Pure Data as an embeddable audio synthesis library
Copyright (c) Peter Brinkmann & the libpd team 2010-2023
For documentation of libpd, see the wiki: https://github.com/libpd/libpd/wiki
If you are using Processing, iOS, or Android, see our companion repositories:
The preferred method to download libpd is to use git.
Do not download libpd as a zip or tar.gz file from GitHub.
The "Download ZIP" button may look like a good idea, but currently Github does not include submodule files when compiling zip files. This means the zip file is missing the main pd source files and you will not be able to build libpd, with errors such as: No rule to make target pure-data/src/d_arithmetic.o or No such file or directory: pure-data/extra/bonk~/bonk~.c.
To download libpd & check out the pure-data submodule, do the following:
git clone --recurse-submodules https://github.com/libpd/libpd.git
You should now have a libpd
directory and the libpd/pure-data
directory should contain the pd sources.
Note: If your version of git does not support "--recurse-submodules", you can run the git submodule commands in the libpd directory itself after cloning:
cd libpd
git submodule update --init --recursive
For most uses, it is recommended to check out the latest stable release version via a git tag. For example, to switch to libpd version 0.8.3 after cloning:
git checkout 0.8.3
git submodule update
The master branch contains the latest libpd development and can be considered generally stable. However, we make no guarantees. :)
The directory containing the sources of Pd Vanilla and it's standard externals. This is a git submodule of Miller Puckette's official Pd git repository:
git:https://git.code.sf.net/p/pure-data/pure-data
This directory contains the source files that make up the core of libpd.
Build support for various platforms. Feel free to improve the build system in any way you see fit.
Glue for using libpd with C++, C#, Java, Objective-C, and Python. Feel free to improve or add support for other languages such as Lua.
Small sample programs and tests in the various supported languages.
The build result location and required software libraries for the various supported languages.
Core build requirments:
- Unix command shell: bash, dash, etc
- C compiler chain: gcc/clang & make
Note: The various language wrappers may have additional requirements.
Currently the main Makefile builds a dynamic lib on Windows (in MinGW), Linux, & macOS and has the following targets:
- libpd: build the libpd C core, default if no target is specified
- csharplib: build libpdcsharp
- javalib: build libpdnative and the jni wrapper
- javadoc: generate Java HTML documentation
- javasrc: create a Java source jar
- clean: remove object files
- clobber: remove linked library files
- install: install libpd C library and C/C++* headers, set location with prefix= (default: /usr/local)
- uninstall: remove libpd C library and C/C++ headers, set location with prefix= (default: /usr/local)
* C++ headers are only installed if the C utility layers were built as well (ie. UTIL=true), see below.
Makefile options allow for conditional compilation of libpd util and pd extra externals sources into libpd as well as other options:
- UTIL=true: compile utilities in
libpd_wrapper/util
(default) - EXTRA=true: compile
pure-data/extra
externals which are then inited in libpd_init() (default) - MULTI=true: compile with multiple instance support
- DOUBLE=true: compile with double-precision support
- DEBUG=true: compile with debug symbols & no optimizations
- STATIC=true: compile static library (in addition to shared library)
- FAT_LIB=true: compile universal "fat" lib with multiple architectures (macOS only)
- LOCALE=false: do not set the LC_NUMERIC number format to the default "C" locale* (default)
- PORTAUDIO=true: compile with portaudio support (currently JAVA jni only)
- JAVA_HOME=/path/to/jdk: specify the path to the Java Development Kit
To build the libpd C core with default options:
make
To build libpd without the util libs and extra externals:
make UTIL=false EXTRA=false
Note: The C++ wrapper requires UTIL=true as it uses the ringbuffer.
* See the Known Issues section for more info.
If you need to add custom search paths to the CFLAGS or LDFLAGS, you can specify them when building via the ADDITIONAL_* variables:
make ADDITIONAL_CFLAGS="-I/usr/local/include" \
ADDITIONAL_LDFLAGS="-L/usr/local/lib"
Once libpd has built successfully, the compiled libraries will be found in the libs
directory.
