I installed LLVM from source (using CMake and make install) but I am unable to uninstall it because there is no make uninstall avalaible.
This is LLVM version 3.5.2.
I use ArchLinux.
How can I uninstall LLVM in an automated way?
From the CMake FAQ / Can I do "make uninstall" with CMake :
...Unix users could enter this command in the shell:
xargs rm < install_manifest.txt
(after you cd to the LLVM build directory).
Related
I have a project writen with C99. It links to libcurl, generate makefile with cmake. Now, i want to explore the project to window env. I tried cygwin first, everything works except that final user need to install cygwin too.
After some search, mingw64 may archive my goal (build and use directly in window), but i could not find a full use guild for how to build linux project with mingw64 in ubuntu(linux).
Thanks for the help.
After some research, a solutions works:
install window cmake
install mingw-w64
mkdir build && cd build && cmake -G "MinGW Makefiles" ..
mingw32-make
MinGW may not support all API, some compatible job should be done.
I'm a little familiar with command line but I'm still having problems running GNU Typist/gtypist. http://www.gnu.org/software/gtypist/
Can anyone give me really basic step-by-step instructions for a Mac?
Install using Homebrew
brew install gnu-typist
gtypist
You can also install it directly from the sources (note that for building from the sources you have to install XCode - it's free and you can download it from App Store):
Get the source of gnu typist
http://ftp.gnu.org/gnu/gtypist/gtypist-2.9.tar.gz
Get the source of ncurses
I am downloading it from GNU pages here:
ftp://ftp.gnu.org/gnu/ncurses//ncurses-6.0.tar.gz.sig
But you can get most recent version directly from maintainer's page:
ftp://ftp.invisible-island.net/ncurses/ncurses.tar.gz
Create place where you want to have all the stuff
mkdir ~/opt/usr
Build ncurses - pay attention to --enable-widec
./configure --enable-widec --with-shared --prefix=$HOME/opt/usr/
make; make install
Build gnutypist
export LDFLAGS="-L$HOME/opt/usr/lib"
export CFLAGS="-I$HOME/opt/usr/include"
./configure --prefix=$HOME/opt/usr/
make; make install
Run gnutypist
cd $HOME/opt/usr/bin
./gtypist
Alternatively to brew you can use MacPorts (macos package manager)
Installation command:
sudo port install gtypist
Link to the port summary page:
https://ports.macports.org/port/gtypist/summary
Am looking for installing protobuf 2.5.0 on Arch Linux, so that protoc-2.5.0.so is installed on the OS, so that I can go ahead with building hadoop 2.6.0 from source and make my life easy! :)
BTW, protobuf 2.6.0 does not compile when hadoop is built from source I have tried that as well. Ubuntu 14.04 comes with protoc 2.5.0. I DO NOT want to use Ubuntu.
Please check the screenshot first (there is no protobuf 2.5.0), since the problem lies there.. I guess
am getting the following exception, I am aware that protoc is not installed in arch linux currently.
[ERROR] Failed to execute goal
org.apache.hadoop:hadoop-maven-plugins:2.6.0:protoc (compile-protoc)
on project hadoop-common:
org.apache.maven.plugin.MojoExecutionException: 'protoc --version' did
not return a version -> [Help 1]
Please help me out, since, I have spent 4 hours every day from two days, with no luck.
Compiling Google's protobuf is pretty easy.
I originally found out how to do it on this blog post while compiling hadoop myself.
But here is my version:
$ cd /usr/local/src/
$ wget https://github.com/google/protobuf/releases/download/v2.5.0/protobuf-2.5.0.tar.gz
$ tar xvf protobuf-2.5.0.tar.gz
$ cd protobuf-2.5.0
$ ./autogen.sh
$ ./configure --prefix=/usr
$ make
$ make install
$ protoc --version
Install protobuf for java
$ cd java
$ mvn install
$ mvn package
You should be good to go.
To enable you to install different versions of protobuf, install stow
then change ./configure --prefix=/usr to ./configure --prefix=/usr/local/stow/protobuf-2.5.0
Then link protobuf into your system with stow:
$ cd /usr/local/stow
$ stow protobuf-2.5.0
Note: stow uses /usr/local/bin by default. Make sure thats in your $PATH
To unlink that version of protobuf,
$ stow -D protobuf-2.5.0
Hope this helped.
I wonder why the above answer got downvoted,even I had to perform few more steps (in addition to the accepted answer by Rudker) to get protobuf 2.5 installed on Ubuntu Xenial.
Leaving the additional steps here for everyone's benefit:
apt-get install autoconf in response to the error : ./autogen.sh: autoreconf: not found for command : ./autogen.sh
apt-get install libtool in response to the error : autoreconf: libtoolize is needed because this package uses Libtool for command : ./autogen.sh
apt install g++ in response to the error : configure: error: C++ preprocessor "/lib/cpp" fails sanity check for command : ./configure --prefix=/usr
An easier but not future proof solution (for future queries) would be to head over to Arch Linux Archives: http://seblu.net/a/archive/packages/p/protobuf/
Uninstall the newer version of protobuf and install the downloaded package via pacman -U protobuf-2.5.x..
Though whenever you upgrade the Arch Linux packages via pacman you'd need to ensure you are doing: sudo pacman -Syu --ignore protobuf
I currently don't have enough reputation to comment, so I add a answer here to update the top voted answer.
