot be compiled.  The search mechanism can be changed
using the configure variables <code>F77</code> and <code>F2C</code> which specify
the commands that run the FORTRAN 77 compiler and FORTRAN-to-C
converter, respectively.  If <code>F77</code> is given, it is used to compile
FORTRAN; otherwise, if <code>F2C</code> is given, f2c is used even if a
FORTRAN compiler would be be available.  If your FORTRAN compiler is in
a non-standard location, you should set the environment variable
<code>PATH</code> accordingly before running <code>configure</code>, or use the
configure variable <code>F77</code> to specify its full path.

   <p>If your FORTRAN libraries are in slightly peculiar places, you should
also look at <code>LD_LIBRARY_PATH</code> or your system's equivalent to make
sure that all libraries are on this path.

   <p>You must set whatever compilation flags (if any) are needed to ensure
that FORTRAN <code>integer</code> is equivalent to a C <code>int</code> pointer and
FORTRAN <code>double precision</code> is equivalent to a C <code>double</code>
pointer.  This is checked during the configuration process.

   <p>Some of the FORTRAN code makes use of <code>COMPLEX*16</code> variables, which
is a FORTRAN 90 extension.  This is checked for at configure
time<a rel="footnote" href="#fn-4"><sup>4</sup></a>, but you may need to avoid
compiler flags<a rel="footnote" href="#fn-5"><sup>5</sup></a> asserting
FORTRAN 77 compliance.

   <p>For performance reasons<a rel="footnote" href="#fn-6"><sup>6</sup></a> you may want to choose a FORTRAN 90/95 compiler.

   <p>If you use <code>f2c</code> you may need to ensure that the FORTRAN type
<code>integer</code> is translated to the C type <code>int</code>.  Normally
<code>f2c.h</code> contains <code>typedef long int integer;</code>, which will work
on a 32-bit platform but not on a 64-bit platform.

<a name="Compile%20and%20load%20flags"></a>

<h3 class="section">B.5 Compile and load flags</h3>

<p>A wide range of flags can be set in the file <code>config.site</code> or as
configure variables on the command line.  We have already mentioned

     <dl>
<dt><code>CPPFLAGS</code>
     <dd>header file search directory (<code>-I</code>) and any other miscellaneous
options for the C and C++ preprocessors and compilers
<br><dt><code>LDFLAGS</code>
     <dd>path (<code>-L</code>), stripping (<code>-s</code>) and any other miscellaneous
options for the linker
</dl>

<p>and others include

     <dl>
<dt><code>CFLAGS</code>
     <dd>debugging and optimization flags, C
<br><dt><code>MAIN_CFLAGS</code>
     <dd>ditto, for compiling the main program
<br><dt><code>SHLIB_CFLAGS</code>
     <dd>for shared libraries
<br><dt><code>FFLAGS</code>
     <dd>debugging and optimization flags, FORTRAN
<br><dt><code>MAIN_FFLAGS</code>
     <dd>ditto, for compiling the main program
<br><dt><code>SHLIB_FFLAGS</code>
     <dd>for shared libraries
<br><dt><code>MAIN_LDFLAGS</code>
     <dd>additional flags for the main link
<br><dt><code>SHLIB_LDFLAGS</code>
     <dd>additional flags for linking the shared libraries
</dl>

<p>Library paths specified as <code>-L/lib/path</code> in <code>LDFLAGS</code> are
collected together and prepended to <code>LD_LIBRARY_PATH</code> (or your
system's equivalent), so there should be no need for <code>-R</code> or
<code>-rpath</code> flags.

   <p>To compile a profiling version of R, one might for example want to
use <code>MAIN_CFLAGS=-pg</code>, <code>MAIN_FFLAGS=-pg</code>,
<code>MAIN_LDFLAGS=-pg</code> on platforms where <code>-pg</code> cannot be used
with position-independent code.

   <p><strong>Beware:</strong> it may be necessary to set <code>CFLAGS</code> and
<code>FFLAGS</code> in ways compatible with the libraries to be used: one
possible issue is the alignment of doubles, another is the way
structures are passed.

