Are there specific steps I can take to build the Xuggle Xuggler source code from Windows 32-bit, Windows 64-bit, Linux 32-bit, and Linux 64-bit? I've tried multiple times on multiple systems and keep getting lots of different errors.
Update
I spent several days trying to get Xuggle Xuggler to compile (and
cross-compile). I successfully tackled compiling both the original GPL
version of the code and an LGPL version. I thought I'd post an
answer to my own question on Stack Overflow to share my knowledge.
Update on Raspberry Pi
I was also able to build and run Xuggler on the Raspberry Pi following these same basic instructions below. I just used
my LGPL version of the code that I maintain on Github, and made modifications for the Pi. I can use the compiled JAR file and binaries on my Radxa Rock (another ARM device) too. If you're interested
in building on the Pi, you can use my pi branch:
https://github.com/e-d/xuggle-xuggler
If you are lazy and just want the precompiled .jar files for the Pi/ARM:
GPL Version (supports H.264)
LGPL Version (no H.264 support)
Here is a formatted version of my answer in a published Google Document.
For completeness (and in case the link goes dead one day), here is less-nicely-formatted text:
Building Xuggle Xuggler (GPL and LGPL Licensed Versions)
[Linux 32-bit, Linux 64-bit, Windows 32-bit, Windows 64-bit]
To build the Xuggle Xuggler library, you will need two Linux virtual machines running Ubuntu 11.10 (32-bit and 64-bit operating systems). A 32-bit version of the OS is required to build Linux 32-bit, cross compiling Windows 32-bit, and cross compiling Windows 64-bit binaries. A 64-bit version of the OS is required to build Linux 64-bit binaries.
Using VirtualBox, I created the two virtual machines discussed above with the ubuntu-11.10-server-i386.iso and ubuntu-11.10-server-amd64.iso disk images. These are headless server versions of Ubuntu. After installation of the OS, follow these steps to build Xuggler (you are welcome to try different dependency versions and not use the root user, but this is what I did to build successfully):
Change to root user:
sudo su
Just use root’s home directory:
cd /root
Update apt-get to use specific repository:
apt-get install python-software-properties
add-apt-repository ppa:ferramroberto/java
apt-get update
Install Java:
apt-get install sun-java6-jdk sun-java6-plugin
Verify the HotSpot Java 6 JVM is the default java:
java -version
If the incorrect version of Java appears, configure the default by running:
update-alternatives --config java
Install gcc, g++, make and all the other build essentials:
apt-get install build-essential
Install YASM:
apt-get install yasm
Install Open SSL:
apt-get install openssl
Install Package Config:
apt-get install pkg-config
Install Git:
apt-get install git
Install Ant:
apt-get install ant-optional
Install JUnit:
apt-get install junit
Install MingGW to be able to build for Windows (mingw-w64 can do 32 and 64-bit Windows):
apt-get install mingw-w64
Download the LGPL configured Xuggle source code (Ed’s fork of the code from Jeff Wallace’s fork from the original GPL xuggle code) or the original GPL version:
LGPL: git clone https://github.com/e-d/xuggle-xuggler.git
GPL: git clone https://github.com/xuggle/xuggle-xuggler.git
Compile and build the JAR files (with binaries inside). Be sure to run the 64-bit Linux build on the 64-bit version of Ubuntu. Also note that between builds you will need to run “ant clobber” to remove all of the compiled files from the previous architecture. To build run:
(32/64-bit Linux): ant stage
(64-bit Windows): ant -Dbuild.configure.os=x86_64-w64-mingw32 stage
(32-bit Windows): ant -Dbuild.configure.os=i686-w64-mingw32 stage
The JAR files will be in the /dist/lib directory.
If you need the Linux binaries to additionally work on CentOS, you’ll now need to change the version of GCC and G++ to use 3.4 instead of 3.6.
Install GCC 4.4:
apt-get install gcc-4.4
Update symbolic links to use 4.4 (the arch-specific link will be different on 32-bit VM):
rm /usr/bin/gcc
ln -s /usr/bin/gcc-4.4 /usr/bin/gcc
rm /usr/bin/x86_64-linux-gnu-gcc
ln -s /usr/bin/x86_64-linux-gnu-gcc-4.4 /usr/bin/x86_64-linux-gnu-gcc
Install C++ (G++) 4.4:
apt-get install c++-4.4
Update symbolic links to use 4.4 (the arch-specific link will be different on 32-bit VM):
rm /usr/bin/cpp
ln -s /usr/bin/cpp-4.4 /usr/bin/cpp
rm /usr/bin/x86_64-linux-gnu-cpp
ln -s /usr/bin/x86_64-linux-gnu-cpp-4.4 /usr/bin/x86_64-linux-gnu-cpp
rm /usr/bin/g++
ln -s /usr/bin/g++-4.4 /usr/bin/g++
rm /usr/bin/x86_64-linux-gnu-g++
ln -s /usr/bin/x86_64-linux-gnu-g++-4.4 /usr/bin/x86_64-linux-gnu-g++
Verify default versions:
gcc --version
c++ --version
cpp --version
gcc --version
You can now run the builds the same way as before (you only need to re-build Linux binaries). The binaries will now be compatible with slightly older versions of many Linux distros (including CentOS compatibility). These 4.4 compiled binaries should still work everywhere the 4.6 compiled versions would run too.
