I have this configuration file in my CI where I'm specifying a header file and some CMAKE flags on one line.
The configuration file looks like this (filelist):
./settings6.h -DMY_COMPILE_FLAGS="-m32 -fstrict-aliasing"
./settings7.h -DMY_FEATURE_1=ON
./settings8.h -DMY_FLAG=ON -DMY_FEATURE_2=ON -DMY_INCLUDE_DIR=/usr/include/
Now, I'm using a bash script to process this configuration file:
#!/bin/bash
SCRIPTDIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
while read i; do
HEADERFILE=$(echo $i | cut -d ' ' -f 1)
CMAKEFLAGS=$(echo $i | cut -s -d ' ' -f 2-)
if [[ "$HEADERFILE" == "" ]]; then
continue
fi
CFLAGS="-Werror" cmake "my_build_dir" "$CMAKEFLAGS" -G "Ninja" -DMY_EXTRA_INCLUDE="$SCRIPTDIR/$HEADERFILE" -B"build_env_dir" > /dev/null
ninja -C "build_env_dir"
done <<ENDOFINPUT
$(grep -v '^#' $SCRIPTDIR/filelist)
ENDOFINPUT
When I have the bash script as above, the line with settings6.h gets processed properly, i.e. the MY_COMPILE_FLAGS are set to -m32 -fstrict-aliasing.
However, settings8.h is failing because the value of MY_FLAG is seen by CMAKE as ON -DMY_FEATURE_2=ON -DMY_INCLUDE_DIR=/usr/include/, so MY_FEATURE_2 and MY_INCLUDE_DIR are not processed correctly.
After googling around a bit, I thought, well, surely a quoting issue, probably I have to remove the quotes around $CMAKEFLAGS like this:
CFLAGS="-Werror" cmake "my_build_dir" $CMAKEFLAGS -G "Ninja" -DMY_EXTRA_INCLUDE="$SCRIPTDIR/$HEADERFILE" -B"build_env_dir" > /dev/null
In fact, this lets settings8.h work as expected (all three options are processed), but now, settings6.h is suddenly failing since CMAKE complains:
CMake Error: The source directory "/src/-fstrict-aliasing"" does not exist
Can someone guide me please how I read the settings correctly from my filelist so that settings6.h and settings8.h both succeed?
Here's a Makefile which refactors this into a sequence of recipes.
Cases := $(patsubst %.h,%,$(wildcard ./settings*.h))
all_done := $(patsubst %,.%.done,$(Cases))
.PHONY: all
all: $(all_done)
cases.mk: filelist.txt
sed 's%^\./%case_%;s% % := %' $< >$#
include cases.mk
.%_done: ./%.h
CFLAGS="-Werror" cmake "my_build_dir" $(case_$*) -G "Ninja" \
-DMY_EXTRA_INCLUDE="$<" -B"build_env_dir" > /dev/null
ninja -C "build_env_dir"
How to find the operating system using bash script? I found this answer: Detect the OS from a Bash script. It is not clear it would work on Mac OS X.
I would like to find it on Mac OS X vs different linux OS's.
For Linux you can type in the following bash command:
$ cat /etc/*-release
For Mac OS X you can try one of these commands:
$ sw_vers -productVersion
$ system_profiler SPSoftwareDataType
Derived from other answers, this worked for me:
CURRENT_OS="OSX" #CENTOS, UBUNUTU are other valid options
function findCurrentOSType()
{
echo "Finding the current os type"
echo
osType=$(uname)
case "$osType" in
"Darwin")
{
echo "Running on Mac OSX."