<a name="Building%20the%20GNOME%20interface"></a>

<h3 class="section">B.6 Building the <small>GNOME</small> interface</h3>

<p>This interface is experimental and incomplete.  It provides a console
and two graphics devices named <code>gtk()</code> and <code>gnome()</code>.  The
console offers a basic command line editing and history mechanism, along
with tool and button bars that give a point-and-click interface to some
R commands.  Many of the features of the console are currently
stubs. The <code>gtk()</code> graphics device is a port of the <code>x11()</code>
device to <small>GDK</small> (the GIMP Drawing Kit). The <code>gnome()</code>
device uses the <small>GNOME</small> canvas.

   <p>Due to its experimental nature, the <small>GNOME</small> interface for R
will not be built automatically.  You must specify it by running
configure with the <code>--with-gnome</code> option.  For example, you might
run

<pre class="smallexample">     ./configure --with-gnome
     </pre>

<p>but please check you have all the requirements first. You need at least
the following libraries (or later) installed

<pre class="smallexample">     audiofile-0.2.1
     esound-0.2.23
     glib-1.2.10
     gtk+-1.2.10
     imlib-1.9.10
     ORBit-0.5.12
     gnome-libs-1.4.1.2
     libxml-1.8.16
     libglade-0.17
     </pre>

<p>It is preferable to have a complete installation of the <small>GNOME</small>
desktop environment.  If you use Linux, then this should be provided
with your distribution.  In addition, packaged binary distributions of
GNOME are available from <a href="http://www.ximian.com">http://www.ximian.com</a> for the most
popular Linux distributions and for Solaris.

   <p>Remember that some package management systems (such as <small>RPM</small> and
deb) make a distinction between the user version of a package and the
developer version.  The latter usually has the same name but with the
extension <code>-devel</code>.  If you use a pre-packaged version of
<small>GNOME</small> then you must have the developer versions of the above
packages in order to compile the R-GNOME interface.

   <p>The full list of <small>GNOME</small> options to configure is

     <dl>
<dt><code>--with-gnome</code>
     <dd>use <small>GNOME</small>, or specify its prefix [no]
<br><dt><code>--with-gnome-includes=</code><var>DIR</var><code></code>
     <dd>specify location of <small>GNOME</small> headers
<br><dt><code>--with-gnome-libs=</code><var>DIR</var><code></code>
     <dd>specify location of <small>GNOME</small> libs
<br><dt><code>--with-libglade-config=</code><var>LIBGLADE_CONFIG</var><code></code>
     <dd>specify location of <code>libglade-config</code>
</dl>

<a name="Platform%20notes"></a>

<h3 class="section">B.7 Platform notes</h3>

<p>This section provides some notes on building R on different Unix-like
platforms.  These notes are based on tests run on one or two systems in
each case with particular sets of compilers and support libraries.
Success in building R depends on the proper installation and functioning
of support software; your results may differ if you have other versions
of compilers and support libraries.

<a name="Linux"></a>

<h3 class="subsection">B.7.1 Linux</h4>

<p>Linux is the main development platform for R, so compilation from the
sources is normally straightforward.

   <p>Remember that some package management systems (such as <small>RPM</small> and
deb) make a distinction between the user version of a package and the
developer version.  The latter usually has the same name but with the
extension <code>-devel</code> or  <code>-dev</code>: you need both versions
installed.  So please check the <code>configure</code> output to see if the
expected features are detected: if for example <code>readline</code> is
missing add the package containing its headers.

   <p>When R has been installed from a binary distribution there are
sometimes problems with missing components such as the Fortran
compiler.  Searching the <code>R-help</code> archives will normally reveal
what is needed.

   <p>It seems that the <code>gcc</code> compilers normally produce PIC code on
<code>ix86</code> Linux but do not necessarily do so on 64-bit versions such
as that for AMD Opteron.  So care can be needed with BLAS libraries and
when building R as a shared library to ensure that
position-independent code is used in any static libraries (such as the
Tcl/Tk libraries, <code>libpng</code>, <code>libjpeg</code> and <code>zlib</code>) which
might be linked against.  Fortunately these are normally built as shared
libraries with the exception of the ATLAS BLAS libraries.

