Archive for category Software

A look back: Roughly 9 years of email behavior analyzed

In passing I have seen several plugins/apps/sites that offer to retrieve your email history and present some sort of analysis. My wife mentioned this recently and I thought this idea sounded like fun so I did a little processing on my own email. I have email records going back to late 2003 – just about the time I started undergrad. So with some grep/java/gnuplot magic I present to you a graph of the timestamps for every email I have sent since then. I just processed sent emails because my behavior should be more evident than in received.


The graph contains a red dataset – my personal emails and a green dataset – emails sent from my university account. The Y axis is time of day and the X axis is the date. Each dot is an email sent at that time on the given day.


  • The first region that you see is before I started logging school email. It just has personal emails… spanning all times of day/night. Seriously? I sent more email from 0000-0800 then 1600-2400???
  • Second region introduces university emails. During this time I also worked for the university part-time 0900-1800. Nothing too surprising here.
  • Third region is what I call early grad school. A general increase in both of my email volumes.
  • Fourth region – the ASSERT period. The ASSERT project was a massive 2-2.5 year research lab project with a lot of people working on it and a lot of work. Check out the huge increase in school emails. I bet you could spot deadlines in there too.
  • Fifth region – Summer internship
  • Sixth region – Late grad school. Post-quals, post-course work, post-ASSERT, just relaxed set my own schedule time
  • Final region – Real Job. Strict no-email hours, waking up early, less email, yup, that’s real life for ya.

Global observations

  • Maybe I imagine it but I swear you see that the time that I start sending emails in the morning gradually becomes later and later — same for cease time. What/who is to blame? Grad school or my GF/fiance/wife?
  • I sent more school email than personal: 5,164 messages to 4,849 messages
  • You can vaguely spot summer breaks. Look for swaths of lower green density bordered by bursts.
  • Sadly you can’t really spot certain events like: marriage, quals, proposal or defense.


Number of emails sent per year:

Year Personal Count School Count
2003 - 57
2004 145 310
2005 412 310
2006 474 354
2007 728 815
2008 831 1,378
2009 804 1,140
2010 893 438
2011 438 464
2012 – to May 68 59

Cross Compile libmysql

The key to cross-compiling libmysql is to google “cmake cross compile” instead of “libmysql cross compile”.  Took me a bit to think of this but it makes sense because cmake is the build system for libmysql.

My How-To

My how-to assumes you have a Linux system with standard development tools and compilers installed as well as GNU cross-compilers for the target platform.

  1. Download libmysql source: In the drop down box change to source. Otherwise you will be presented with pre-compiled binaries.
  2. Extract libmysql: tar -xzf mysql-connector-c-x.y.z.tar.gz
  3. run: cd mysql-connector-c-x.y.z
  4. Create a toolchain.cmake file (See section below)
  5. run: cmake -G “Unix Makefiles” -DCMAKE_INSTALL_PREFIX=`pwd`/install -DCMAKE_TOOLCHAIN_FILE=toolchain-arm-linux.cmake
  6. run: make
  7. run: make install
  8. Your libs are in `pwd`/install/lib

Toolchain file

The toolchain file sets a few parameters for the cmake system to override the default behavior of searching for the system compilers. I modified a sample from CMake wiki. Here is my modified sample that cross-compiles for an ARM processor running Linux.

# this one is important
#this one not so much
# specify the cross compiler
SET(CMAKE_C_COMPILER   /usr/local/bin/arm-linux-gcc)
SET(CMAKE_CXX_COMPILER /usr/local/bin/arm-linux-g++)
# where is the target environment
#SET(CMAKE_FIND_ROOT_PATH  /opt/eldk-2007-01-19/ppc_74xx /home/alex/eldk-ppc74xx-inst)
# search for programs in the build host directories
# for libraries and headers in the target directories

Enable OpenGL hardware acceleration for Java 2D

For a couple years now I have been disappointed that one of my 2D Java programs had HORRIBLE performance on my laptop. I noticed the problem got horrible when I upgraded to Vista and didn’t get fixed with my upgrade to Windows 7. The program is a 2D graphing application that draws multiple data streams as scrolling waveforms on the screen in real-time. It is written using Java Swing libraries and the Graphics2D API. On my laptop I was getting less than 1 update per second. On my work computer I was exceeding 200!

Over the last couple years I tried everything to no avail. Finally today I stumbled across this article that describes several unsupported Java system properties that affect the 2D graphics system.

I tried adding the following line to my program and voila! Awesome performance. Problem solved.