Since protobuf move to different repo, the new wget command should be:
wget https://github.com/protocolbuffers/protobuf/releases/download/v2.5.0/protobuf-2.5.0.tar.gz
And in order to run ./autogen.sh, you may need install these:
sudo yum install libtool automake autoconf
For OSX prerequisites, try SunitaKoppar's answer (I don't know why the down-votes).
Thanks for the steps. Just wanted to add that, to get autogen.sh to work, I had to install the below packages (commands for mac below):
brew install gtk
brew install autoconf
brew install automake
I've downloaded and installed libpng from http://www.libpng.org/pub/png/libpng.html, which is version 1.6.16, following their simple instructions:
./configure
./make check
sudo ./make install
It runs with everything successful, and says it's been installed. However, after doing this I still read:
$ libpng-config --cflags --ldflags
-I~/anaconda/include/libpng15
-L~/anaconda/lib -lpng15
This is breaking other builds and installs, which sometimes detect 15 while requiring 16. How can I fix this?
Fedora Linux 20 x 64
If you want libpng16 to go in your $HOME/ then, in your libpng directory, run
./configure --prefix=$HOME
./make
./make install
This will install png*.h in $HOME/include/, libpng-config in $HOME/bin/,
and the library itself in $HOME/lib/.
If you want to put it in some other directory where you don't have write permission, the final command is
sudo ./make install
I am trying to install the gnu arm toolchain for ubuntu. I first downloaded the tar from CodeSourcery. However when I go into the bin folder, I cannot run any of the binaries. I have tried with ./ and without and putting it in the PATH and it keeps telling me "Command not Found" yet the file is there in the folder right in front of me. Then I tried sudo apt-get install gcc-arm-linux-gnueabi except after it says it has installed successfully, I cannot find it with whereis gcc-arm-linux-gnueabi. Can anyone help?
fixed, using:
sudo apt-get install gcc-arm*
Are you compiling on a 64-bit OS? Try:
sudo apt-get install ia32-libs
I had the same problem when trying to compile the Raspberry Pi kernel. I was cross-compiling on Ubuntu 12.04 64-bit and the toolchain requires ia32-libs to work on on a 64-bit system.
See http://hertaville.com/2012/09/28/development-environment-raspberry-pi-cross-compiler/
CodeSourcery convention is to use prefix arm-none-linux-gnueabi- for all executables, not gcc-arm-linux-gnueabi that you mention. So, standard name for CodeSourcery gcc would be arm-none-linux-gnueabi-gcc.
After you have installed CodeSourcery G++, you need to add CodeSourcery directory into your PATH.
Typically, I prefer to install CodeSourcery into directory like /opt/arm-2010q1 or something like that. If you don't know where you have installed it, you can find it using locate arm-none-linux-gnueabi-gcc, however you may need to force to update your locate db using sudo updatedb before locate will work properly.
After you have identified where your CodeSourcery is installed, add it your PATH by editing ~/.bashrc like this:
PATH=/opt/arm-2010q1/bin:$PATH
Also, it is customary and very convenient to define
CROSS_COMPILE=arm-none-linux-gnueabi-
in your .bashrc, because with CROSS_COMPILE defined, most tools will automatically use proper compiler for ARM compilation without you doing anything.
if you are on 64 bit os then you need to install this additional libraries.
sudo apt-get install lib32z1 lib32ncurses5 lib32bz2-1.0
got the same error when trying to cross compile the raspberry pi kernel on ubunto 14.04.03 64bit under VM. the solution was found here:
-Install packages used for cross compiling on the Ubuntu box.
sudo apt-get install gcc-arm-linux-gnueabi make git-core ncurses-dev
-Download the toolchain
cd ~
git clone https://github.com/raspberrypi/tools
-Add the toolchain to your path
PATH=$PATH:~/tools/arm-bcm2708/gcc-linaro-arm-linux-gnueabihf-raspbian:~/tools/arm-bcm2708/gcc-linaro-arm-linux-gnueabihf-raspbian-x64/bin
notice the x64 version in the path command
I was also facing the same issue and resolved it after installing the following dependency:
sudo apt-get install lib32z1-dev
If you are on a 64bit build of ubuntu or debian (see e.g. 'cat /proc/version') you should simply use the 64bit cross compilers, if you cloned
git clone https://github.com/raspberrypi/tools
then the 64bit tools are in
tools/arm-bcm2708/gcc-linaro-arm-linux-gnueabihf-raspbian-x64
use that directory for the gcc-toolchain.
A useful tutorial for compiling that I followed is available here Building and compiling Raspberry PI Kernel (use the -x64 path from above as ${CCPREFIX})
You have installed a toolchain which was compiled for i686 on a box which is running an x86_64 userland.
Use an i686 VM.
Its a bit counter-intuitive. The toolchain is called gcc-arm-linux-gnueabi. To invoke the tools execute the following: arm-linux-gnueabi-xxx
where xxx is gcc or ar or ld, etc
try the following command:
which gcc-arm-linux-gnueabi
Its very likely the command is installed in /usr/bin.
I had to cross compile C code in Ubuntu for ARM. This worked for me:
$ sudo apt install gcc-arm-none-eabi
Later, tested it on the qemu emulator
#Install qemu
sudo apt-get install qemu qemu-user-static qemu-system-arm
#Cross compile "helloworld.c"
$ arm-none-eabi-gcc --specs=rdimon.specs -Wl,--start-group -lgcc -lc -lm -lrdimon -Wl,--end-group helloworld.c -o helloworld
#Run
qemu-arm-static helloworld