<a name="MacOS%20X"></a>

<h3 class="subsection">B.7.2 MacOS X</h4>

<p>You can build R as a Unix application on MacOS X.  You will need the
DevTools, <code>f2c</code> or <code>g77</code>, and the <code>dlcompat</code>
library.  You will also need to install an X sub-system or configure
with <code>--without-x</code>.

   <p><code>f2c</code>, <code>g77</code>, the <code>dlcompat</code> library, and X server
and support libraries are available from the Fink project
(<a href="http://fink.sourceforge.net">http://fink.sourceforge.net</a>).  At the time of writing
<code>f2c</code> and <code>g77</code> were not available as part of the Fink
binary distribution and needed to be installed directly; for example for
<code>g77</code> use

<pre class="smallexample">     fink install g77
     </pre>

   <p>The <code>vecLib</code> library of MacOS &gt;= 10.2.2 can be used <em>via</em> the
configuration options
<pre class="smallexample">     --with-blas="-framework vecLib" --with-lapack
     </pre>

<p>to provide higher-performance versions of the BLAS and LAPACK routines.
With <code>gcc 3.1</code> that appears to be the only way to build R, as
the Fortran support routines in <code>libg2c</code> cannot be linked into a
dynmaic library.  (We have had reports of success with pre-release
versions of <code>gcc 3.3</code>.)

<a name="Solaris"></a>

<h3 class="subsection">B.7.3 Solaris on Sparc</h4>

<p>R has been built successfully on Solaris 8 aka Solaris 2.8 aka SunOS
5.8 using <code>gcc</code>/<code>g77</code> and the SunPro WorkShop&nbsp;6 (aka
Forte 6) compilers and the `Sun ONE Studio 7 Compiler Suite' (aka Forte
7), and less regularly on Solaris 2.5.1, 2.6, 2.7 and 9.  GNU
<code>make</code> is needed prior to Solaris 2.7 for building other than in
the source tree, and perhaps even then.

   <p>The Solaris versions of several of the tools needed to build R
(e.g. <code>make</code>, <code>ar</code> and <code>ld</code>) are in
<code>/usr/ccs/bin</code>, so if using those tools ensure this is in your
path.

   <p><code>gcc</code> 3.2.1 and 3.2.2 generate incorrect code on 32-bit Solaris
builds with optimization, but versions 3.1, 3.2, 3.2.3  and 3.3.x work
correctly.  At least files <code>src/main/engine.c</code>,
<code>src/main/graphics.c</code> and <code>src/modules/devX11.c</code> are affected.

   <p>If using <code>gcc</code>, do ensure that the compiler was compiled for the
version of Solaris in use.  (This can be ascertained from <code>gcc
-v</code>.)  <code>gcc</code> makes modified versions of some header files, and
so (for example) <code>gcc</code> compiled under Solaris 2.6 will not
compile R under Solaris 2.7.  Also, do ensure that it was compiled
for the assembler/loader in use.  If you download <code>gcc</code> from
<a href="http://www.sunfreeware.com">http://www.sunfreeware.com</a> then you need to download
<code>binutils</code> too.  To avoid all these pitfalls we strongly
recommended you compile <code>gcc</code> from the sources yourself.

   <p>When using the SunPro compilers do <em>not</em> specify <code>-fast</code>, as
this disables <small>IEEE</small> arithmetic and <code>make check</code> will
fail.  The maximal set of optimization options known to work is

<pre class="smallexample">     -xlibmil -xO5 -dalign
     </pre>

   <p>We have found little performance difference between <code>gcc</code> and
<code>cc</code> but considerable benefit from using a SunPro Fortran
compiler: the <code>gcc</code>/<code>f77</code> combination works well.  For
many C++ applications (e.g. package <strong>Matrix</strong>) Forte 7 requires
<code>-lCstd</code>, which the configure script will add to
<code>SHLIB_CXXLDFLAGS</code> if it identifies the compiler correctly.

   <p>To compile for a 64-bit target on Solaris (which needs an UltraSparc
chip and for support to be enabled in the OS) with the Forte 6 and 7
compilers we used

<pre class="smallexample">     CC="cc -xarch=v9"
     CFLAGS="-xO5 -xlibmil -dalign"
     F77="f95 -xarch=v9"
     FFLAGS="-xO5 -xlibmil -dalign"
     CXX="CC -xarch=v9"
     CXXFLAGS="-xO5 -xlibmil -dalign"
     </pre>

<p>in <code>config.site</code>.

