binfalse

Today I had a conversation with a scientist from Bulgaria who is working with microarrays. He told me some practical experiences of his work. It was very interesting and I learned a lot of things, in spite of the fact that I gave a lecture about microarrays some time ago.

In this talk he said a wonderful sentence:

Früher dachte ich immer die Russen wären dumm, bis ich die Amerikaner kennen gelernt habe!

English translation: Some years ago I thought the Russians are stupid, until I got to know the Americans.

Topic was the structuring of websites of companies. If he has a question he always has to search through the web because everyone tells him the answer is anywhere in there! affymetrix for example has thousands of user manuals, the intersection of all of these papers is very small, but one paper has hundreds of pages… And I think he is totally right. The arrangement of information today is very terrible, to find what you are searching about is some kind of art! But he doesn’t mince matters. I really like Eastern Europeans ;)

He invented me to his lab tomorrow so I can see how this affymetrix machinery produces the data that I get to analyze.

Oh no, not that kind of quickie you might think about! Rumpel an me decided more or less spontaneously to go to Bonn and visit one of our former employee Martin and additionally take a little look at SIGINT in Cologne.

So we rent a car at Sixt on Friday morning and met Martin at 5 pm in his flat. Of course our trip was very analog, we didn’t have any navigation device, just printed a route calculated by Google maps and rely to male instinct on the way through Germany and the high traffic in Ruhr Valley at Friday afternoon before holiday… What should I say, of course everything went totally well and we had a lot of fun in our little car! You can see some pictures at picasa.

Quickie through Germany

Of course it was a great weekend! We’ve seen a lot of fascination places of Bonn and Cologne like Cologne cathedral, big ships on Rhine or Media Center in Cologne. The events at SIGINT were also interesting, where it cannot be compared with the Chaos Communcation Congress in Berlin. In Cologne you’ll always get a chair and the queues are very short. Nevertheless the topics are of high quality.

All in all it was an excellent trip, even it was very expensive.

One of the people that are working with me on some crazy stuff always forgets to pull the newest revision of the repository before changing the content and so he has very often trouble with different versions when he decides to push his work to the master repository. His actual workaround is to check out the complete repository in a new directory and merge his changes by hand into this revision… Here is a little instruction to maximize his productivity and minimize the network traffic.

Lets assume we have a repository, created like this:

/tmp % git init --bare root

And we have one user, that clones this new repository and inits:

/tmp % git clone root slave1
/tmp % cd slave1
/tmp/slave1 (git)-[master] % echo "line1\\nline2" >> testfile
/tmp/slave1 (git)-[master] % cat testfile
line1
line2
/tmp/slave1 (git)-[master] % git add .
/tmp/slave1 (git)-[master] % git commit -m "init"
[master (root-commit) bc7e4da] init
 1 files changed, 2 insertions(+), 0 deletions(-)
 create mode 100644 testfile
/tmp/slave1 (git)-[master] % git push ../root master

So we have some content in our root repo. Another user (our bad guy) clones that repository too:

/tmp % git clone root slave2

So let a bit of time elapse, while user one is changing the root repository so that the testfile may look like this:

/tmp/slave1 (git)-[master] % cat testfile | sed 's/line1/&\\nline1a/' > testfile.tmp && mv testfile.tmp testfile
/tmp/slave1 (git)-[master] % cat testfile
line1
line1a
line2

And of course, the changer commits his changes:

/tmp/slave1 (git)-[master] % git commit -a -m "haha, root has changed..."
[master e18f637] haha, root has changed...
 1 files changed, 1 insertions(+), 0 deletions(-)
/tmp/slave1 (git)-[master] % git push ../root master
Counting objects: 5, done.
Writing objects: 100% (3/3), 265 bytes, done.
Total 3 (delta 0), reused 0 (delta 0)
Unpacking objects: 100% (3/3), done.
To ../root
   bc7e4da..a04d363  master -> master

Ok, nothing bad happened, but now our special friend decides to work:

/tmp/slave2 (git)-[master] % cat testfile | sed 's/line1/&\\nline1b/' > testfile.tmp && mv testfile.tmp testfile
/tmp/slave2 (git)-[master] % cat testfile
line1
line1b
line2
/tmp/slave2 (git)-[master] % git commit -a -m "oops, i am very stupid..."
[master d691ada] oops, i am very stupid...
 1 files changed, 1 insertions(+), 0 deletions(-)

What do you think will happen if he tries to push his changes to the master repo? Your right, nothing but a error:

/tmp/slave2 (git)-[master] % git push ../root master
To ../root
 ! [rejected]        master -> master (non-fast-forward)
error: failed to push some refs to '../root'
To prevent you from losing history, non-fast-forward updates were rejected
Merge the remote changes before pushing again.  See the 'Note about
fast-forwards' section of 'git push --help' for details.

