setup: i3wm, fedora 31
I am facing this issue, in which some rpm's even after being installed,
For e.g. I have Bluejeans installed, but when I try running the app, it results in command not found, the baffling part being I have no such issue opening it in gnome.
$ rpm -qa | grep -i bluejeans
bluejeans-v2-2.0.0-188.x86_64
$ bluejeans
zsh: bluejeans: command not found...
same goes when I try accessing using dmenu.
solved the issue, bluejeans was not getting installed in /usr/bin or any other similar paths.
rpm -ql bluejeans
package bluejeans is not installed
while they also named package as bluejeans-v2 which made it harder to search, while in gnome I could start using bluejeans even from the terminal.
solution:
create a symlink to /usr/bin where you have bluejean-v2 installed
ls -s /usr/bin/bluejean-v2 ${install_path}/bluejeans-v2
So, I want to use Mariadb. There is this Connector-C for it.
https://downloads.mariadb.org/connector-c/
How do I install it? Quiet frankly, the documentation for it is horrible. Even the src file for 3.0.5 is linked to 3.0.4 page.
I did not find a way to install the binary, and the documentation for building from src is quiet vague. I would prefer to know how to install both ways (binary and build from source)
I'm using CentOS7 64bit.
The easiest way to install it would be to use the MariaDB package repository.
curl -sS https://downloads.mariadb.com/MariaDB/mariadb_repo_setup | sudo bash
sudo yum -y install MariaDB-devel
As for building from source, these steps should work on CentOS 7.
sudo yum -y install git gcc openssl-devel make cmake
git clone https://github.com/MariaDB/mariadb-connector-c.git
mkdir build && cd build
cmake ../mariadb-connector-c/ -DCMAKE_INSTALL_PREFIX=/usr
make
sudo make install
And for Ubuntu 20.04...
sudo apt-get install libmariadb3 libmariadb-dev
Raspberry Pi OS
cd to preferred build location. Then install (thanks to #markusjm!):
sudo apt install git gcc make cmake libssl-dev
git clone https://github.com/MariaDB/mariadb-connector-c.git
mkdir build && cd build
cmake ../mariadb-connector-c/ -DCMAKE_INSTALL_PREFIX=/usr
make
sudo make install
Then add installation directory to LD_LIBRARY_PATH. Note: my installation directory is /usr/lib/mariadb. If you cannot find this after your installation, search for e.g. libmariadb.so, a file that should reside in your installation folder.
export LD_LIBRARY_PATH=/usr/lib/mariadb:$LD_LIBRARY_PATH
Afterwards you can finally pip3 install mariadb, if, like me, you tried to do that in the first place.
After you download MariaDB Connector/C, untar and cd. Then mv the executable first.
sudo mv -f bin/mariadb_config /usr/bin/
Now you can execute mariadb_config and will know where to put header and library files to build wheel for mariadb.
For example,
Ubuntu 18.04
sudo mv -f include/mariadb /usr/local/include/
sudo mv -f lib/mariadb /usr/local/lib/
CentOS 7 & Ubuntu 20.04
sudo mv -f include/mariadb /usr/include/
sudo mv -f lib/mariadb /usr/lib/
Finally, you could pip install mariadb. (Or, export CFLAGS=-std=c99 may help.)
After, in the case you cannot import mariadb,
export LD_LIBRARY_PATH=/PATH/TO/where/you/mv/lib/mariadb
I made the attempt at enabling EPEL on my Fedora 22 machine by running
wget http://dl.fedoraproject.org/pub/epel/7/x86_64/e/epel-release-7-5.noarch.rpm
Then
sudo rpm -ivh epel-release-7-5.noarch.rpm
And now when I run
dnf repolist
I get the following
[sinux1#horrible-host ~]$ sudo dnf repolist
Extra Packages for Enterprise Linux 7 - x86_64 1.8 MB/s | 9.0 MB 00:05
Failed to open: /var/cache/dnf/epel-2b6dfc5904c26562/repodata/49c4e3bd54f19136521be9b254830c954369cc22ce1b661db502ebef13b0004c-updateinfo.xml.bz2.