   <p>For 64-bit compilation with <code>gcc</code> 3.2.x and 3.3.x we used

<pre class="smallexample">     CC="gcc -m64"
     FFLAGS="-m64 -g -O2"
     CXXFLAGS="-m64 -g -O2"
     LDFLAGS="-L/usr/local/lib/sparcv9 -L/usr/local/lib"
     </pre>

   <p>Note that <code>/usr/local/lib/sparcv9</code> will need to be in the
<code>LD_LIBRARY_PATH</code> during configuration.

   <p>Note that using <code>f95</code> allows the Sun performance library
<code>libsunperf</code> to be selected: it will not work with <code>f77</code>,
nor with <code>g77</code>.  <code>libsunperf</code> contains both BLAS and LAPACK
code, and <code>--with-lapack</code> is recommended for 32-bit builds using
<code>f95</code>, but not for 64-bit builds where on our test system it
failed in both Forte 6U1 and 7, albeit in different ways.  Our
experience has been that ATLAS's BLAS is faster than <code>libsunperf</code>,
especially for complex numbers.

   <p>Some care is needed to ensure that libraries found by
<code>configure</code> are compatible with the R executable and modules, as
the testing process will not detect many of the possible problems.  For
32-bit builds under <code>cc</code> the flag <code>-dalign</code> is needed for
some of the Sun libraries: fortunately the equivalent flag for
<code>gcc</code>, <code>-mno-unaligned-doubles</code>, is the default.  In
theory, libraries such as <code>libpng</code>, <code>libjpeg</code>, <code>zlib</code> and
the ATLAS libraries need to be built with a <code>pic</code> or <code>PIC</code>
flag, which could be a problem if static libraries are used.  In
practice this seems to give little problem for 32-bit builds.

   <p>For a 64-bit build, 64-bit libraries must be used.  As the configuration
process by default sets <code>LDFLAGS</code> to <code>-L/usr/local/lib</code>,
you may need to set it to avoid finding 32-bit addons (as in the
<code>gcc -m64</code> example above).  It is possible to build Tcl/Tk as
64-bit libraries with the configure option <code>--enable-64bit</code>, but
only with the Forte compiler (and not with <code>gcc</code>) as of Tcl/Tk 8.4.5.

<a name="HP-UX"></a>

<h3 class="subsection">B.7.4 HP-UX</h4>

<p>R has been built successfully on HP-UX 10.2 and HP-UX 11.0 using both
native compilers and <code>gcc</code>.  However, 10.2 has not been tested
since R 1.4.0.  By default, R is configured to use <code>gcc</code> and
<code>g77</code> on HP-UX (if available).  Some installations of
<code>g77</code> only install a static version of the <code>g2c</code> library
that cannot be linked into a shared library since its files have not
been compiled with the appropriate flag for producing position
independent code (<small>PIC</small>).  This will result in <code>make</code>
failing with a linker error similar to

<pre class="smallexample">     ld: CODE_ONE_SYM fixup to non-code subspace in file foo.o -
     shared library must be position independent. Use +z or +Z to recompile.
     </pre>

<p>(<code>+z</code> and <code>+Z</code> are the <small>PIC</small> flags for the native
compiler <code>cc</code>.)  If this is the case you either need to modify
your <code>g77</code> installation or configure with

<pre class="smallexample">     F77=fort77
     </pre>

<p>to specify use of the native <small>POSIX</small>-compliant FORTRAN 77
compiler.

   <p>You may find that <code>configure</code> detects other libraries that R
needs to use as shared libraries but are only available as static
libraries.  If you cannot install shared versions you will need to tell
<code>configure</code> not to use these libraries, or make sure they are
not in the library path.  The symptom will the linker error shown in the
last paragraph.  Static libraries that might be found and would cause
problems are

<pre class="smallexample">     BLAS                   use --without-blas
     Tcl/Tk                 use --without-tcltk
     GNOME                  not built by default
     libpng                 use --without-libpng
     jpeg                   use --without-jpeglib
     zlib                   use --without-zlib
     </pre>

<p>and <code>bzip2</code> and <code>pcre</code> are problematic when building
<code>libR.so</code>, only.  These can be avoided by <code>--without-bzlib</code>
and <code>--without-pcre</code> respectively, but these are the defaults.

