Blog 2011-11-13 - 2012-09-01

Micro controllers
Vector graphics
Newer entries

Training at
Shaolin Si Wushu Guan Göteborg starts again next week :-D
No Tai Ji or San Da for me this time, though. Only Gong Fu.
I really like my Kindle Touch. I am also sure Amazon likes my Kindle too. It is far too easy to buy stuff with it...

I have recently had quite a bit of problems with a
OpenCV install on windows with MinGW. Initially I used the prebuilt opencv 2.3.1 win superpack, but any call to namedWindow (or other parts of the highgui lib) would cause a crash, in even the most basic test program. This seems to be a fairly common problem, and was not solved until OpenCV was rebuilt from sources instead. The OpenCV site has instructions on how to build OpenCV with MinGW. If you also experience this problem, just rebuild OpenCV from sources instead, and you should be fine too.
Repaired the
page rating functionality available on some of the pages here. Apparently I broke it when moving to the new server. Through the magic of find -exec it is fixed now.
I have dug out my old
Arduino and started experimenting with it. What I don't get is how compiling even the tiniest program can be so slow in Windows. Is all of the Arduino environment being recompiled every time? There are no speed issues compiling the same program, on the same machine, when it is booted up in Ubuntu instead.
Why is almost all
referrer spam I get, Russian?
The only remaining thing I still missed after the new server install was a text-only usenet reader. In the past I've used slrn, but I have not been able to find a prebuilt slrn rpm in any of the standard Scientific Linux 6 or Enterprise Linux 6 repos, and satisfying the build requirements of the
slrn source rpm turned out to be problematic. Instead I was recommended to use alpine. That worked and made me a bit nostalgic (pine was the first email client I ever used), but alpine had enough things bugging me that I started missing slrn again.

Time for another attempt to build slrn. First minor stumbling block is that building slrn requires inews. There is no inews source rpm. Instead inews is built from the inn source rpm. Picking the newest inn source rpm (intended for Fedora 17) led to rather a lot of dependency problems. Building the older inn 2.3.5-5, on the other hand, was easy. The last hurdle was that one of the source rpms tried to link to libdb 4.2, instead of the installed version 4.7, but that was easily solved with a symlink. Now, with slrn installed there is no longer anything I miss from the old server.
Today the new implementation of the old
auto-updating fractal gallery has come far enough so I can activate it again (according to the logs I turned the old version off in 2007). This is a complete re-write from scratch, partially as a test of Eclipse CDT. In the past I have often not bothered with an IDE and stuck to using an editor and gcc in a terminal window, but the more I use Eclipse the more I like it (yes, I know I'm late to the party).
The server switch-over is now complete. Faster hardware and a new distro :-D
Everything looks good so far, although there are a couple of minor things I still miss (jmacs and slrn). Gnome also seems to be a bit screwed up in this install, but since this will run as a server I will almost never be logged in to X, so I don't really care.

Happy new year!

Today I've worked a bit more with the new server replacement. Turns out I don't like Debian very much. Next stop -
Scientific Linux.

Making a bootable USB out of the Scientific Linux LiveDVD iso resulted in a kernel panic, but I did have more luck with the bootable USB made from the LiveCD iso instead. The OS is now installed, but more testing is needed until I know if I like it or not. Looks promising so far, though.
Breaking news! My taste in music doesn't follow a power law distribution!
I've been using for a couple of years. Spotify and the player in my mobile are also set to scrobble. This has collected a rather large data set that I've been meaning to take a closer look at, but never got around to until now. What I expected to find was that the distribution of plays per artist would follow a power law(obviously), but...
graph: distribution of plays per artist
Distribution of plays per artist. doesn't. A lot of things may look like a straight line in a log-log plot, but that doesn't necessarily mean that it is a power law. However, this doesn't even look like a straight line in the graph, so it isn't a power law. What does all this mean then? Probably not that much, but possibly the plays, where I don't have full control over what is played, follows a different distribution from the scrobble sources where I do have full control (Spotify and mobile player). It is at least easy to imagine that the plays would have a more pronounced long-tail component in the distribution.
The new tiny server has got
Debian installed now, but the old server and its successor will have to live in parallel for a while, until all customization and configuration is complete and the switch-over can be done. Won't be any time to complete the process until after christmas, though.
I''ve finally received the hardware for the new server that I ordered. Thank you,
webhallen, but was it really necessary to wait a full month to tell me that the drive I ordered wasn't in stock and wasn't going to get re-stocked either?
I've ordered new hardware for my tiny server. The Intel D525MW mobo should be significantly faster than the old EPIA board that is currently being used. This time I've also decided to give
Debian a go, and see if I like it.
Older entries