Qte.dk

We are currently busy writing our masters thesis, and as always we use LaTeX to write the report. LaTeX documents consists of LaTeX files and a bibliography, and in order to get all references correct for both chapters, sections, figures, and citations to bibliography entries the LaTeX document has to be compiled in 4 steps.

First a list of all labels has to be generated. The second step is to compile the bibliography. The third and fourth steps walk through the document and insert correct references, generate the table of contents etc.

Even though we have a Makefile that make it easy to compile the LaTeX files into a PDF document, the compilation still takes a while, and as the document grows so does the compilation time. Yesterday I discovered a clever little program called rubber, that can greatly decrease the compilation time to a minimum.

Instead of having to go through all four steps every time, the rubber document build system analyzes the document files and determines what compilations steps are necessary to produce the PDF document. If I only modify text in a single file, and do not add any new labels, rubber only performs a single LaTeX compilation. If I add a new reference it performs two LaTeX compilations. Finally if I add a new bibliography entry and reference it, rubber will perform all four compilation steps.

Under normal use, it reduces the compilation time with a factor of three. To use rubber, execute the following command

rubber --texpath=packages --pdf report.tex

If you do not have any extra LaTeX packages, you can skip the --texpath option. As an extra bonus for all the university students - rubber is already installed on the servers, so update your Makefiles today! :-)

Christian Jørgensen just posted a link to an animation on the Hitachi GST website explaining how perpendicular magnetic recording works.

I found it quite amusing - maybe we should start explaining technical topics in this manner :-)

[http://www.hitachigst.com/hdd/research/recording_head/pr/PerpendicularAnimation.html]

Yesterday - thursday - after two days of snow, Tinus, Eckhart, and I went for a ride on one of the snow-covered local hills. Since we are all poor students, we bought a set of large black plastic-bags and used those instead of a real sledge.

They worked fine, except that they gave a bumpy ride down the hill :-)

The next time there is enough snow, we will definitely buy a real sledge and test it againist the plastic-bags.

A month ago I decided it was time to finally make the famous biscuit-cake, that a co-student and I have been talking about for ages. Tinus helped me out, and it was a great success. I have saved the recipe with photos for future reference.

See the recipe at http://qte.dk/projects/recipes/kiksekage/

After the post last year :-) I decided to hack integration with the Mantis bug tracker into the StatCVS. After some quick reading, it did not take long to hack the code, compile, and test it. Everything seemed to work fine, so I generated a diff and sent it to the developers. This morning when I woke up, there were already to possitive comments and one of the developers had committed to their CVS repository.

Great way to start the new year.

The patch can be seen in action on the Fundanemt installation of StatCVS. You might have to scroll down or go to the next page to find some of the bug links.