CURRENT_OS="OSX"
} ;;
"Linux")
{
# If available, use LSB to identify distribution
if [ -f /etc/lsb-release -o -d /etc/lsb-release.d ]; then
DISTRO=$(gawk -F= '/^NAME/{print $2}' /etc/os-release)
else
DISTRO=$(ls -d /etc/[A-Za-z]*[_-][rv]e[lr]* | grep -v "lsb" | cut -d'/' -f3 | cut -d'-' -f1 | cut -d'_' -f1)
fi
CURRENT_OS=$(echo $DISTRO | tr 'a-z' 'A-Z')
} ;;
*)
{
echo "Unsupported OS, exiting"
exit
} ;;
esac
}
use uname
$(uname -s)
this will give you the os name (Darwin = OSX)
$(uname -v)
will give you the os version
see uname manual here
Is there a way to make gcc ignore an invalid option, instead of dying with "unrecongized option"? Reason is I want to use an option only available in later versions of gcc (-static-libstdc++), but it should also compile on older compilers. I could check for gcc version in the makefile but it is a bit ugly.
no, but you can set the flags based on the gcc version as follows:
version=`gcc --version | head -1 | cut -d ' ' -f3`
if [ `echo -e "$version\n4.6.1" | sort -V -C; echo $?` == 0 ]; then
flags = -static-libstdc++;
fi
gcc $flags ...
(Disclaimer: I'm not sure which version first uses static-libstdc++, 4.6.1 is just a guess).
John
You can run gcc and check if it accepts the flag:
STATIC_LIBCPP_FLAG := $(shell if gcc -static-libstdc++ --version 2>&1 | grep -q 'unrecognized option'; then true; else echo -static-libstdc++; fi)
CFLAGS += $(STATIC_LIBCPP_FLAG)
On this line:
GCCVER:=$(shell a=`mktemp` && echo $'#include <stdio.h>\nmain() {printf("%u.%u\\n", __GNUC__, __GNUC_MINOR__);}' | gcc -o "$a" -xc -; "$a"; rm "$a")
I get:
*** unterminated call to function `shell': missing `)'. Stop.
What's wrong with my stupidly circuitous variable?
Update0
$ make --version
GNU Make 3.81
$ bash --version
GNU bash, version 4.2.8(1)-release (x86_64-pc-linux-gnu)
$ uname -a
Linux 2.6.38-10-generic #46-Ubuntu SMP x86_64 GNU/Linux
$ gcc --version
gcc (Ubuntu/Linaro 4.5.2-8ubuntu4) 4.5.2
when using $ for Bash inside a Makefile, you need to double them: $$a for example. I'm not familiar with the notation $' but I'll have to assume you know what you're doing with that. unless it's a Makefile construct, you need to double the dollar sign on that one too.
also, the hash sign # is terminating the shell expansion in Make's evaluation, which is why it never sees the right paren. escaping it helps, but I don't have it working quite right yet.
I'm debugging it by having two steps: first is setting GCCVER to be the list of commands without the enclosing $(shell), then in the 2nd step setting GCCVER := $(shell $(GCCVER)). you might want to try that too, commenting out the $(shell) step when it doesn't work, using export, and making a "set" recipe:
GCCVER := some commands here
#GCCVER := $(shell $(GCCVER)) # expand the commands, commented out now
export # all variables available to shell
set:
set # make sure this is prefixed by a tab, not spaces
Then:
make set | grep GCCVER
[update] this works:
GCCVER := a=`mktemp` && echo -e '\#include <stdio.h>\nmain() {printf("%u.%u\\n", __GNUC__, __GNUC_MINOR__);}' | gcc -o "$$a" -xc -; "$$a"; rm "$$a"
GCCVER := $(shell $(GCCVER))
export
default:
set
jcomeau#intrepid:/tmp$ make | grep GCCVER
GCCVER=4.6
And full circle, having gotten rid of the extra step:
jcomeau#intrepid:/tmp$ make | grep GCCVER; cat Makefile
GCCVER=4.6
GCCVER := $(shell a=`mktemp` && echo -e '\#include <stdio.h>\nmain() {printf("%u.%u\\n", __GNUC__, __GNUC_MINOR__);}' | gcc -o "$$a" -xc -; "$$a"; rm "$$a")
export
default:
set
Using the $' Bash construct:
jcomeau#intrepid:/tmp$ make | grep GCCVER; cat Makefile
GCCVER=4.6
GCCVER := $(shell a=`mktemp` && echo $$'\#include <stdio.h>\nmain() {printf("%u.%u\\n", __GNUC__, __GNUC_MINOR__);}' | gcc -o "$$a" -xc -; "$$a"; rm "$$a")
export
default:
set
Since your system doesn't work the same as mine, I'm going to cop out and say either use reinierpost's suggestion or, alternatively:
GCCVER := $(shell gcc -dumpversion | cut -d. -f1,2)
I would like to keep my .bashrc and .bash_login files in version control so that I can use them between all the computers I use. The problem is I have some OS specific aliases so I was looking for a way to determine if the script is running on Mac OS X, Linux or Cygwin.