   <p>Some versions of <code>gcc</code> may contain what appears to be a bug at
the <code>-O2</code> optimization level that causes

<pre class="smallexample">     &gt; 2 %/% 2
     [1] 1
     &gt; 1:2 %/% 2
     [1] 0 0     # wrong!!
     </pre>

<p>which will cause <code>make check</code> to fail.  If this is the case, you
should use <code>CFLAGS</code> to specify <code>-O</code> as the optimization
level to use.

   <p>Some systems running HP-UX 11.0 may have a <code>gcc</code> that was
installed under HP-UX 10.2.  Between versions 10.2 and 11.0 HP-UX
changed its support functions for <small>IEEE</small> arithmetic from the
recommended functions of the <small>IEEE</small> standard to the ones
specified in the C9x draft standard.  In particular, this means that
<code>finite</code> has been replaced by <code>isfinite</code>.  A <code>gcc</code>
configured for HP-UX 10.2 run on 11.0 will not find <code>isfinite</code>, and
as a result <code>configure</code> does not recognize the machine as fully
supporting <small>IEEE</small> arithmetic and does not define <code>IEEE_754</code>
when compiling C code.  This results in a failure in <code>make
check</code>.  The best solution is to install a properly configured
<code>gcc</code>.  An alternative work-around is to add <code>-DIEEE_754</code>
to the <code>CFLAGS</code> variable.

   <p>You can configure R to use both the native <code>cc</code> and
<code>fort77</code> with

<pre class="smallexample">     ./configure CC=cc F77=fort77
     </pre>

   <p><code>f90</code> insists on linking against a static <code>libF90.a</code> which
typically resides in a non-standard directory (e.g.,
<code>/opt/fortran90/lib</code>).  Hence, to use <code>f90</code> one needs to
add this directory to the linker path via the configure variable
<code>LDFLAGS</code> (e.g., <code>./configure F77=f90
LDFLAGS=/opt/fortran90/lib</code>).

<a name="IRIX"></a>

<h3 class="subsection">B.7.5 IRIX</h4>

<p>R has been built successfully on IRIX64 6.5 using
<code>gcc</code>/<code>f77</code> or <code>cc</code>/<code>f77</code> for 32-bit
executables and the native compilers for a 64-bit executable.  The
command

<pre class="smallexample">     ./configure CC="cc -64" F77="f77 -64" --with-tcltk=no
     </pre>

<p>was used to create the 64-bit executable.  It was necessary to
explicitly omit Tcl/Tk because <code>configure</code> would find the 32-bit
version but not detect that is was incompatible with a 64-bit build.

   <p>A 32-bit build using <code>gcc</code>/<code>g77</code> passed <code>make
check</code> but failed <code>make test-all-extras</code> in the complex LAPACK
tests.

<a name="Alpha%2fOSF1"></a>

<h3 class="subsection">B.7.6 Alpha/OSF1</h4>

<p>R has been built successfully on an Alpha running OSF1 V4.0 using
<code>gcc</code>/<code>g77</code> and <code>cc</code>/<code>f77</code>.  Mixing
<code>cc</code> and <code>g77</code> fails to configure.  The
<code>configure</code> option <code>--without-blas</code> was used since the
native blas seems not to have been built with the flags needed to
suppress <code>SIGFPE</code>'s.  Currently R does not set a signal handler for
<code>SIGFPE</code> on platforms that support <small>IEEE</small> arithmetic, so
these are fatal.

<a name="Alpha%2fFreeBSD"></a>

<h3 class="subsection">B.7.7 Alpha/FreeBSD</h4>

<p>Attempts to build R on an Alpha with FreeBSD 4.3 have been only
partly successful.  Configuring with <code>-mieee</code> added to both
<code>CFLAGS</code> and <code>FFLAGS</code> builds successfully, but tests fail with
<code>SIGFPE</code>'s.  It would appear that <code>-mieee</code> only defers these
rather than suppressing them entirely.  Advice on how to complete this
port would be greatly appreciated.

