Build LBNE Software Suites

This package is used to build the software suites for LBNE. It makes use of waf and worch.

Note: this is not yet ready to use. Soon!

Prerequisites

On all platforms, some OS-level packages are implicitly required.

compiler

a compiler capable to build the version of GCC from source

autoconf

and related tools

Red Hat Enterprise based (eg. Scientific Linux)

A base install of SL6.4 has most of the packages required. The exceptions are:

• t.b.d.

Debian based (eg. Ubuntu)

You will need at least these:

$ sudo apt-get install \
       libbz2-dev libx11-dev  libxmu-dev \
       libgl1-mesa-dev libglu1-mesa-dev \
       doxygen-latex libtool libbz2-dev \
       libreadline-dev automake libncurses5-dev \
       tcl8.5-dev

Fedora Core 20

Starting from a base netinst of FC20 here are the packages that need to be installed:

$ sudo yum install python-virtualenv git gcc-c++ m4 subversion patch automake flex \
                   ncurses-devel zlib-devel libtool bzip2-devel \
                   libX11-devel libXmu-devel libXpm-devel libXft-devel \
                   mesa-libGL-devel mesa-libGLU-devel \
                   doxygen-latex tcl-devel \
                   libtool libbz2-dev libreadline-dev automake \
                   expat-devl readline-devel krb5-devel

Instructions

The build environment is provided by a few Python modules. It is recommended that these be installed into a virtualenv virtual environment. To set up for this do:

$ virtualenv /path/to/venv
$ source /path/to/venv/bin/activate

The rest of the instructions assume this.

Install the latest release build environment

Releases of lbne-build and its few dependencies are placed on PyPI and may be installed simply via:

$ pip install lbne-build

Installing an older release

T.B.D.

Selecting the suite to build

The installation is driven by a top level configuration file which will be found:

$ ls /path/to/venv/share/worch/config/lbne/suite-*.cfg

Prepare for building

The build will need a substantial amount of disk space (10’s of GB) for intermediate build products. This is best located on fast, local disk. After the build is complete, it can be removed or, if more than one suite is desired, reused to avoid rebuilding shared packages multiple times. A final installation location must also be determined. This will require many GBs of space. With the suite name determined from the above listing the build is configured with:

$ cd /path/to/build
$ cp /path/to/venv/share/worch/wscripts/lbne/wscript .
$ waf --prefix=/path/to/install \
      --orch-config=lbne/suite-<SUITE>.cfg \
        configure 

Building the software

Finally, to kick off the build one simply runs:

$ waf

The intermediate files will be populated below the directory ./tmp/.

Development

To develop lbne-build itself you need a way to edit its files, install and exercise them.

Either create a new Virtual Environment (run ”deactivate” to deactivate the current one from any activated shells) or reuse the one from above. The initial setup goes like:

$ virtualenv /path/to/venv
$ source /path/to/venv/bin/activate
$ mkdir -p /path/to/work
$ cd /path/to/work

$ git clone git@github.com:brettviren/worch-ups.git
$ cd worch-ups 
$ python setup.py sdist
$ pip install dist/worch-ups-X.Y.tar.gz 
$ cd ..

$ git clone git@github.com:LBNE/lbne-build.git
$ cd lbne-build
$ python setup.py sdist
$ pip install dist/lbne-build-X.Y.tar.gz
$ cd ..

Replace X.Y with whatever version is built.

After hacking on the lbne-build file reinstall everything like:

$ cd lbne-build
$ pip uninstall -y lbne-build
$ python setup.py sdist
$ pip install dist/lbne-build-X.Y.tar.gz
$ cd -

To actually run the build one does the same as if installing an official release:

$ cd /path/to/build
$ cp /path/to/venv/share/worch/wscripts/lbne/wscript .
$ waf --prefix=/path/to/install \
      --orch-config=lbne/suite-<SUITE>.cfg \
        configure 
$ waf

Note, this still installs Worch and other required Python packages automatically. If you need to hack on them you can similarly ”pip uninstall” them from the Virtual Environment, “git clone” their source, ”python setup.py sdist” to make the dist/*.tar.gz file and ”pip install” that.