Install the core build requirements using your distribution's package manager. For Debian, you can install the compiler chain, autotools, & gettext with:
sudo apt-get install build-essentials
macOS is built on top of a BSD system and the bash or zsh commandline can be accessed with the Terminal application in the /Applications/Utility directory.
The clang compiler and associated tools are provided by Apple. If you are running macOS 10.9+, you do not need to install the full Xcode application and can install the Commandline Tools Package only by running the following:
xcode-select --install
If you are running macOS 10.6 - 10.8, you will need to install Xcode from the Mac App Store or downloaded from https://developer.apple.com
By building with the FAT_LIB=true
Makefile option, libpd will be compiled with support for multiple architectures depending on the detected system version:
- macOS <= 10.13: i386 (32 bit Intel) & x86_64 (64 bit Intel)
- macOS >= 11.0: arm64 (64 bit Arm) & x86_64 (64 bit Intel)
To override autodetection, specify the -arch
flags directly using the FAT_ARCHS
Makefile option:
make FAT_LIB=true FAT_ARCHS="-arch i386 -arch x86_64"
libpd on Windows can be built with MinGW which provides the core build requirements: a compiler chain & shell environment. It is recommended to use the Msys2 distribution which provides both a Unix command shell and MinGW. Download the Msys2 "x86_64" 64 bit installer (or "i686" if you are using 32 bit Windows) from:
Then install to the default location (C:\msys32 or C:\msys64) and follow the setup/update info on the Msys2 webpage.
Msys2 provides both 32 and 64 bit MinGW and command shells which are used to compile for 32 or 64 bit, respectively. Due to how MinGW is designed, you cannot build a 64 bit libpd with a 32 bit MinGW and vice versa.
Note: Msys2 development seems to change frequently, so some of the package names below may have changed after this document was written.
Open an Msys2 shell and install the compiler chain & make via:
# 32 bit
pacman -S mingw-w64-i686-toolchain mingw-w64-i686-clang make
# 64 bit
pacman -S mingw-w64-x86_64-toolchain mingw-w64-x86_64-clang make
You can also search for packages in Msys2 with pacman -S -s <searchterm>
.
Once the packages are installed, you should now be ready to build libpd.
Note: For 64 bit Windows, build Pd with the following additional C flags to ensure long integers are the correct length:
make ADDITIONAL_CFLAGS='-DPD_LONGINTTYPE="long long"'
If you run into strange errors such as /bin/sh: cc: command not found
, try closing and reopening your shell window before building again.
By default, libpd computes numbers and samples internally as single-precision 32-bit floats. This is fast and good enough for most general usage. If you are working with small numbers beyond 6 decimal points, however, you will need a higher degree of precision.
To enable double-precision 64-bit floating point support, build libpd with -DPD_FLOATSIZE=64
in the CFLAGS. The libpd makefile provides the DOUBLE
makefile variable for this:
make DOUBLE=true
Now utilize the libpd API which use the double type, such as the libpd double hook, libpd_add_double()
, and libpd_process_double()
.
To double-check your build, the following will print a 1 if double-precision support is enabled:
printf("double-precision %d\n", (int)(sizeof(t_float)/8));
Libpd can load pre-compiled dynamic libraries of external objects in the same manner as desktop Pd if it is compiled with the -ldl
LDFLAG. This is done by default in the Makefile. The main difference is that libpd does not inherit the same search paths by default, so paths outside of those specified by the loading patch with [declare]
objects need to be added via libpd's "add to search path" function.
If libpd is compiled with multi-instance support via make MULIT=true
and -DPDINSTANCE -DPDTHREADS
are defined, it may have trouble loading dynamic externals which are compiled without them:
vbap.pd_linux: vbap.pd_linux: undefined symbol: pd_this
vbap
... couldn't create
In this case, the vbap external needs to be recompiled with the -DPDINSTANCE -DPDTHREADS
CFLAGS defined to add multi-instance support to match.
The C++ wrapper is inspired by the Java wrapper and provides a PdBase class as well as listener, list, and message type classes. This is a header only library so you only need to include the cpp
directory in your project. You also may need to add libpd_wrapper/util
to you include paths.
