**** Copyright (c) 2004 by Charles A. Loomis, Jr. and Le Centre National de
**** la Recherche Scientifique (CNRS).  All rights reserved.
****
**** The software was distributed with and is covered by the European
**** DataGrid License.  A copy of this license can be found in the included
**** LICENSE file.  This license can also be obtained from
**** http://www.eu-datagrid.org/license.

This software is intended to transform a comps.xml file (used to
describe package groups for RedHat-like systems) into a set of PAN
templates suitable for use with quattor.  The transformation
faithfully preserves the group structure of the source comps.xml file.

Prerequisites:

1) The source comps.xml file for your RedHat-like distribution.  The
most reliable place to get the file is from the ISO image of the first
installation disk.  Mount the image and look in the directory:
distname/base where distname is the name of the distribution.  For
Fedora this is "Fedora/base" and for Scientific Linux, "SL/base".

NOTE: You must edit the comps.xml file and delete the DOCTYPE tag.
Most XML parsers will try to validate the file against the DTD and
will fail if it isn't available. 

2) All of the RPMs of the distribution must be locally accessible.
Either mount the ISO images of the installation disks or copy the full
set of RPMs from a mirror.

3) An XSLT processor compliant with XSLT 2.0!  The transform has been
tested with the Saxon processor; use version 8.5 or later.  It can be
downloaded from: http://saxon.sourceforge.net/.

4) The Saxon processor is written in Java and requires a Java Virtual
Machine.  Install a real commercial JVM which supports Java 1.4 or
later.  The GJC (GNU Java) will not work!  You can tell which java you
have by running "java -version" from the command line.

5) Perl.  Scripts are written in perl; you won't get far without it. 

Template Generation:

1) Extract the "provides" information from the distribution's RPMs.
The command for doing this is:

./rpmProvides.pl \
    /mnt/fc1/Fedora/RPMS/ \
    /mnt/fc2/Fedora/RPMS/ \
    /mnt/fc3/Fedora/RPMS/ \
    /home/loomis/fedora/updates/ > rpm-provides-fc1.txt

The arguments are the list of directories containing the
distribution's RPMs.  If you have updates to the standard packages,
put them in a separate directory and put it LAST on the command line.
Redirect the output to a file. 

2) Extract the "requires" information from the RPMs and from the
"provides" information of the above step.  The command for doing this
is: 

./rpmRequires.pl \
    rpm-provides-fc1.txt \
    /mnt/fc1/Fedora/RPMS/ \
    /mnt/fc2/Fedora/RPMS/ \
    /mnt/fc3/Fedora/RPMS/ \
    /home/loomis/fedora/updates/ > rpm-requires-fc1.txt

The first argument is the name of the file with the provides
information.  Follow this with the same list of directories containing
the RPMs.  Redirect the output to a file. 

3) Transform the comps.xml file into a set of pan templates.  

export CLASSPATH=<full path to saxon jar file>
java net.sf.saxon.Transform \
    comps.xml \
    comps2pan.xsl \
    depdb=rpm-requires-fc1.txt \
    output-dir=`pwd` \
    pan-prefix=pro_fc1_

The CLASSPATH must be set to the full path of the Saxon jar file.  The
first two arguments of the transform are the comps.xml file and the
transformation stylesheet.  Replace the value of the depdb option with
the file you produced in step 2.  Define output-dir to be the existing
directory where you want to write the generated pan templates.  (This
is used in an URL, so use must use '/' as the directory separator for
all platforms.)  Set the pan-prefix to the desired prefix for the pan
templates.  This is typically "pro_<distname>_" if you don't use
namespaces and the empty string if you do.

4) Verify that the syntax of the generated files are correct. 

panc --check `find <output-dir> *.tpl`

where <output-dir> is replaced with the output directory you used
above.  Typically you'll need to fix by hand some of the global
variables which have been generated for the kernel and kernel modules.