[sinux1#horrible-host ~]$
I'm stuck and not sure what my next step is. How can I reverse what I did, and what would have been the proper way of enabling EPEL?
Thanks for any input
This is what appears to have worked for me.
Remove the epel repo
rm -rf /etc/yum.repos.d/epel*
then try install epel via dnf
dnf clean all
dnf install epel-release
I did the same as Sinux Tine and ended up in some sort of deadlock.
First try to do like Rup says.
If everything fails, try the following as a last resort. It worked for me.
> su
> cd /var/cache/dnf/epel-2b6dfc5904c26562/repodata
> mkdir throwaway
> mv 49c4e3bd54f19136521be9b254830c954369cc22ce1b661db502ebef13b0004c-updateinfo.xml.bz2 throwaway
That effectively deletes the file which dnf says it cannot open.
When and if the problem is solved, you can delete the throwaway directory and its contents.
Current libsolv version does not have bz2 compression.
Solution: enable bunzip2 compression in libsolv, Basically 'yum update libsolv', then reinstall epel-release.
To complement the answer for Centos 7, I followed the next steps:
Download the libsolv-0.6.14-1.el7 wich was patched
Install it
sudo rpm -ivh libsolv-0.6.14-1.el7.centos.x86_64.rpm
If we get a conflict with the previous version
warning: libsolv-0.6.14-1.el7.centos.x86_64.rpm: Header V3 RSA/SHA1 Signature, key ID 1b6f9f55: NOKEY
Preparing... ################################# [100%]
file /usr/lib64/libsolv.so.0 from install of libsolv-0.6.14-1.el7.centos.x86_64 conflicts with file from package libsolv-0.6.11-1.el7.x86_64
file /usr/lib64/libsolvext.so.0 from install of libsolv-0.6.14-1.el7.centos.x86_64 conflicts with file from package libsolv-0.6.11-1.el7.x86_64
We proceed to remove the previous version
sudo yum erase libsolv-0.6.11-1.el7.x86_64
and install the patched libsolv package
sudo rpm -ivh libsolv-0.6.14-1.el7.centos.x86_64.rpm
We need to reinstall the dnf package manager, because it was removed after to remove the libsolv so we install with the following command.
sudo yum install dnf
We validate it using sudo dnf repolist now it, give us the right output.
Using metadata from Mon Jun 13 10:53:11 2016
repo id repo name status
base CentOS-7 - Base 9,007
For context, it on a remote server which has a firewall. I'm setting up my environment through a proxy. I have ruby 1.8.7. When I try to gem install..
sudo gem install --http-proxy <host address>:<port> json
I get the following error:
Building native extensions. This could take a while...
ERROR: Error installing json:
ERROR: Failed to build gem native extension.
/usr/bin/ruby extconf.rb
mkmf.rb can't find header files for ruby at /usr/lib/ruby/ruby.h
Gem files will remain installed in /usr/lib64/ruby/gems/1.8/gems/json-1.8.1 for inspection.
Results logged to /usr/lib64/ruby/gems/1.8/gems/json-1.8.1/ext/json/ext/generator/gem_make.out
Since I was unsure what the problem is, I googled and found these
gem install: Failed to build gem native extension (can't find header files) - the instructions here seem to be specific to the gem being installed.
How to install json gem - Failed to build gem native extension This seems to be slightly different error.
Any hints? Thanks!
Modern era update, as stated by mimoralea:
In case that you are using ruby 2.0 or 2.2 (thanks #patrick-davey).
sudo apt-get install ruby2.0-dev
sudo apt-get install ruby2.2-dev
sudo apt-get install ruby2.3-dev
or, generic way:
sudo apt-get install ruby-dev
or
sudo apt-get install ruby`ruby -e 'puts RUBY_VERSION[/\d+\.\d+/]'`-dev
The first link you’ve posted is exactly your case: there is no ruby development environment installed. Development env is needed to compile ruby extensions, which are mostly written in C. Proxy has nothing to do with the problem: everything is downloaded fine, just compilation fails.