Sample programs are found in samples/cpp
.
By default, PdBase always wraps to the single main libpd instance, so it is recommended to only use one instance of PdBase.
If PDINSTANCE
is defined, each PdBase will wrap a separate libpd instance.
To enable multiple instance support:
- Build libpd with
make MULTI=true
or with CFLAGS-DPDINSTANCE -DPDTHREADS
- Build the C++ program using PdBase.hpp with CPPFLAGS
-DPDINSTANCE
The libpd C# library is available as a NuGet package:
https://www.nuget.org/packages/LibPdBinding
If your platform's native libpdcsharp.(so/dll) is not included, you have to build it yourself following and copy the resulting file to the output directory. Batch scripts for compilation on Windows with MinGW are included.
The C# library expects a file libpdcsharp.(so/dll) in its directory. Before using the project, you need to compile it:
make csharplib
Include csharp/LibPdBinding.csproj
in your solution and reference the project in your application. See csharp/README.txt
for details.
The wrapper can be built with MinGW. See the previous "Windows" section for instructions on setting up a MinGW-based build environment using Msys2.
Build libpdcsharp using the the .bat DOS batch file wrappers for make to match which MinGW you are using:
# 32 bit
./mingw32_build_csharp.bat
# 64 bit
./mingw64_build_csharp.bat
Usually you want the 32 bit version, as it will work on 64 bit Windows as well. However so C# environments require a 64 bit version, Unity 5 for instance.
Once the build is finished, a libpdcsharp.(so/dll) library should be found in the libs
directory
You also may need to use the libwinpthread library along with libpdcsharp. This included with libpd in the libs
directory, either within libs/mingw32
or libs/mingw64
. For a current version of libwinpthread-1.dll
search in your Msys2 installation's bin
directory.
Note: If you have installed Msys2 to a non-default location, you will need to change the variable for %MSYS2%
in the .bat files.
If you want to use the library on Linux with Mono, you need the following changes to the LibPdBinding project:
- Compile the .so file with
make csharplib
. - Remove
libpdcsharp.dll
andlibwinpthread-1.dll
from LibPdBinding project. - Add
libpdcsharp.so
to the LibPdBinding project. - Set "Copy to Output Directory" for
libpdcsharp.so
to "Copy always"
May be out of date
Ready-made binaries for Java are available at the libpd-java-build repository:
https://github.com/wivlaro/libpd-java-build
You will need the Java Development Kit (JDK) to build the libpd Java lib. Make sure the JDK/bin path is added to your $PATH shell variable and, optionally, the JAVA_HOME variable points to the JDK location.
Build the libpd Java lib with:
make javalib
This should result in a libpd.jar and pdnative.(so/dll) in the libs
directory.
Optionally, you can build libpd with Eclipse using the .classpath
& .project
workspace files.
You can also build a libpd source jar and Java HTML documentation:
make javasrc
make javadoc
This should result in a libs/libpd-sources.jar
and a javadoc
directory.
Install the JDK via your distributions package manager.
Install the JDK either by downloading an installer package or by using one of the open source package managers for macOS:
- Homebrew: https://brew.sh (recommended)
- MacPorts: https://www.macports.org
The wrapper can be built with MinGW. See the previous "Windows" section for instructions on setting up a MinGW-based build environment using Msys2.
Install the JDK by downloading an installer package, then add the path to JDK/bin to your $PATH shell variable and the JDK path to $JAVA_HOME (optional). If the JDK is installed to C:\Program Files\Java\jdk1.8.0_152
, add the following to your ~/.bash_profile:
# add JDK bin path
export PATH=$PATH:'C:\Program Files\Java\jdk1.8.0_152\bin'
# JDK path (optional)
export JAVA_HOME=C:/Program\ Files/Java/jdk1.8.0_152
Restart your shell if it's open.
Build the libpd javalib with:
make javalib
You can also set the JAVA_HOME path when running make with:
make javalib JAVA_HOME=C:/Program\ Files/Java/jdk1.8.0_152
Once the build is finished, you should find libpd.jar and pdnative.(so/dll) in the libs
directory.