<a name="AIX"></a>

<h3 class="subsection">B.7.8 AIX</h4>

<p>On AIX 4.3.3 and AIX 5.1, it was found that the use of "run time
linking" (as opposed to normal AIX style linking) was required.  For
this, the R main program must be linked to the runtime linker with the
<code>-brtl</code> linker option, and shareable objects must be enabled for
runtime linking with the <code>-G</code> linker option.  Without these
options, the AIX linker will not automatically link to any shared object
with a <code>.so</code> extension.  Also, the R main program is unable to
dynamically load modules (such as X11) with the <code>dlopen</code> call.

   <p>When setting <code>MAIN_LDFLAGS</code> and <code>SHLIB_LDFLAGS</code> accordingly,
note that linker flags must be escaped using <code>-Wl,</code> if
<code>gcc</code> is used for linking: use <code>MAIN_LDFLAGS="-Wl,brtl"</code>
and <code>SHLIB_LDFLAGS="-Wl,-G"</code> in this case.

   <p>Harald Servat Gelabert &lt;harald at cepba dot upc dot es&gt; reported
success building R 1.7.0 under AIX 5.1 with

<pre class="smallexample">     CC=xlc
     F77=xlf
     CXX=xlC
     CFLAGS=-O3 -qstrict -qmaxmem=8192
     FFLAGS=-O3 -qstrict -qmaxmem=8192
     CXXFLAGS=-O2 -qmaxmem=8192
     MAIN_LDFLAGS=-Wl,-brtl
     SHLIB_LDFLAGS=-Wl,-G
     </pre>

<p>but was unable to use the X libraries or the native BLAS (ESSL) and so
used <code>--without-x --without-blas</code>.

<a name="New%20platforms"></a>

<h2 class="appendix">Appendix C New platforms</h2>

<p>There are a number of sources of problems when installing R on a new
hardware/OS platform.  These include

   <p><strong>Floating Point Arithmetic</strong>: R supports the <small>POSIX</small>,
SVID and <small>IEEE</small> models for floating point arithmetic.  The
<small>POSIX</small> and SVID models provide no problems.  The <small>IEEE</small>
model however can be a pain.  The problem is that there is no agreement
on how to set the signalling behaviour; Sun/Sparc, SGI/IRIX and ix86
Linux require no special action, FreeBSD requires a call to (the macro)
<code>fpsetmask(0)</code> and OSF1 requires that computation be done with a
<code>-ieee_with_inexact</code> flag etc.  On a new platform you must find
out the magic recipe and add some code to make it work.  This can often
be done via the file <code>config.site</code> which resides in the top level
directory.

   <p>Beware of using high levels of optimization, at least initially.  On
many compilers these reduce the degree of compliance to the
<small>IEEE</small> model.  For example, using <code>-fast</code> on the Solaris
SunPro compilers causes R's <code>NaN</code> to be set incorrectly.

   <p><strong>Shared Libraries</strong>: There seems to be very little agreement
across platforms on what needs to be done to build shared libraries.
there are many different combinations of flags for the compilers and
loaders.  GNU libtool cannot be used (yet), as it currently does not
fully support FORTRAN (and will most likely never support <code>f2c</code>:
one would need a shell wrapper for this).  The technique we use is to
first interrogate the X window system about what it does (using
<code>xmkmf</code>), and then override this in situations where we know
better (for tools from the GNU Compiler Collection and/or platforms we
know about).  This typically works, but you may have to manually
override the results.  Scanning the manual entries for <code>cc</code> and
<code>ld</code> usually reveals the correct incantation.  Once you know the
recipe you can modify the file <code>config.site</code> (following the
instructions therein) so that the build will use these options.

   <p>If you do manage to get R running on a new platform please let us
know about it so we can modify the configuration procedures to include
that platform.

