I have created django-publications (it is actually a github fork, but has by now evolved into completely separate direction than the original project) primarily to address our lab’s publication repository problem. A production version of the system is available here, and even though the site needs some front-end polish, the system is working well. Anyone interested in setting up a similar service and extending the code is welcome to fork the project on GitHub.
Since I do not have anything new to post this week, here is an old logo that I made for the CogX EU research project that our lab was a part of. I also did some derivative event logos for three spring schools that were organized within the project.
I have stopped using a powerful desktop computer at work about a year ago and switched to a laptop (an Zenbook actually). With the switch I have also decided to move to remote virtual machine for all processor hungry computations, many of which are done in Matlab.
I connect to the remote machine using NX protocol and from the beginning I have noticed a strange lag only present in Matlab interface. Actually I think it is a Java problem, because Matlab interface is written in Java, but that is not the point. The point is that this was driving me nuts for half a year and then I have finally decided to solve it and I am posting the solution here for everyone to see. After a bit of searching I found this post that describes a correct configuration for official NX client. Unfortunately I do not use a client with all these options present, so I went looking for a server-based solution.
I have found a configuration file for the NX node, called /usr/NX/etc/node.cfg. In the file there is a (commented out) option for extra client options with the key AgentExtraOptions. Uncomment the line and modify it to look like this:
AgentExtraOptions = "-nocomposite -noshpix -noshmem"
This way a new session will be started with disabled shared memory extension and disabled emulation of shared pixmaps, which for some reason cause problems with Java Swing GUI (I have no idea why). After that you have to start a new session to see a difference. Now Matlab should work like a charm. Unfortunately other, native applications on Linux (which I am using on the virtual machine) work less smoothly, but I find it less annoying than previous Matlab unresponsive behavior.
Earlier this month, almost a year after being accepted, my scientific paper was finally published in the IEEE Transactions on Pattern Recognition and Machine Intelligence. In between my current chores, worries about PhD study and life in general I would probably never have noticed it as the entire publishing procedure more or less vanished from my mental focus as soon as I have the proofed the final version. However, I owe it to the young naive me from a few years back, only starting an academic career and dreaming of such a feat … to look back at the entire process and smile.
:)
It took some time, far too much time in fact, but now the paper is out … and it is time to move on … here is to many more!
So, here is the thing: my sister listed me on her blog and gave me a Leibster award which I am very thankful for even though I personally find a kind of a blog analogy of a chain letter. Here is the idea behind the award:
And here is a little about what the award is all about: This award was designed to be a blog award in the “pay it forward” fashion. Once you’ve been nominated, you award it to five blogs that you like that have fewer than 200 followers, to encourage new visitors to visit these blogs.
The rules for accepting this award are: You need to thank the person who gave you the award and link back to their blog. Post the award onto your blog. Give the award to five bloggers who you appreciate that have fewer than 200 followers. Leave a comment on their blog letting them know that you have given them this awesome award!
Next you need to PAY IT FORWARD!
Despite the fact that I am not very keen on these kind of practices, I find the idea perfect to revive my blog after its numerous migrations and a lot of neglect content-wise. So, without further ado I would like to thank my sister for the award and her encouragement to start posting stuff again. As for my awards I will not really follow the rule about 200 followers as I do not have this information. But the blogs that I follow or will follow are:
The revival of the blog will probably be slow, I fill first try to publish something every week week (more likely every two) and then I will see where it will go.
Gabe Newell
With the upgrade to the new online Moodle classroom at our faculty this school year I have become increasingly annoyed by the HTTPS-only access option with no redirect from plain HTTP links. Apparently a redirect is a security issue. Nevertheless I consider no-redirect a usability issue.
So, instead of arguing with the administrator, I have written a rule for HTTPS Everywhere Firefox addon that redirects automatically. Below is a ruleset that you can copy to Ucilnica.xml and save it to HTTPSEverywhereUserRules directory in your Firefox profile directory.
<ruleset name="FRI Ucilnica">
<target host="*.fri.uni-lj.si" />
<rule from="^http://(ucilnica[0-9]*)\.fri\.uni\-lj\.si/"
to="https://$1.fri.uni-lj.si/"/>
</ruleset>
Restart the Firefox and everything should work.
A while ago my WD MyBook Pro external hard drive that I have used for additional storage at work died. The disks (it is a 2TB unit with two 1TB disks) were fine, but the enclosure started misbehaving. Since I did not want to lose the data I have plugged the disks into a computer with a live CD Linux distro. The problem was that the disks were configured in RAID 0 with unknown parameters. The Internet gave no definite answer so I have played around with the mdadm tool (you probably have to install it on live distro) a bit and found out that the following set of commands works for WD MyBook Pro (RAID 0):
$ mknod /dev/md0 b 9 4
$ mdadm --build /dev/md0 -l0 -n2 -c256 /dev/sda /dev/sdb
In my case /dev/sda and /dev/sdb were the two disks. The order of the disks is important and I have determined it by examining the two disks on their own using Disk Utility. The first disk should contain a recognizable partition (ext2 in my case) while the second disk should contain only an unknown partition (in a way it makes sense) for RAID 0. After that I was able to mount the resulting device although the Disk Utility complained about the partition misalignment, which would be bad had I configured a production system … but I only wanted to rescue my data.