What is the proper way to detect the operating system in a Bash script?
I think the following should work. I'm not sure about win32 though.
if [[ "$OSTYPE" == "linux-gnu"* ]]; then
# ...
elif [[ "$OSTYPE" == "darwin"* ]]; then
# Mac OSX
elif [[ "$OSTYPE" == "cygwin" ]]; then
# POSIX compatibility layer and Linux environment emulation for Windows
elif [[ "$OSTYPE" == "msys" ]]; then
# Lightweight shell and GNU utilities compiled for Windows (part of MinGW)
elif [[ "$OSTYPE" == "win32" ]]; then
# I'm not sure this can happen.
elif [[ "$OSTYPE" == "freebsd"* ]]; then
# ...
else
# Unknown.
fi
For my .bashrc, I use the following code:
platform='unknown'
unamestr=$(uname)
if [[ "$unamestr" == 'Linux' ]]; then
platform='linux'
elif [[ "$unamestr" == 'FreeBSD' ]]; then
platform='freebsd'
fi
Then I do somethings like:
if [[ $platform == 'linux' ]]; then
alias ls='ls --color=auto'
elif [[ $platform == 'freebsd' ]]; then
alias ls='ls -G'
fi
It's ugly, but it works. You may use case instead of if if you prefer.
The bash manpage says that the variable OSTYPE stores the name of the operation system:
OSTYPE Automatically set to a string that describes the operating system on which bash is executing. The default is system-
dependent.
It is set to linux-gnu here.
$OSTYPE
You can simply use pre-defined $OSTYPE variable e.g.:
case "$OSTYPE" in
solaris*) echo "SOLARIS" ;;
darwin*) echo "OSX" ;;
linux*) echo "LINUX" ;;
bsd*) echo "BSD" ;;
msys*) echo "WINDOWS" ;;
cygwin*) echo "ALSO WINDOWS" ;;
*) echo "unknown: $OSTYPE" ;;
esac
However it's not recognized by the older shells (such as Bourne shell).
uname
Another method is to detect platform based on uname command.
See the following script (ready to include in .bashrc):
# Detect the platform (similar to $OSTYPE)
OS="`uname`"
case $OS in
'Linux')
OS='Linux'
alias ls='ls --color=auto'
;;
'FreeBSD')
OS='FreeBSD'
alias ls='ls -G'
;;
'WindowsNT')
OS='Windows'
;;
'Darwin')
OS='Mac'
;;
'SunOS')
OS='Solaris'
;;
'AIX') ;;
*) ;;
esac
You can find some practical example in my .bashrc.
Here is similar version used on Travis CI:
case $(uname | tr '[:upper:]' '[:lower:]') in
linux*)
export TRAVIS_OS_NAME=linux
;;
darwin*)
export TRAVIS_OS_NAME=osx
;;
msys*)
export TRAVIS_OS_NAME=windows
;;
*)
export TRAVIS_OS_NAME=notset
;;
esac
Detecting operating system and CPU type is not so easy to do portably. I have a sh script of about 100 lines that works across a very wide variety of Unix platforms: any system I have used since 1988.
The key elements are
uname -p is processor type but is usually unknown on modern Unix platforms.
uname -m will give the "machine hardware name" on some Unix systems.
/bin/arch, if it exists, will usually give the type of processor.
uname with no arguments will name the operating system.
Eventually you will have to think about the distinctions between platforms and how fine you want to make them. For example, just to keep things simple, I treat i386 through i686 , any "Pentium*" and any "AMD*Athlon*" all as x86.