Special thanks to this blog for pointing me in the right direction and giving me the majority of what I detailed above.
Related
When I install llvm using command
bash -c "$(wget -O - https://apt.llvm.org/llvm.sh)"
The binaries that get installed looks like following
All the binaries have version number. For e.g. llvm-objdump-15.
But the tools I use are looking for llvm-objdump, while installer installed it as llvm-objdump-15. I can create a symbolic link with name llvm-obdump, but I want to know how to install it correctly and not have version numbers in all these binaries?
I fixed it by using apt-get install
sudo apt-get install clang-format clang-tidy clang-tools clang clangd libc++-dev libc++1 libc++abi-dev libc++abi1 libclang-dev libclang1 liblldb-dev libllvm-ocaml-dev libomp-dev libomp5 lld lldb llvm-dev llvm-runtime llvm python3-clang
It did install an older version, but it was fine for my needs
The names without version numbers are controlled by the llvm-defaults package on your distribution. It picks a specific version to make the default, and only that one has un-versioned symlinks installed into the system PATH.
As a consequence, on Debian based systems only one version (controlled by the distro) is going to be available there and it may not be the one from https://apt.llvm.org/. On these systems, the recommended way to use a specific version is to add the suffix.
If you can't do that, you should install the distro-provided version using the normal process rather than the versions on https://apt.llvm.org/.
To read more details about how all of this works, you can check out the documentation for the llvm-defaults package set here: https://salsa.debian.org/pkg-llvm-team/llvm-defaults/-/blob/experimental/debian/README.Debian
How can one build v8 from source on most recent Centos 7?
I tried, but ninja build always fails right away with "centos /lib64/libc.so.6: version `GLIBC_2.18' not found" message.
Plus, dependency installer script tells that Centos platform is not supported.
So, is there a way?
Thanks!
That error is due a not compatible C compiler, try compiling a newer GCC. On CentOS 7:
sudo yum install bzip2
cd /usr/local/src
wget https://www.mirrorservice.org/sites/sourceware.org/pub/gcc/releases/gcc-8.3.0/gcc-8.3.0.tar.gz
tar zxf gcc-8.3.0.tar.gz
cd gcc-8.3.0/
./contrib/download_prerequisites
./configure --disable-multilib --enable-languages=c,c++
make
sudo make install
export LD_LIBRARY_PATH=/usr/local/lib:/usr/lib:/usr/local/lib64:/usr/lib64
echo "/usr/local/lib64" > /etc/ld.so.conf.d/gcc-8.3.0.x86_64.conf
ldconfig
(From their wiki)
You can try using docker to build V8.
See https://github.com/gengjiawen/v8-build.
How to install gcc version 4.8 on centos or scientific linux operating systems which require yum for installing.
I tried to download gcc from https://ftp.gnu.org/gnu/gcc/gcc-4.8.2/ and then ran ./configure and then make. After running make it gives me the error: configure: error: cannot compute suffix of object files: cannot compile
See `config.log' for more details.
Is there some way by which I may install gcc version 4.8 on linux using yum install
I tried:
1). I tried with yum --enablerepo=testing-2-devtools-6 install devtoolset-2-gcc devtoolset-2-gcc-c++ it gives me repository not found
2). yum group install "Development Tools". It gives me Package gcc-4.4.7-11.el6.x86_64 already installed and latest version
3). building it from scratch, you'll have to do ./contrib/download_prerequisites first to get MPFR, GMP and MPC in the GCC source tree, then make a separate directory and run /path/to/gcc/source/configure.
It gives:
configure: error: building out of tree but /home/Softwares/gcc-4.8.2
contains host-x86_64-unknown-linux-gnu
https://access.redhat.com/documentation/en-US/Red_Hat_Developer_Toolset/3/html/3.0_Release_Notes/DTS3.0_Release.html#Features
Install the Red Hat Developer Toolset 3.0 (or 2.X) in a way similar to what is described here. (Basically you use a repo someone else built for CentOS).
Google says to try
yum group install "Development Tools"
check out: http://www.cyberciti.biz/faq/centos-rhel-7-redhat-linux-install-gcc-compiler-development-tools/
Using
gcc -m32 myprog.c
should compile in 32 bit version the file myprog.c.