   <p>If you are having trouble getting R to work on your platform please
feel free to get in touch to ask questions.  We have had a fair amount
of practice at porting R to new platforms <small class="enddots">....</small>

<a name="Function%20and%20variable%20index"></a>

<h2 class="unnumbered">Function and variable index</h2>

<ul class="index-vr" compact>
<li><code>configure</code>: <a href="#Using%20make">Using make</a>, <a href="#Configuration%20variables">Configuration variables</a>, <a href="#Installation">Installation</a>, <a href="#Simple%20compilation">Simple compilation</a>
<li><code>install.packages</code>: <a href="#Installing%20packages">Installing packages</a>
<li><code>make</code>: <a href="#Using%20make">Using make</a>
<li><code>R_HOME</code>: <a href="#Simple%20compilation">Simple compilation</a>
<li><code>remove.packages</code>: <a href="#Removing%20packages">Removing packages</a>
<li><code>rsync</code>: <a href="#Using%20rsync">Using rsync</a>
<li><code>update.packages</code>: <a href="#Updating%20packages">Updating packages</a>
</ul>
<a name="Concept%20index"></a>

<h2 class="unnumbered">Concept index</h2>

<ul class="index-cp" compact>
<li>AIX: <a href="#AIX">AIX</a>
<li>BLAS library: <a href="#Solaris">Solaris</a>, <a href="#MacOS%20X">MacOS X</a>, <a href="#Using%20FORTRAN">Using FORTRAN</a>, <a href="#Linear%20algebra">Linear algebra</a>
<li>FORTRAN: <a href="#Using%20FORTRAN">Using FORTRAN</a>
<li>Help pages: <a href="#Simple%20compilation">Simple compilation</a>
<li>HP-UX: <a href="#HP-UX">HP-UX</a>
<li>Installation: <a href="#Installation">Installation</a>
<li>Installing under Unix: <a href="#Installing%20R%20under%20Unix">Installing R under Unix</a>
<li>Installing under Windows: <a href="#Installing%20R%20under%20Windows">Installing R under Windows</a>
<li>IRIS: <a href="#IRIX">IRIX</a>
<li>LAPACK library: <a href="#Solaris">Solaris</a>, <a href="#MacOS%20X">MacOS X</a>, <a href="#Linear%20algebra">Linear algebra</a>
<li>Linux: <a href="#Linux">Linux</a>, <a href="#Installing%20R%20under%20Unix">Installing R under Unix</a>
<li>MacOS X: <a href="#MacOS%20X">MacOS X</a>, <a href="#Installing%20R%20under%20Unix">Installing R under Unix</a>
<li>Manuals: <a href="#Making%20the%20manuals">Making the manuals</a>
<li>Manuals, installing: <a href="#Installation">Installation</a>
<li>Obtaining R: <a href="#Obtaining%20R">Obtaining R</a>
<li>Packages: <a href="#Add-on%20packages">Add-on packages</a>
<li>Packages, installing: <a href="#Installing%20packages">Installing packages</a>
<li>Packages, removing: <a href="#Removing%20packages">Removing packages</a>
<li>Packages, updating: <a href="#Updating%20packages">Updating packages</a>
<li>Solaris: <a href="#Solaris">Solaris</a>
<li>Sources for R: <a href="#Getting%20and%20unpacking%20the%20sources">Getting and unpacking the sources</a>
</ul>


   <div class="footnote">
<hr>
<h4>Footnotes</h4>
<ol type="1">
<li><a name="fn-1"></a>
<p>if you have the source-code package
files installed</p>

   <li><a name="fn-2"></a>
<p>Using the SunPro aka Forte aka Sun ONE <code>cc</code> and
<code>f95</code> compilers</p>

   <li><a name="fn-3"></a>
<p>On HP-UX <code>fort77</code> is the <small>POSIX</small>
compliant FORTRAN compiler, and comes second in the search list.</p>

   <li><a name="fn-4"></a>
<p>as well as its equivalence to the <code>Rcomplex</code>
structure defined in <code>R_ext/Complex.h</code>.</p>

   <li><a name="fn-5"></a>
<p>In particular, avoid <code>g77</code>'s
<code>-pedantic</code>, which gives confusing error messages.</p>

   <li><a name="fn-6"></a>
<p>e.g., to use an optimized BLAS on
Sun/Sparc</p>

   </ol><hr></div>

   </body></html>