My ~/.profile runs an a script at startup which sets one variable to a string indicating the combination of CPU and operating system. I have platform-specific bin, man, lib, and include directories that get set up based on that. Then I set a boatload of environment variables. So for example, a shell script to reformat mail can call, e.g., $LIB/mailfmt which is a platform-specific executable binary.
If you want to cut corners, uname -m and plain uname will tell you what you want to know on many platforms. Add other stuff when you need it. (And use case, not nested if!)
I recommend to use this complete bash code
lowercase(){
echo "$1" | sed "y/ABCDEFGHIJKLMNOPQRSTUVWXYZ/abcdefghijklmnopqrstuvwxyz/"
}
OS=`lowercase \`uname\``
KERNEL=`uname -r`
MACH=`uname -m`
if [ "{$OS}" == "windowsnt" ]; then
OS=windows
elif [ "{$OS}" == "darwin" ]; then
OS=mac
else
OS=`uname`
if [ "${OS}" = "SunOS" ] ; then
OS=Solaris
ARCH=`uname -p`
OSSTR="${OS} ${REV}(${ARCH} `uname -v`)"
elif [ "${OS}" = "AIX" ] ; then
OSSTR="${OS} `oslevel` (`oslevel -r`)"
elif [ "${OS}" = "Linux" ] ; then
if [ -f /etc/redhat-release ] ; then
DistroBasedOn='RedHat'
DIST=`cat /etc/redhat-release |sed s/\ release.*//`
PSUEDONAME=`cat /etc/redhat-release | sed s/.*\(// | sed s/\)//`
REV=`cat /etc/redhat-release | sed s/.*release\ // | sed s/\ .*//`
elif [ -f /etc/SuSE-release ] ; then
DistroBasedOn='SuSe'
PSUEDONAME=`cat /etc/SuSE-release | tr "\n" ' '| sed s/VERSION.*//`
REV=`cat /etc/SuSE-release | tr "\n" ' ' | sed s/.*=\ //`
elif [ -f /etc/mandrake-release ] ; then
DistroBasedOn='Mandrake'
PSUEDONAME=`cat /etc/mandrake-release | sed s/.*\(// | sed s/\)//`
REV=`cat /etc/mandrake-release | sed s/.*release\ // | sed s/\ .*//`
elif [ -f /etc/debian_version ] ; then
DistroBasedOn='Debian'
DIST=`cat /etc/lsb-release | grep '^DISTRIB_ID' | awk -F= '{ print $2 }'`
PSUEDONAME=`cat /etc/lsb-release | grep '^DISTRIB_CODENAME' | awk -F= '{ print $2 }'`
REV=`cat /etc/lsb-release | grep '^DISTRIB_RELEASE' | awk -F= '{ print $2 }'`
fi
if [ -f /etc/UnitedLinux-release ] ; then
DIST="${DIST}[`cat /etc/UnitedLinux-release | tr "\n" ' ' | sed s/VERSION.*//`]"
fi
OS=`lowercase $OS`
DistroBasedOn=`lowercase $DistroBasedOn`
readonly OS
readonly DIST
readonly DistroBasedOn
readonly PSUEDONAME
readonly REV
readonly KERNEL
readonly MACH
fi
fi
echo $OS
echo $KERNEL
echo $MACH
more examples examples here: https://github.com/coto/server-easy-install/blob/master/lib/core.sh
I would suggest avoiding some of these answers. Don't forget that you can choose other forms of string comparison, which would clear up most of the variations, or ugly code offered.
One such solution would be a simple check, such as:
if [[ "$OSTYPE" =~ ^darwin ]]; then
Which has the added benefit of matching any version of Darwin, despite it's version suffix. This also works for any variations of Linux one may expect.
You can see some additional examples within my dotfiles here
uname
or
uname -a
if you want more information
In bash, use $OSTYPE and $HOSTTYPE, as documented; this is what I do. If that is not enough, and if even uname or uname -a (or other appropriate options) does not give enough information, there’s always the config.guess script from the GNU project, made exactly for this purpose.
Try using "uname". For example, in Linux: "uname -a".
According to the manual page, uname conforms to SVr4 and POSIX, so it should be available on Mac OS X and Cygwin too, but I can't confirm that.
