Building Julia (Detailed))
Be sure to also configure your system to use the appropriate proxy settings, e.g. by setting the https_proxy
and http_proxy
variables.
When compiled the first time, the build will automatically download pre-built external dependencies. If you prefer to build all the dependencies on your own, or are building on a system that cannot access the network during the build process, add the following in Make.user
:
USE_BINARYBUILDER=0
Building Julia requires 5GiB if building all dependencies and approximately 4GiB of virtual memory.
To perform a parallel build, use make -j N
and supply the maximum number of concurrent processes. If the defaults in the build do not work for you, and you need to set specific make parameters, you can save them in Make.user
, and place the file in the root of your Julia source. The build will automatically check for the existence of Make.user
and use it if it exists.
You can create out-of-tree builds of Julia by specifying make O=<build-directory> configure
on the command line. This will create a directory mirror, with all of the necessary Makefiles to build Julia, in the specified directory. These builds will share the source files in Julia and deps/srccache
. Each out-of-tree build directory can have its own Make.user
file to override the global Make.user
file in the top-level folder.
If everything works correctly, you will see a Julia banner and an interactive prompt into which you can enter expressions for evaluation. (Errors related to libraries might be caused by old, incompatible libraries sitting around in your PATH. In this case, try moving the julia
directory earlier in the PATH). Note that most of the instructions above apply to unix systems.
To run julia from anywhere you can:
add an alias (in
bash
:echo "alias julia='/path/to/install/folder/bin/julia'" >> ~/.bashrc && source ~/.bashrc
), oradd a soft link to the
julia
executable in thejulia
directory to/usr/local/bin
(or any suitable directory already in your path), oradd the
julia
directory to your executable path for this shell session (inbash
:export PATH="$(pwd):$PATH"
; incsh
ortcsh
:
set path= ( $path $cwd )
), or
add the
julia
directory to your executable path permanently (e.g. in.bash_profile
), orwrite
prefix=/path/to/install/folder
into and then runmake install
. If there is a version of Julia already installed in this folder, you should delete it before runningmake install
.
Now you should be able to run Julia like this:
If you are building a Julia package for distribution on Linux, macOS, or Windows, take a look at the detailed notes in .
cd julia
git pull && make
Assuming that you had made no changes to the source tree that will conflict with upstream updates, these commands will trigger a build to update to the latest version.
Over time, the base library may accumulate enough changes such that the bootstrapping process in building the system image will fail. If this happens, the build may fail with an error like
As described, running
make clean && make
is usually sufficient. Occasionally, the stronger cleanup done bymake cleanall
is needed.New versions of external dependencies may be introduced which may occasionally cause conflicts with existing builds of older versions.
a. Special
make
targets exist to help wipe the existing build of a dependency. For example,make -C deps clean-llvm
will clean out the existing build ofllvm
so thatllvm
will be rebuilt from the downloaded source distribution the next timemake
is called.make -C deps distclean-llvm
is a stronger wipe which will also delete the downloaded source distribution, ensuring that a fresh copy of the source distribution will be downloaded and that any new patches will be applied the next timemake
is called.b. To delete existing binaries of
julia
and all its dependencies, delete the./usr
directory in the source tree.If you’ve moved the source directory, you might get errors such as
CMake Error: The current CMakeCache.txt directory ... is different than the directory ... where CMakeCache.txt was created.
, in which case you may delete the offending dependency underdeps
In extreme cases, you may wish to reset the source tree to a pristine state. The following git commands may be helpful:
git reset --hard #Forcibly remove any changes to any files under version control
git clean -x -f -d #Forcibly remove any file or directory not under version control
To avoid losing work, make sure you know what these commands do before you run them.
git
will not be able to undo these changes!
Platform-Specific Notes
Notes for various operating systems:
Notes for various architectures:
Building Julia requires that the following software be installed:
- [GNU make] — building dependencies.
- [gcc & g++][gcc] (>= 5.1) or [Clang][clang] (>= 3.5, >= 6.0 for Apple Clang) — compiling and linking C, C++.
- [libatomic][gcc] — provided by [gcc] and needed to support atomic operations.
- [python] (>=2.7) — needed to build LLVM.
- [gfortran] — compiling and linking Fortran libraries.
- [perl] — preprocessing of header files of libraries.
- [wget], [curl], or [fetch] (FreeBSD) — to automatically download external libraries.
- [m4] — needed to build GMP.
- [awk] — helper tool for Makefiles.
- [patch] — for modifying source code.
- [cmake] (>= 3.4.3) — needed to build
libgit2
. - [pkg-config] — needed to build
libgit2
correctly, especially for proxy support. - [powershell] (>= 3.0) — necessary only on Windows.
- [which] — needed for checking build dependencies.