I would suggest you to install ruby-dev (ruby-devel for rpm-based distros) package onto you target machine.
gcc package might be needed as well.
Try:
$ sudo apt-get install ruby-dev
Or, for Redhat distro:
$ sudo yum install ruby-devel
Or, for [open]SuSE:
$ sudo zypper install ruby-devel
For Xcode 11 on macOS 10.14, this can happen even after installing Xcode and installing command-line tools and accepting the license with
sudo xcode-select --install
sudo xcodebuild -license accept
The issue is that Xcode 11 ships the macOS 10.15 SDK which includes headers for ruby2.6, but not for macOS 10.14's ruby2.3. You can verify that this is your problem by running
ruby -rrbconfig -e 'puts RbConfig::CONFIG["rubyhdrdir"]'
which on macOS 10.14 with Xcode 11 prints the non-existent path
/Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX10.15.sdk/System/Library/Frameworks/Ruby.framework/Versions/2.3/usr/include/ruby-2.3.0
However, Xcode 11 installs a macOS 10.14 SDK within /Library/Developer/CommandLineTools/SDKs/MacOS10.14.sdk. It isn't necessary to pollute the system directories by installing the old header files as suggested in other answers. Instead, by selecting that SDK, the appropriate ruby2.3 headers will be found:
sudo xcode-select --switch /Library/Developer/CommandLineTools
ruby -rrbconfig -e 'puts RbConfig::CONFIG["rubyhdrdir"]'
This should now correctly print
/Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk/System/Library/Frameworks/Ruby.framework/Versions/2.3/usr/include/ruby-2.3.0
Likewise, gem install should work while that SDK is selected.
To switch back to the current Xcode SDK, use
sudo xcode-select --switch /Applications/Xcode.app
In case that you are using ruby 2.0 or 2.2 (thanks #patrick-davey) or 2.3 (thanks #juanitofatas).
sudo apt-get install ruby-dev
sudo apt-get install ruby2.0-dev
sudo apt-get install ruby2.2-dev
sudo apt-get install ruby2.3-dev
And you get the pattern here...
I also encountered this problem because I install Ruby on Ubuntu via brightbox, and I thought ruby-dev is the trunk of ruby. So I did not install. Install ruby2.3-dev fixes it:
sudo apt-get install ruby2.3-dev
For those who are getting this on Mac OS X you may need to run the following command to install the XCode command-line tools, even if you already have XCode installed:
sudo xcode-select --install
Also you must agree the terms and conditions of XCode by running the following command:
sudo xcodebuild -license
I had a similar problem using cygwin to run the following command:
$ gem install rerun
I solved it by installing the following cygwin packages:
ruby-devel
libffi-devel
gcc-core
gcc-g++
make
automake1.15
Most voted solution didn't work on my machine (linux mint 18.04).
After a careful look, i found that g++ was missing.
Solved with
sudo apt-get install g++
in case you use SUSE
sudo yast2 -i ruby-devel
Xcode 11 / macOS Catalina
On Xcode 11 / macOS Catalina, the header files are no longer in the old location and the old /Library/Developer/CommandLineTools/Packages/macOS_SDK_headers_for_macOS_10.14.pkg file is no longer available.
Instead, the headers are now installed to the /usr/include directory of the current SDK path:
/Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX.sdk/usr/include
Most of this directory can be found by using the output of xcrun --show-sdk-path. And if you add this path to the CPATH environment variable, then build scripts (including those called via bundle) will generally be able to find it.
I resolved this by setting my CPATH in my .zshrc file:
export CPATH="$(xcrun --show-sdk-path)/usr/include"
After opening a new shell (or running source .zshrc), I no longer receive the error message mkmf.rb can't find header files for ruby at /usr/lib/ruby/ruby.h and the rubygems install properly.