BTW: $OSTYPE is also set to linux-gnu here :)
I wrote these sugars in my .bashrc:
if_os () { [[ $OSTYPE == *$1* ]]; }
if_nix () {
case "$OSTYPE" in
*linux*|*hurd*|*msys*|*cygwin*|*sua*|*interix*) sys="gnu";;
*bsd*|*darwin*) sys="bsd";;
*sunos*|*solaris*|*indiana*|*illumos*|*smartos*) sys="sun";;
esac
[[ "${sys}" == "$1" ]];
}
So I can do stuff like:
if_nix gnu && alias ls='ls --color=auto' && export LS_COLORS="..."
if_nix bsd && export CLICOLORS=on && export LSCOLORS="..."
if_os linux && alias psg="ps -FA | grep" #alternative to pgrep
if_nix bsd && alias psg="ps -alwx | grep -i" #alternative to pgrep
if_os darwin && alias finder="open -R"
Below it's an approach to detect Debian and RedHat based Linux OS making use of the /etc/lsb-release and /etc/os-release (depending on the Linux flavor you're using) and take a simple action based on it.
#!/bin/bash
set -e
YUM_PACKAGE_NAME="python python-devl python-pip openssl-devel"
DEB_PACKAGE_NAME="python2.7 python-dev python-pip libssl-dev"
if cat /etc/*release | grep ^NAME | grep CentOS; then
echo "==============================================="
echo "Installing packages $YUM_PACKAGE_NAME on CentOS"
echo "==============================================="
yum install -y $YUM_PACKAGE_NAME
elif cat /etc/*release | grep ^NAME | grep Red; then
echo "==============================================="
echo "Installing packages $YUM_PACKAGE_NAME on RedHat"
echo "==============================================="
yum install -y $YUM_PACKAGE_NAME
elif cat /etc/*release | grep ^NAME | grep Fedora; then
echo "================================================"
echo "Installing packages $YUM_PACKAGE_NAME on Fedorea"
echo "================================================"
yum install -y $YUM_PACKAGE_NAME
elif cat /etc/*release | grep ^NAME | grep Ubuntu; then
echo "==============================================="
echo "Installing packages $DEB_PACKAGE_NAME on Ubuntu"
echo "==============================================="
apt-get update
apt-get install -y $DEB_PACKAGE_NAME
elif cat /etc/*release | grep ^NAME | grep Debian ; then
echo "==============================================="
echo "Installing packages $DEB_PACKAGE_NAME on Debian"
echo "==============================================="
apt-get update
apt-get install -y $DEB_PACKAGE_NAME
elif cat /etc/*release | grep ^NAME | grep Mint ; then
echo "============================================="
echo "Installing packages $DEB_PACKAGE_NAME on Mint"
echo "============================================="
apt-get update
apt-get install -y $DEB_PACKAGE_NAME
elif cat /etc/*release | grep ^NAME | grep Knoppix ; then
echo "================================================="
echo "Installing packages $DEB_PACKAGE_NAME on Kanoppix"
echo "================================================="
apt-get update
apt-get install -y $DEB_PACKAGE_NAME
else
echo "OS NOT DETECTED, couldn't install package $PACKAGE"
exit 1;
fi
exit 0
Output example for Ubuntu Linux:
delivery#delivery-E5450$ sudo sh detect_os.sh
[sudo] password for delivery:
NAME="Ubuntu"
===============================================
Installing packages python2.7 python-dev python-pip libssl-dev on Ubuntu
===============================================
Ign http://dl.google.com stable InRelease
Get:1 http://dl.google.com stable Release.gpg [916 B]
Get:2 http://dl.google.com stable Release [1.189 B]
...
You can use the following:
OS=$(uname -s)
then you can use OS variable in your script.
I wrote a personal Bash library and scripting framework that uses GNU shtool to do a rather accurate platform detection.
