I often use the linux watch command to monitor the status of certain commands. When I’m copying lots of files say, I’d watch the files in the target directory to see what files have already been copied across with the following command:

watch ls -l

The watch program clears the screen and displays the output of “ls -l” every 2 seconds.

Sometimes I’ll want to monitor a command that only outputs a single line. If I wanted to see the total number of files in a directory rather than the files themselves I could use the command “ls -l | wc -l”. The fact that watch clears the whole screen can be a little annoying here though, because the command is only outputting a single line. That is why I came up with the following small bash script, linewatch.

Linewatch repeatedly calls any arguments passed to it every 2 seconds (in the same way watch does), but only clears a single line rather than the whole screen. Here is the code:

#!/bin/bash
clearline="\b\033[2K\r"
command=$@

while true
do
    eval "$command"
    sleep 2
    echo -n -e "$clearline"
done

And here is an example of how to call it:

$ ./linewatch "ls -l | wc -l"
24

The number of files in the current directory (24 in the example) will keep update every 2 seconds. Just hit Ctrl-C when you want to quit,

I was really pleased to come across the “SQL Antipatterns Stirke Back” presentation recently, which discusses common mistakes with SQL database design. It gives some really good advice on how best to design databases to avoid these issues. I’ve certainly made some of the mistakes mentioned, and I’m sure I’ll be referring back to this presentation again and again!

I came up with the following shell script recently to monitor code changes in a subversion repository. On the first run it will emails out the 10 most recent changes. After that the script mails out all changes since the last time it was run. You can set it up to run as a daily cron job which mails you all changes made to you favourite open source project!

It wouldn’t take much to get it working with other version control systems such as Git or Bazaar, or to do some nice formatting of the output instead of outputting the raw svn log as-is. Let me know if you find it useful!

#!/bin/bash
# Shell script to email the latest changes in an SVN
# repsitory to a specified email address.
# Ben Dowling - wwww.coderholic.com

svnUrl="http://anonsvn.wireshark.org/wireshark/trunk/"
lastRevisionFile="./.last-revision"
mailto="ben@coderholic.com"

function getCurrentRevision {
  # Get the current SVN revision, eg. "r4670"
  currentRevision=$(svn log "$svnUrl" -r HEAD 2>/dev/null | head -n2 | grep -v -- "-------" | awk '{ print $1 }')
  # Strip off the 'r'
  currentRevision="${currentRevision:1}"
  echo "$currentRevision"
}

currentRevision=$(getCurrentRevision)

# If we've run this program before then we've stored the SVN revision at the time
if [ -f "$lastRevisionFile" ]
then
  lastRevision=$(cat "$lastRevisionFile")
  #  Check what the current revision is, and exit if there
  # haven't been any changes since we last checked
  if [ $currentRevision -lt $lastRevision ]
  then
      echo "No changes since last check"
      exit
  fi
else
  # We haven't run this program before, so set the last revision to the current revision - 10
  lastRevision=$(echo "$currentRevision - 10" | bc)
fi

# Mail the SVN changes
svn log "$svnUrl" -r "HEAD:${lastRevision}" | mail -s "SVN changes for $svnUrl" $mailto

# Store the current revision + 1 as the last revision
revision=$(echo "$currentRevision + 1" | bc)
echo "$revision" > "$lastRevisionFile"