Note on Building to Non-macOS Platforms
If you are building to non-macOS platforms, such as iOS/tvOS/watchOS, this change will attempt to include the macOS SDK in those platforms, causing build errors. To resolve, either don't set CPATH environment variable on login, or temporarily set it to blank when running xcodebuild like so:
CPATH="" xcodebuild --some-args
In Fedora 21 and up, you simply open a terminal and install the Ruby Development files as root.
dnf install ruby-devel
On Mac 10.14, the header files don't seem to be installed in the correct place. Rather than changing paths like the other fixes, I was able to just run this:
open /Library/Developer/CommandLineTools/Packages/macOS_SDK_headers_for_macOS_10.14.pkg
Follow the instructions and it resolved this problem for me.
You may need to install gcc after install ruby-devel
Xcode -> Preferences -> Locations
change Command Line Tools to Xcode 11.2.1
You need to install the entire ruby and not just the minimum package. The correct command to use is:
sudo apt install ruby-full
The following command will also not install a complete ruby:
sudo apt-get install ruby2.3-dev
For Ubuntu 18, after checking log file mentioned while install
Results logged to /var/canvas/vendor/bundle/ruby/2.5.0/extensions/x86_64-linux/2.5.0/nio4r-2.5.2/gem_make.out
with
less /var/canvas/vendor/bundle/ruby/2.5.0/extensions/x86_64-linux/2.5.0/nio4r-2.5.2/gem_make.out
I noticed that make is not found. So installed make by
sudo apt-get install make
everything worked.
I faced a similar issue on Xcode 12 with macOS 10.15 and cocoapods. Just make sure that the xcode-select command points to the SDK you want to build against. It should build without issues afterwards.
BEFORE you follow the tip from Joki's answer (below) and IF :
you have MacOS 10.14.6
at /Library/Developer/CommandLineTools/SDKs/ you have folders
MacOSX.sdk(symbolic), MacOSX10.14.sdk, MacOSX10.15.sdk
Move MacOSX10.15.sdk to anywhere (admin privileges needs)
Delete symbolic link (admin privileges needs)
At /Library/Developer/CommandLineTools/SDKs/ create another symbolic link now to MacOSX10.14.sdk folder using (admin privileges needs)
sudo ln -s /Library/Developer/CommandLineTools/SDKs/MacOSX10.14.sdk MacOSX.sdk
Now you can follow Joki's answer
WARNING!
If you move MacOSX10.15.sdk folder to /Library/Developer/CommandLineTools/SDKs/ again, the command
ruby -rrbconfig -e 'puts RbConfig::CONFIG["rubyhdrdir"]'
will show MacOSX10.15.sdk folder like default again, nowadays I dunno how to fix it! My suggestion, compress the folder and put the original folder until fix will be available.
macOS RubyMine Gem installation failure
My problem with this error message was when trying to install a Gem via RubyMine. It didn't like that I had changed the global Ruby version with rbenv, so I fixed it by changing back to the system default Ruby version with:
rbenv global system
and restarted RubyMine.
sudo apt-get --reinstall install ruby
try it for ubuntu 16.04
I have a running (in production) mysql instance on my linux server (ubuntu-10.10) however I cannot find my mysql_config file.
command and output:
~$ locate mysql_config
~$
I've heard/read that I need the libmysqlclient-dev package installed to be able to use mysql_config but I don't want to break my current production instance. I want to make sure installing this dev package is not going to have adverse effects on my current mysql databases.
Furthermore, where can I find the source download for libmysqlclient-dev to install manually? In my current situation (behind corporate proxy) I am not permitted to use apt-get's.
UPDATE
this is stemming from attempting to install python-MySQLdb from source. the setup.py file is requiring the mysql_config path and continues to break when trying to use anything but that file.
The mysql_config executable is by default located in the bin directory of the MySQL server installation if you install it from precompiled binaries. But if you install it using apt-get it may not exist on your server.
Try:
sudo apt-get install libmysqlclient-dev
yum install mariadb-devel works for Centos 7 too, this puts the mysql_config into all the places that pip install mysqlclient requires/looks for it.
In Ubuntu 13.04 installation, it is at /usr/bin/mysql_config
Try /etc/my.cnf, that is the standard file for mysql config.
You can also do:
find / -name my.cnf -type f
On Fedora, install the mariadb-devel package, and you should have the binary.