GNU shtool is a very portable set of scripts that contains, among other useful things, the 'shtool platform' command. Here is the output of:
shtool platform -v -F "%sc (%ac) %st (%at) %sp (%ap)"
on a few different machines:
Mac OS X Leopard:
4.4BSD/Mach3.0 (iX86) Apple Darwin 9.6.0 (i386) Apple Mac OS X 10.5.6 (iX86)
Ubuntu Jaunty server:
LSB (iX86) GNU/Linux 2.9/2.6 (i686) Ubuntu 9.04 (iX86)
Debian Lenny:
LSB (iX86) GNU/Linux 2.7/2.6 (i686) Debian GNU/Linux 5.0 (iX86)
This produces pretty satisfactory results, as you can see. GNU shtool is a little slow, so I actually store and update the platform identification in a file on the system that my scripts call. It's my framework, so that works for me, but your mileage may vary.
Now, you'll have to find a way to package shtool with your scripts, but it's not a hard exercise. You can always fall back on uname output, also.
EDIT:
I missed the post by Teddy about config.guess (somehow). These are very similar scripts, but not the same. I personally use shtool for other uses as well, and it has been working quite well for me.
try this:
DISTRO=$(cat /etc/*-release | grep -w NAME | cut -d= -f2 | tr -d '"')
echo "Determined platform: $DISTRO"
This should be safe to use on all distros.
$ cat /etc/*release
This produces something like this.
DISTRIB_ID=LinuxMint
DISTRIB_RELEASE=17
DISTRIB_CODENAME=qiana
DISTRIB_DESCRIPTION="Linux Mint 17 Qiana"
NAME="Ubuntu"
VERSION="14.04.1 LTS, Trusty Tahr"
ID=ubuntu
ID_LIKE=debian
PRETTY_NAME="Ubuntu 14.04.1 LTS"
VERSION_ID="14.04"
HOME_URL="http://www.ubuntu.com/"
SUPPORT_URL="http://help.ubuntu.com/"
BUG_REPORT_URL="http://bugs.launchpad.net/ubuntu/"
Extract/assign to variables as you wish
Note: On some setups. This may also give you some errors that you can ignore.
cat: /etc/upstream-release: Is a directory
You can use following if clause and expand it as needed:
if [ "${OSTYPE//[0-9.]/}" == "darwin" ]
then
aminute_ago="-v-1M"
elif [ "${OSTYPE//[0-9.]/}" == "linux-gnu" ]
then
aminute_ago="-d \"1 minute ago\""
fi
This is what I use if anyone is interested in detecting WSL vs WSL verion 2 as well.
#!/usr/bin/env bash
unameOut=$(uname -a)
case "${unameOut}" in
*Microsoft*) OS="WSL";; #must be first since Windows subsystem for linux will have Linux in the name too
*microsoft*) OS="WSL2";; #WARNING: My v2 uses ubuntu 20.4 at the moment slightly different name may not always work
Linux*) OS="Linux";;
Darwin*) OS="Mac";;
CYGWIN*) OS="Cygwin";;
MINGW*) OS="Windows";;
*Msys) OS="Windows";;
*) OS="UNKNOWN:${unameOut}"
esac
echo ${OS};
I tend to keep my .bashrc and .bash_alias on a file share that all platforms can access. This is how I conquer the problem in my .bash_alias:
if [[ -f (name of share)/.bash_alias_$(uname) ]]; then
. (name of share)/.bash_alias_$(uname)
fi
And I have for example a .bash_alias_Linux with:
alias ls='ls --color=auto'
This way I keep platform specific and portable code separate, you can do the same for .bashrc
I tried the above messages across a few Linux distros and found the following to work best for me. It’s a short, concise exact word answer that works for Bash on Windows as well.
OS=$(cat /etc/*release | grep ^NAME | tr -d 'NAME="') #$ echo $OS # Ubuntu
This checks a bunch of known files to identfy if the linux distro is Debian or Ubunu, then it defaults to the $OSTYPE variable.
os='Unknown'
unamestr="${OSTYPE//[0-9.]/}"
os=$( compgen -G "/etc/*release" > /dev/null && cat /etc/*release | grep ^NAME | tr -d 'NAME="' || echo "$unamestr")
echo "$os"
Doing the following helped perform the check correctly for ubuntu:
if [[ "$OSTYPE" =~ ^linux ]]; then
sudo apt-get install <some-package>
fi