This Xfire thing... it kinda sucks.

He has a point: http://www.xfire.com/forums/182686/topic/2108938/

I have a tendency to put up quite a random status text when I am sleeping to indicate that I will not be replying to messages sent to me immediately.

Last night the text was "/quit" and it was apparently a huge success since I got reactions from three people about it throughout the night:

Quote

[01:45:12] [Xf Helper] 4CH!LL3S: wtf
[05:47:10] XƒίЯξ: what's with /quit?

The third one was just Gamma trying to do something with a chat window to me at [10:04:59] local time but he never sent actual text...

So, to collectively answer those people what the status was about, I posted the answer in this blog since I know they read this text right here.

Okay, so /quit is the command I give on cmdxfire to, well, quit the program, so that the Xfire client running on my computer will lose access to cmdxfire running in proxy mode and so I can start cmdxfire again but logging in locally to keep me logged in to Xfire but that I can shut down my main computer.

And of course in the morning I give /quit to again quit the cmdxfire logged in with my account and then start it again in proxy mode so I can log in with the real Xfire client on my main computer.

So I put /quit in my status message since it was relevant and was descriptive in what is going on (well, for me at least) and to see if that message would get some reaction from people. And it did! Success.

Finland uses comma instead of a dot as the decimal point in numbers, but apparently LaTeX thinks that everyone should use dots even if Finnish style was enabled with babel package. The difference is that if you put a comma in math mode, LaTeX will insert space after it, which is incorrect.

After doing some research I found out a trick to redefine what comma means in math mode:

Code:



Do not ask me what that exactly does or why it works, I do not feel like searching for all the documentations through for the definition of \DeclareMathSymbol and its parameters, but in any case after including that command before \begin{document}, my equations with decimal numbers looked correct with no extra space.

A problem, or actually two problems. The main problem is that the nights are becoming dark, so I need me some lights. The other problem is that it is still very hot in here and using my halogen desk lamp to counter the darkness is therefore not the best idea as it would produce even more heat.

Of course the easy solution would be to just go to a shop and buy a lamp that doesn't produce much heat such as those "energy saving" lamps, or LED lamps. But where would be the fun in that? Instead I am going to build my own LED light system from the boxful of stuff looted from University (the same series of random stuff as you have seen earlier in my blog).

To begin I needed to have some way to test the LEDs that are still attached to all kinds of parts such as bulk Dell computer case front panel circuit boards. First I needed a power source, I chose to test with two rechargeable 1.2 V batteries (AAA sized). Apparently using only one of them was not enough to lit the diodes but two of them providing a total of 2.4 volts seemed to do the trick and I got some green light from one of the test subjects.

But keeping the batteries oriented that way by hand and at the same time trying to hold wires to the LEDs wasn't very nice so I started making a testbed where I could seat the batteries in. It wasn't perfect, I would still need to apply some force when testing to make the batteries have good enough contact with each other.

In the end I found some metal piece that I believe has once been a connector for a CPU fan or something similar. I used that to make a larger connection area to the positive end of my power source and at the same time it was applying the necessary tension to keep the batteries in contact.

So the result of this night is:

• Testbed with batteries and a "contact bridge"
  
• 9 tested green leds
  
• learning the pin orientation of the Dell circuits inside out

And here's an IrfanViewed photo for you to drool on (click to view in full size):


I wonder who wrote that POWER...

Sorry fans, I forgot to post blog entries for you! So here's the picture that my phone uploaded on Facebook (ugh!) some weeks ago:



That's my room fan, trying to keep me cool on days with 37 degrees Celcius and looking at the failing-at-life posts on forums.

At first it seemed very bad... for days. Then there was a slight sign of light, maybe all hope is not lost after all. Next morning it all went down again. What next? I hope I knew...

Read the subject as "Don't do this at production server".

So last week we had a great idea: upgrade the server operating system. It had been running Ubuntu 9.04 amd64 server edition and the lack of the newer versions of libraries such as glibc started to show. Since they cannot just give all new versions as regular updates but instead force you to upgrade the OS version to get bigger updates (I still don't know why) I had to try it out.

Of course I took some precautions by backing up important kind of data such as /var/www contents, MySQL database files and my personal files with development projects and such. Well, since 9.04 couldn't be directly upgraded into 10.04, I had to do an intermediate upgrade into 9.10 first. That worked surprisingly well; only some MySQL tables related to Xfire accounts (do you have any idea what kind of data my database has??) couldn't be automatically converted from MySQL 5.0 to MySQL 5.1 database format. Well, those tables were still usable and if needed I could rebuild them manually myself.

Everything else seemed fine after the first upgrade so it was time for the second attempt, to get to the latest version. Everything went just like with the previous upgrade, files getting updated, services getting restarted and so on. Then something unexpected happened. Just when PowerDNS was being updated the Internet connection dropped for yet unknown reason. While this should normally not have any effect on update because all the packages were downloaded already, it seemed to cause something when PowerDNS service was being restarted. The upgrade stopped there, nothing happening.

Looking at process tree, the powerdns script was marked as [defunct]. Braaaiiiiins! Even after getting connection back it was still stuck there, so I had no choice but try to kill the stuck subprocess in hopes the main upgrade script could then continue.

Unfortunately the upgrade script was made so that if any part of the upgrade returned something else than "success", it...