sudo in Debian squeeze

Debian squeeze creates an advanced config file for sudo package. File contains the following lines:

  %sudo ALL=(ALL) ALL
  #includedir /etc/sudoers.d

The first one allows all users from sudo group to do everything they want and the second one demonstrates a way to create a flexible sudo configuration.

Unattended shutdown

My notebook seems to have a hardware defect. It might self power on after being powered off for some time.
This bug quickly degrades the battery making me recharge it too often. To prevent the system from discharging the battery I've wrote the following startup script:

#!/bin/sh

### BEGIN INIT INFO
# Provides:        haltIfNousers
# Required-Start:
# Required-Stop:
# Should-Start:        gdm
# Default-Start:    2 3
# Default-Stop:        4 5
# Short-Description: Execute the halt command.
# Description:
### END INIT INFO

#Powers off the system if no  user is logged in in 30 seconds

BLOCK_FILE=/tmp/haltIfNousers

case "$1" in
    start)
        $0 wait &
        ;;
    wait)
        touch "$BLOCK_FILE"
        sleep 30
        [ -e "$BLOCK_FILE" ] || exit 0
        if ! who|grep '.*' -q ; then
            echo $0: No users are logged in. Halting now.
            shutdown -h now
            exit 0
        fi
        rm -f "$BLOCK_FILE"
    ;;  
    restart|reload|force-reload)
        echo "Error: argument '$1' not supported" >&2
        exit 3
    ;;
    stop)
            rm -f "$BLOCK_FILE"
        ;;
  *)
        echo "Usage: $0 start|stop" >&2
        exit 3
        ;;
esac

This script checks if a user is logged on in 30 seconds after system boot and if he is not, shuts the system down.

Commented lines at the top of the script are Debian LSB header that drives its new dependecy based boot system. This header requests starting the script after gdb making measured time interval for user to login more precise.

Multiple make

To clean multiple folders at once:

echo base_path/{subfolder1,subfolder2} | xargs -n 1 make clean -C

RPM atomated building

If your are providing sources or binaries of your software as RPMs you should be aware of a smart way of automating RPM creation.
To create RPM building environment:

1. create ~/.rpmmacros with the following content:
   

   
   %_topdir .../rpm #Arbitrary
   %_tmppath /tmp/rpm #Optional 
   

   
   Three dots here denote your favorite RPM building directory infrastructure root. (/home/username will do)
   
2. use the following script:
   

   
   #!/bin/sh
   for d in _topdir _sourcedir _rpmdir _srcrpmdir _specdir _builddir _tmppath; do
     mkdir -p `rpm --eval %{$d}`
   done
   

   
   

In your RPM generation script use constructions like

# Doesn't override previously set value. See Parameter Expansion
SOURCES=${SOURCES:-`rpm --eval %{_sourcedir}`} 
#Assumes your sources are in $PROJECTS/$PROJECT directory
tar --wildcards --exclude \*\.svn\* --exclude \*\~ \ -C "$PROJECTS" -zcf "$SOURCES/$PROJECT.tar.gz" $PROJECT
rpmbuild -ba "$PROJECTS/$PROJECT/$PROJECT.spec"

Following this recommendation you will never need to force your packagers to use a harcoded SOURCES folder or to do configuration of your package with their RPM settings.

Again, thanks to comrade Buzykaev, for his tar command idea

Bash versioning

Here is the bash command:

cp -i aaa/bbb/ccc/ddd/xxxx_{6,7}.tpl

This expands to

cp -i aaa/bbb/ccc/ddd/xxxx_6.tpl aaa/bbb/ccc/ddd/xxxx_7.tpl

This feature of bash is called Brace Expansion and can ease versioning of files based on filesystem. If you have a sequence of files with versions assigned to them as incremented part of the name, you can easily create one more with the given command:

> ls
aaa/bbb/ccc/ddd/xxxx_4.tpl
aaa/bbb/ccc/ddd/xxxx_5.tpl
aaa/bbb/ccc/ddd/xxxx_6.tpl
> cp -i aaa/bbb/ccc/ddd/xxxx_{6,7}.tpl
> ls
aaa/bbb/ccc/ddd/xxxx_4.tpl
aaa/bbb/ccc/ddd/xxxx_5.tpl
aaa/bbb/ccc/ddd/xxxx_6.tpl
aaa/bbb/ccc/ddd/xxxx_7.tpl

Even nicer usage:

cp filename{,_`date '+%F'`}

A floating point and C++ templates

Find a bug in the following line:
double sum = std::accumulate(V.begin(), V.end(), 0);


Think more...


This bug can't be easiliy avoided as the connection between internal accumulator variable, return value and third argument types is not obvious and is very well incapsulated by implementation.

The right answer is:
double sum = std::accumulate(V.begin(), V.end(), double(0)); //0.L gives off the same effect

Here is GNU implementation of template function in question:

  /**
   *  @brief  Accumulate values in a range.
   *
   *  Accumulates the values in the range [first,last) using operator+().  The
   *  initial value is @a init.  The values are processed in order.
   *
   *  @param  first  Start of range.
   *  @param  last  End of range.
   *  @param  init  Starting value to add other values to.
   *  @return  The final sum.
   */
  template
    inline _Tp
    accumulate(_InputIterator __first, _InputIterator __last, _Tp __init)
    {
      // concept requirements
      __glibcxx_function_requires(_InputIteratorConcept<_InputIterator>)
      __glibcxx_requires_valid_range(__first, __last);

      for (; __first != __last; ++__first)
        __init = __init + *__first;
      return __init;
    }

Empty structures in C and C++

//test.c
#include <stdio.h>

struct Temp0{};
struct Temp1{char a;};

int main() {
    printf("Temp0 size: %d\n", sizeof(struct Temp0));
    printf("Temp1 size: %d\n", sizeof(struct Temp1));
    return 0;
}

This code being compiled with GNU toolchain gives the following results:

$ gcc test.c -o test && ./test
Temp0 size: 0
Temp1 size: 1



$ g++ test.c -o test && ./test
Temp0 size: 1
Temp1 size: 1


The reasons explained

I wonder are other toolchains give the same results...