The Objective-C wrapper is designed to be used on iOS and macOS and includes a (currently iOS-only) audio unit and audio manager for sound I/O.
libpd.xcodeproj provides an Xcode project to build libpd + the Obj-C wrapper as a static library for iOS & macOS. Drag the libpd project into your existing Xcode project, then add libpd-ios (or libpd-osx) to the Linked Frameworks and Libraries in the General tab of your project target.
The Xcode project builds the following targets:
- libpd-ios: libpd and the Obj-C wrapper for iOS
- libpd-osx: libpd and the Obj-C wrapper for macOS
- libpd-ios-multi: libpd for iOS with multiple instance support
- libpd-osx-multi: libpd for macOS with multiple instance support
For detailed instructions, see Working with libpd in Xcode
If you are unfamiliar with how static libraries work or how to use them in Xcode, see this useful tutorial.
Note: libpd is tested with the release versions of Xcode. It is recommended that you avoid using beta or developer preview versions.
If you are using Xcode to build iOS apps, you can use CocoaPods to add libpd to your project.
Use the following in your CocoaPods podfile:
pod 'libpd', :git => 'https://github.com/libpd/libpd', :submodules => true
To specify a stable version tag, use the :tag
option:
pod 'libpd', :git => 'https://github.com/libpd/libpd', :submodules => true, :tag => '0.12.1'
The Python wrapper provides a "pylibpd" module mirroring the libpd C API. Build the wrapper with:
cd python
make
See the sample programs in samples/python
.
If you have multiple versions of Python on your system, you can specify which is used to build the module via the PYTHON makefile option:
make PYTHON=python3
If you are building for 64-bit Windows, you may need to set the additional MS_WIN64 flag:
make CFLAGS="-DMS_WIN64=1"
Some samples require the "pyaudio" Portaudio library.
If you install pyaudio with pip
, you will need to install Portaudio first. On macOS, for example, you can install it with Homebrew:
brew install portaudio
There is currently a third party Rust wrapper for libpd called libpd-rs which is actively maintained.
libpd-rs uses a fork of libpd which is frequently synchronized with upstream.
Please visit the repository for detailed information.
CMake can be used to build the libpd C library.
CMake is a cross-platform, open-source build system. CMake is used to control the software compilation process using simple platform and compiler independent configuration files, and generate native makefiles and workspaces that can be used in the compiler environment of your choice.
- CMake: You can download CMake for your platform here. The minimum version is 2.8.11.
- pthreads: On Windows, you need to provide a pthreads library.
If you are using MinGW, you can use the libwinpthread-1.dll included in the libs/mingw*
directories in this repository. Alternatively, you can also download it or compile it yourself from the sources here. This will typically result in pthreadGC2.(dll/lib).
If you are using Visual Studio, you need to provide a pthreads library compiled for Visual Studio either by downloading it or compiling it yourself. See here. Be careful to download / compile the right version for your setup. This would typically be pthreadVC2.(dll/lib).
One way to configure CMake is to use the CMake GUI. The GUI will list the variables that can be provided to configure the build. The variables can also be specified in the command-line interface (See below for an example)
In this step you can select the features to be included with PD_EXTRA
, PD_LOCALE
, PD_MULTI
and PD_UTILS
as described above. You can also enable building the C sample programs using PD_BUILD_C_EXAMPLES
.
When using Microsoft Visual Studio (MSVC), you will be requested to provide a path to the pthreads library and its headers using variables CMAKE_THREAD_LIBS_INIT
and PTHREADS_INCLUDE_DIR
.
On macOS, you can define different deployment target and architectures from your current system using the variables CMAKE_OSX_DEPLOYMENT_TARGET
and CMAKE_OSX_ARCHITECTURES
.
You can specify additional compilation flags using the variable CMAKE_C_FLAGS
.
CMake can now generate Makefiles, a MSVC solution, or an XCode project.
