Release tars available from

The P4P modules requires:

  • Python 2.7 or >=3.5 (>=3.6 for asyncio support)

  • numpy >=1.6

  • Cython >=0.25.2

  • nose2 (Optional, recommended to run unittests)


Current (>= 4.0)

  • EPICS Base >= 3.14.12

  • PVXS >= 0.2.0

Prior to 4.0

  • EPICS >= 7.0.2


  • EPICS Base >= 3.14.12

  • pvDataCPP >=7.1.0

  • pvAccessCPP >=6.1.0

P4P can be built and installed in one of two ways. As a python package, preferably managed by PIP. As an EPICS module.


  • cothread needed by p4p.client.cothread.Context.

  • qtpy needed for p4p.client.Qt.Context.

Local (re)build as Python package

The process for building as a python package using sources from by adding the argument “–no-binary” to prevent the use of pre-built binarys (except for numpy).

python -m virtualenv p4ptest
. p4ptest/bin/activate
python -m pip install -U pip
python -m pip install \
 --only-binary numpy \
 --no-binary epicscorelibs,pvxslibs,p4p \
 nose2 p4p
python -m nose2 p4p   # Optional: runs automatic tests

Bootstrap a virtualenv offline

It may be necessary to create a virtualenv including P4P on a machine with no internet access, or to ensure the use of certain verified binary/source. The following recipe was tested with virtualenv 16.1 and pip 18.1.

First, from a machine with internet access, and having the same archetecture as the target machine, collect the necessary files.

mkdir /tmp/venv
cd /tmp/venv
python -m pip download virtualenv
unzip virtualenv-*.whl
python --never-download prepenv
. prepenv/bin/activate
pip download -d virtualenv_support nose2 p4p
tar -caf p4p-env-`date -u +%Y%m%d`.tar.gz virtualenv_support

Now move the created file p4p-env-*.tar.gz to the target machine. Then prepare the virtualenv env with.

tar -xaf p4p-env-*.tar.gz
python --never-download env
. env/bin/activate
pip install --no-index -f virtualenv_support p4p
python -m nose2 p4p   # Optional: runs automatic tests

Utilities to automate this process include

Build as EPICS Module

P4P can also be built as an EPICS Module with additional python dependencies.

Prepare the host to build python extensions. eg. a Debian Linux host:

sudo apt-get install build-essential python3-dev

Install python dependencies on eg. a Debian Linux host:

sudo apt-get install python3-numpy python3-nose2 cython3

or with PIP:

python3 -m pip install -r requirements-latest.txt

From from versioned source. (may replace “master” with release version number):

git clone --branch master
cd p4p

Set location of EPICS modules. With EPICS >= 7.0.2:

cat <<EOF > configure/RELEASE.local

See below for details on building EPICS from source.

By default P4P will build using ‘python’ found in the system search path. To explicitly specify a particular version.

make distclean
make PYTHON=python3

Alternately, the full path of the interpreter can be used.

make distclean
make PYTHON=/usr/bin/python3

For convenience PYTHON can also be set in configure/CONFIG_SITE

Multiple Python Versions

To build for multiple python versions it is necessary to do a partial clean before building another version. This will not remove the final tree.

make PYTHON=python2
make PYTHON=python2 clean
make PYTHON=python3
make PYTHON=python3 clean


If PYTHON= is ever specified, then it must be specified for all targets except ‘distclean’.

Building EPICS dependencies

If the necessary EPICS modules are not present, then they may be built from source.

sudo apt-get install libreadline6-dev libncurses5-dev perl
git clone
git clone
cat <<EOF > pvxs/configure/RELEASE.local
cat <<EOF > p4p/configure/RELEASE.local
make -C epics-base
make -C pvxs

CLI and unittests

To run the unittests:

make nose

For testing purposes several simple command line client tools are provided. For further information run:

PYTHONPATH=$PWD/python2.7/linux-x86_64 python -m p4p.client.cli -h