On Debian-based distributions (e.g. Ubuntu), you can easily install them with :
Julia uses the following external libraries, which are automatically downloaded (or in a few cases, included in the Julia source repository) and then compiled from source the first time you run make
. The specific version numbers of these libraries that Julia uses are listed in deps/Versions.make:
- [FemtoLisp] — packaged with Julia source, and used to implement the compiler front-end.
- [libuv] (custom fork) — portable, high-performance event-based I/O library.
- [OpenLibm] — portable libm library containing elementary math functions.
- [DSFMT] — fast Mersenne Twister pseudorandom number generator library.
- [OpenBLAS] — fast, open, and maintained [basic linear algebra subprograms (BLAS)]
- [LAPACK] — library of linear algebra routines for solving systems of simultaneous linear equations, least-squares solutions of linear systems of equations, eigenvalue problems, and singular value problems.
- [MKL] (optional) – OpenBLAS and LAPACK may be replaced by Intel’s MKL library.
- [SuiteSparse] — library of linear algebra routines for sparse matrices.
- [PCRE] — Perl-compatible regular expressions library.
- [GMP] — GNU multiple precision arithmetic library, needed for
BigInt
support. - [MPFR] — GNU multiple precision floating point library, needed for arbitrary precision floating point (
BigFloat
) support. - [libgit2] — Git linkable library, used by Julia’s package manager.
- [curl] — libcurl provides download and proxy support.
- [libssh2] — library for SSH transport, used by libgit2 for packages with SSH remotes.
- [mbedtls] — library used for cryptography and transport layer security, used by libssh2
- [utf8proc] — a library for processing UTF-8 encoded Unicode strings.
- [LLVM libunwind] — LLVM’s fork of [libunwind], a library that determines the call-chain of a program.
If you already have one or more of these packages installed on your system, you can prevent Julia from compiling duplicates of these libraries by passing USE_SYSTEM_...=1
to make
or adding the line to Make.user
. The complete list of possible flags can be found in Make.inc
.
Please be aware that this procedure is not officially supported, as it introduces additional variability into the installation and versioning of the dependencies, and is recommended only for system package maintainers. Unexpected compile errors may result, as the build system will do no further checking to ensure the proper packages are installed.
The most complicated dependency is LLVM, for which we require additional patches from upstream (LLVM is not backward compatible).
For packaging Julia with LLVM, we recommend either:
- bundling a Julia-only LLVM library inside the Julia package, or
- adding the patches to the LLVM package of the distribution.
- A complete list of patches is available in
deps/llvm.mk
, and the patches themselves are indeps/patches/
. - The only Julia-specific patch is the lib renaming (
llvm-symver-jlprefix.patch
), which should not be applied to a system LLVM. - The remaining patches are all upstream bug fixes, and have been contributed into upstream LLVM.
- A complete list of patches is available in
Using an unpatched or different version of LLVM will result in errors and/or poor performance. Though Julia can be built with newer LLVM versions, support for this should be regarded as experimental and not suitable for packaging.
Julia uses a custom fork of libuv. It is a small dependency, and can be safely bundled in the same package as Julia, and will not conflict with the system library. Julia builds should not try to use the system libuv.
As a high-performance numerical language, Julia should be linked to a multi-threaded BLAS and LAPACK, such as OpenBLAS or ATLAS, which will provide much better performance than the reference libblas
implementations which may be default on some systems.
Source distributions of releases
Each pre-release and release of Julia has a “full” source distribution and a “light” source distribution.
The full source distribution contains the source code for Julia and all dependencies so that it can be built from source without an internet connection. The light source distribution does not include the source code of dependencies.
For example, julia-1.0.0.tar.gz
is the light source distribution for the v1.0.0
release of Julia, while julia-1.0.0-full.tar.gz
is the full source distribution.
Building Julia from source with a Git checkout of a stdlib
If you need to build Julia from source with a Git checkout of a stdlib, then use make DEPS_GIT=NAME_OF_STDLIB
when building Julia.
For example, if you need to build Julia from source with a Git checkout of Pkg, then use make DEPS_GIT=Pkg
when building Julia. The Pkg
repo is in stdlib/Pkg
, and created initially with a detached HEAD
. If you’re doing this from a pre-existing Julia repository, you may need to make clean
beforehand.
If you need to build Julia from source with Git checkouts of more than one stdlib, then DEPS_GIT
should be a space-separated list of the stdlib names. For example, if you need to build Julia from source with a Git checkout of Pkg, Tar, and Downloads, then use make DEPS_GIT='Pkg Tar Downloads'
when building Julia.
An “assert build” of Julia is a build that was built with both FORCE_ASSERTIONS=1
and LLVM_ASSERTIONS=1
. To build an assert build, define both of the following variables in your Make.user
file:
FORCE_ASSERTIONS=1
Please note that assert builds of Julia will be slower than regular (non-assert) builds.