=head1 NAME

rrdbuild - Instructions for building RRDtool

=head1 DESCRIPTION

=head2 Overview

If you downloaded the source of rrdtool you have to compile it. This
document will give some information on how this is done.

RRDtool relies on services of thrid part libraries. Some of these libraries
may already be installed on your system. You have to compile copies of the other
ones before you can build RRDtool.

This document will tell you about all the necessary steps to get going.

=head2 Building

Before you start to build RRDtool, you have to decide two things:

=over

=item 1.

In which directory you want to build the software.

=item 2.

Where you want to install the software.

=back

Once you have decided. Save the two locations into environment variables.
Depending on the shell you are using, you can do either (bash,zsh):

 BUILD_DIR=/tmp/rrdbuild
 INSTALL_DIR=/usr/local/rrdtool-1.2.9

Or if you run tcsh:

 set BUILD_DIR=/tmp/rrdbuild
 set INSTALL_DIR=/usr/local/rrdtool-1.2.9

Now make sure the BUILD_DIR exists and go there:

 mkdir -p $BUILD_DIR
 cd $BUILD_DIR

Lets first assume you already have all the necessary libraries
per-installed. Note that these instructions assume that your copies of
B<tar> and B<make> are actually B<GNU tar> and B<GNU make> respectively. It
could be that they are installed as B<gtar> and B<gmake> on your system.

 wget http://people.ee.ethz.ch/oetiker/webtools/rrdtool/pub/rrdtool-1.2.9.tar.gz
 tar zxf rrdtool-1.2.9.tar.gz
 cd rrdtool-1.2.9
 ./configure --prefix=/path/to/installation && make && make install

Ok, this was very optimistic. This try will probably have ended with
B<configure> complaining about several missing libraries. If you are on a
Linux or *bsd system you may want to just install the missing bits from your
software repository. When you do that, make sure you also get the B<-dev>
package for each library you install. Once you have the missing bits on
board, just re-run the last line of the instructions above.

But again this may have been too optimistic, and you actually have to
compile your own copies of the required libraries. Here is how:

=over

=item Building cgilib

 cd $BUILD_DIR
 wget http://people.ee.ethz.ch/oetiker/webtools/rrdtool/pub/libs/cgilib-0.5.tar.gz
 tar zxf cgilib-0.5.tar.gz
 cd cgilib-0.5

If you are on Mac OSX you want to fix a little header problem here by doing

 touch malloc.h

and now you are ready to build

 make CC=gcc CFLAGS="-O3 -fPIC -I."
 mkdir -p $BUILD_DIR/lb/include
 cp *.h $BUILD_DIR/lb/include
 mkdir -p $BUILD_DIR/lb/lib
 cp libcgi* $BUILD_DIR/lb/lib

=item Building zlib

 cd $BUILD_DIR
 wget http://people.ee.ethz.ch/oetiker/webtools/rrdtool/pub/libs/zlib-1.2.2.tar.gz
 tar  zxf zlib-1.2.2.tar.gz
 cd zlib-1.2.2
 env CFLAGS="-O3 -fPIC" ./configure --prefix=$BUILD_DIR/lb
 make
 make install

=item Building libpng

Libpng itself requires zlib to build, so we need to help a bit. If you
already have a copy of zlib on your system (which is very likley) you can
drop the settings of LDFLAGS and CPPFLAGS. Note that the backslash (\) at
the end of line 4 means that line 4 and line 5 are on one line.

 cd $BUILD_DIR
 wget http://people.ee.ethz.ch/oetiker/webtools/rrdtool/pub/libs/libpng-1.2.8-config.tar.gz
 tar zxvf libpng-1.2.8-config.tar.gz
 cd libpng-1.2.8-config
 env CPPFLAGS="-I$BUILD_DIR/lb/include" LDFLAGS="-L$BUILD_DIR/lb/lib" CFLAGS="-O3 -fPIC" \
     ./configure --disable-shared --prefix=$BUILD_DIR/lb
 make
 make install

=item Building freetype

 cd $BUILD_DIR
 wget http://people.ee.ethz.ch/oetiker/webtools/rrdtool/pub/libs/freetype-2.1.9.tar.gz
 tar zxvf freetype-2.1.9.tar.gz
 cd freetype-2.1.9
 env CPPFLAGS="-I$BUILD_DIR/lb/include" LDFLAGS="-L$BUILD_DIR/lb/lib" CFLAGS="-O3 -fPIC" \
     ./configure --disable-shared --prefix=$BUILD_DIR/lb
 make
 make install

=item Building libart_lgpl

 cd $BUILD_DIR
 wget http://people.ee.ethz.ch/oetiker/webtools/rrdtool/pub/libs/libart_lgpl-2.3.17.tar.gz
 tar zxvf libart_lgpl-2.3.17.tar.gz
 cd libart_lgpl-2.3.17
 env CFLAGS="-O3 -fPIC" ./configure --disable-shared --prefix=$BUILD_DIR/lb
 make
 make install

=back

Now all the dependent libraries are built and you can try again. Since these
are static libraries, you may have to use F<ranlib> to make them accessible.
Especially BSD systems like Mac OS X may require this, Linux and Solaris
will do just fine without since their F<ar> command does ranlibs job as well.

 ranlib $BUILD_DIR/lb/lib/*.a

This time you tell configure where it should be looking for libraries and
include files. This is done via environment variables. Depending on the
shell you are running, the syntax for setting environment variables is
different. Under csh/tcsh you use:

 set IR=-I$BUILD_DIR/lb/include
 setenv CPPFLAGS "$IR $IR/libart-2.0 $IR/freetype2 $IR/libpng"
 setenv LDFLAGS  -L$BUILD_DIR/lb/lib
 setenv CFLAGS -O3

If you are running bash/sh/ash/ksh/zsh use this:

 IR=-I$BUILD_DIR/lb/include
 CPPFLAGS="$IR $IR/libart-2.0 $IR/freetype2 $IR/libpng"
 LDFLAGS="-L$BUILD_DIR/lb/lib"
 CFLAGS=-O3
 export CPPFLAGS LDFLAGS CFLAGS

And finally try building again. We disable the python bindings because it
seems that a fair number of people have ill configured python setups that
would prevent rrdtool from building if they are included in their current
state.

 cd $BUILD_DIR
 ./configure --prefix=$INSTALL_DIR --disable-python && make && make install

Now go to I<$INSTALL_DIR>B</examples> and run them to see if your
build has been successful.

=head1 AUTHOR

Tobias Oetiker E<lt>oetiker@ee.ethz.chE<gt>