After generation, depending on your platform you can navigate to the directory where CMake generated the build files and then:
- On Linux: run
make
- On Windows: open the MSVC solution and build it
- On macOS: open the XCode project and build it
Of course you can also use CMake itself to build libpd by running this on the command line:
cd <path/to/build/files/generated/by/CMake>
cmake --build .
You can build Windows binaries in Linux using MinGW. This will only build the static and shared libraries, and not the C# bindings.
The easiest way is to set up a toolchain file for CMake pointing to all the necessary MinGW tools. For debian-based distributions, the file would contain the following:
SET(CMAKE_SYSTEM_NAME Windows)
SET(CMAKE_C_COMPILER x86_64-w64-mingw32-gcc)
SET(CMAKE_CXX_COMPILER x86_64-w64-mingw32-g++)
SET(CMAKE_RC_COMPILER x86_64-w64-mingw32-windres)
SET(CMAKE_FIND_ROOT_PATH /usr/x86_64-w64-mingw32)
SET(CMAKE_FIND_ROOT_PATH_MODE_PROGRAM NEVER)
SET(CMAKE_FIND_ROOT_PATH_MODE_LIBRARY ONLY)
SET(CMAKE_FIND_ROOT_PATH_MODE_INCLUDE ONLY)
Note paths and binary names might vary in other distributions.
You can then use the CMAKE_TOOLCHAIN_FILE
option to tell CMake to load the settings from your file. For example, if the above is saved to $HOME/.local/share/cmake/toolchain-mingw-x86_64-w64.cmake
, then CMake can be invoked with
cmake -DCMAKE_TOOLCHAIN_FILE=$HOME/.local/share/cmake/toolchain-mingw-x86_64-w64.cmake
followed by any other CMake options you'd normally use, and the source directory.
Here are examples of how to download, configure and build the latest libpd on the command line using CMake.
Linux:
git clone https://github.com/libpd/libpd
cd libpd
git submodule init
git submodule update
mkdir build && cd build
cmake .. -DPD_MULTI:BOOL=ON -DPD_BUILD_C_EXAMPLES:BOOL=ON
cmake --build .
Windows / MSVC:
git clone https://github.com/libpd/libpd
cd libpd
git submodule init
git submodule update
mkdir build && cd build
cmake .. -DCMAKE_THREAD_LIBS_INIT:PATH=</path/to/pthreadsVC2.lib> -DPTHREADS_INCLUDE_DIR:PATH=</path/to/pthread/header/files>
cmake --build .
Currently the CMake script is not capable of building the C# or the Java bindings. Please use the makefile for that.
Historically, Pd was designed to be built using the open source gcc & make and did not directly support being built in Visual Studio on Windows, mainly due to differences in C compiler versions.
Recently, the code has been adapted and a CMake build script has been developed that should allow you to generate a MSVC solution. The only important thing you need to be careful about is providing a pthreads library compiled for Visual Studio. See the section above about building with CMake.
Another possible approach is building the libpd C library using gcc and make using MinGW in msys on Windows. You can use the resulting .dll, .def, & .lib files with Visual Studio and the cpp wrapper is provided as an all header library so it should work directly within VS as well.
After building libpd in msys, you can "install" it to a temp directory to get only the libs and headers you need:
make install prefix=libpd-build
Pd expects numbers to be in an English format, ie. "0.3". If you are using a non-English language or locale setting on your system, it may be encoding numbers differently, ie. "0,3". This can lead to weird bugs in loaded patches where numbers seem wrong or end up truncated as 0.
By default, libpd is built with the LC_NUMERIC locale set to the "C" default, so this shouldn't be a problem. If you are using libpd within a project that requires specific locale settings, you will need to make sure libpd's LC_NUMERIC is left alone or at least reset it to "C" if working with a different numeric setting. If a non-English LC_NUMERIC is set, you will run into the number parsing issues mentioned above.
If you need to control LC_NUMERIC manually, you can build libpd without the call to setlocale() in libpd_init using the SETLOCALE=false makefile option or by setting the LIBPD_NO_NUMERIC define.
See #130 for more info.