Unfortunately I get this error:
In file included from /usr/include/stdio.h:27:0,
from myprog.c:1:
/usr/include/features.h:374:25: fatal error: sys/cdefs.h: No such file or directory
#include <sys/cdefs.h>
How could I fix this?
If I don't use -m32 it works but uses 64 bit instructions.
To compile 32 bit binaries on 64 bit Linux version, you have to Install libx32gcc development package and 32 bit GNU C Library
try this
sudo apt-get install libx32gcc-4.8-dev
and
sudo apt-get install libc6-dev-i386
You need gcc multilib support. Install the package gcc-multilib. E.g.:
sudo apt-get install gcc-multilib
I encountered to this same problem. when I was trying to build 32 bit so file while i'm in 64 bit ubuntu version.
you can build by switching your lib version into 64 bit
in netbeans
(-> project properties -> C compiler -> additional options) set to
-shared -m64
To fix the issue we first need to know which package will provide the file sys/cdefs.h
In CentOs/RHEL based system you can find it by running below command
yum provides '*sys/cdefs.h'
Similarly, in Debian/Ubuntu systems you can find it by running
apt-file search 'sys/cdefs.h'
Note: If apt-file package is not already present on the system install it with command
apt-get install apt-file
Both above commands will scan the respective package management systems database i.e RPM/Apt and tells the name of the package which supplies the file ending with "sys/cdefs.h".
Install the appropriate package from the packages listed by the commands.
Like
On CentOs/RHEL:
yum install glibc-headers
On Debian/Ubuntu
apt-get install libc6-dev
This approach is not only useful for this particular issue but any similar issue which is reporting some required file being provided by some package being not present.
I have written some effects in C++ (g++) using freeglut on Linux, and I compile them with
g++ -Wall -lglut part8.cpp -o part8
So I was wondering if it is possible to have g++ make static compiled Windows executables that contains everything needed?
I don't have Windows, so it would be really cool, if I could do that on Linux :)
mingw32 exists as a package for Linux. You can cross-compile and -link Windows applications with it. There's a tutorial here at the Code::Blocks forum. Mind that the command changes to x86_64-w64-mingw32-gcc-win32, for example.
Ubuntu, for example, has MinGW in its repositories:
$ apt-cache search mingw
[...]
g++-mingw-w64 - GNU C++ compiler for MinGW-w64
gcc-mingw-w64 - GNU C compiler for MinGW-w64
mingw-w64 - Development environment targeting 32- and 64-bit Windows
[...]
Suggested method gave me error on Ubuntu 16.04: E: Unable to locate package mingw32
===========================================================================
To install this package on Ubuntu please use following:
sudo apt-get install mingw-w64
After install you can use it:
x86_64-w64-mingw32-g++
Please note!
For 64-bit use: x86_64-w64-mingw32-g++
For 32-bit use: i686-w64-mingw32-g++
One option of compiling for Windows in Linux is via mingw. I found a very helpful tutorial here.
To install mingw32 on Debian based systems, run the following command:
sudo apt-get install mingw32
To compile your code, you can use something like:
i586-mingw32msvc-g++ -o myApp.exe myApp.cpp
You'll sometimes want to test the new Windows application directly in Linux. You can use wine for that, although you should always keep in mind that wine could have bugs. This means that you might not be sure that a bug is in wine, your program, or both, so only use wine for general testing.
To install wine, run:
sudo apt-get install wine
Install a cross compiler, like mingw64 from your package manager.
Then compile in the following way: instead of simply calling gcc call i686-w64-mingw32-gcc for 32-bit Windows or x86_64-w64-mingw32-gcc" for 64-bit Windows. I would also use the --static option, as the target system may not have all the libraries.
If you want to compile other language, like Fortran, replace -gcc with -gfortran in the previous commands.
I've used mingw on Linux to make Windows executables in C, I suspect C++ would work as well.
I have a project, ELLCC, that packages clang and other things as a cross compiler tool chain. I use it to compile clang (C++), binutils, and GDB for Windows. Follow the download link at ellcc.org for pre-compiled binaries for several Linux hosts.
From: https://fedoraproject.org/wiki/MinGW/Tutorial
As of Fedora 17 it is possible to easily build (cross-compile) binaries for the win32 and win64 targets. This is realized using the mingw-w64 toolchain: http://mingw-w64.sf.net/. Using this toolchain allows you to build binaries for the following programming languages: C, C++, Objective-C, Objective-C++ and Fortran.
"Tips and tricks for using the Windows cross-compiler": https://fedoraproject.org/wiki/MinGW/Tips
For Fedora:
# Fedora 18 or greater
sudo dnf group install "MinGW cross-compiler"
# Or (not recommended, because of its deprecation)
sudo yum groupinstall -y "MinGW cross-compiler"