Mmmh, so lets try to pull the root repo:

/tmp/slave2 (git)-[master] % git pull ../root master
remote: Counting objects: 5, done.
remote: Total 3 (delta 0), reused 0 (delta 0)
Unpacking objects: 100% (3/3), done.
From ../root
 * branch            master     -> FETCH_HEAD
Auto-merging testfile
CONFLICT (content): Merge conflict in testfile
Automatic merge failed; fix conflicts and then commit the result.

Our friend would now check out the whole repository and insert his changes by hand, but whats the better solution? Merging the file! Git has a function called mergetool , you can merge the conflicts with a program of your choice. Some examples are vimdiff , xxdiff , emerge or also for GUI lovers kdiff3 . In this post I’ll use vimdiff :

/tmp/slave2 (git)-[master|merge] % git mergetool --tool=vimdiff testfile

Normal merge conflict for 'testfile':
  {local}: modified
  {remote}: modified
Hit return to start merge resolution tool (vimdiff): 
3 files to edit

So change the conflicting file(s), you will also see the changes made in root’s and in your local revision. If you’re done just save it and commit your merge:

/tmp/slave2 (git)-[master|merge] % git commit -m "merged"
[master 6be1482] merged

Great, now there is nothing that prevents you from pushing your changes to the root repository:

/tmp/slave2 (git)-[master] % git push ../root master
Counting objects: 10, done.
Delta compression using up to 2 threads.
Compressing objects: 100% (2/2), done.
Writing objects: 100% (6/6), 555 bytes, done.
Total 6 (delta 0), reused 0 (delta 0)
Unpacking objects: 100% (6/6), done.
To ../root
   a04d363..6be1482  master -> master

I think this way of solving such conflicts maybe much more efficient than cloning the whole repository again and again and again ;)

One week passed since I got a package of Spots, this weekend I found some time to hack a little bit with this funny components.

First of all I programmed a tool that visualizes the Spots movement in an OpenGL frame that draws a virtual Spot. Nice for demonstrations, but nothing spectacular.

After that I developed a little mouse emulator, that translates Spot movement to mouse motions on the screen. Here the Spot isn’t doing anything intelligent, it only sends its tilt status every 25 ms as well as switch events to broadcast. Another Spot, working as basestation connected to my machine, is listening to this talking Spot and my host analyzes the received values. To move the mouse on the screen or to generate a click I use the Robot class of the Java AWT package. Long story short, a video may explain it more understandable (via YouTube):

I will continue with working on these libraries before I publish them in another post. So look forward to the release ;-)

The unstable release of Debian is of course tricky in a lot of cases, so there is also a little stumbling stone on your path of Java network programming. On every new system it annoys me.

Before I wrongful blame my preferred Debian release called Sid I have to acknowledge I don’t know whether this feature is also available in other releases… Here is a small program to test/reproduce:

import java.net.URL;
import java.io.BufferedReader;
import java.io.InputStreamReader;

public class WebReader
{
	public static void main (String[] args)
	throws Exception
	{
		BufferedReader reader = new BufferedReader(
			new InputStreamReader (new URL (args[0]).openStream ()));
		String line = reader.readLine ();
		
		while ((line = reader.readLine ()) != null)
			System.out.println (line);
	}
}

Compilation shouldn’t fail, but if you try to launch it you’ll get an exception like that:

Exception in thread "main" java.net.SocketException: Network is unreachable
        at java.net.PlainSocketImpl.socketConnect(Native Method)
        at java.net.PlainSocketImpl.doConnect(PlainSocketImpl.java:333)
        at java.net.PlainSocketImpl.connectToAddress(PlainSocketImpl.java:195)
        at java.net.PlainSocketImpl.connect(PlainSocketImpl.java:182)
        at java.net.SocksSocketImpl.connect(SocksSocketImpl.java:366)
        at java.net.Socket.connect(Socket.java:525)
        at java.net.Socket.connect(Socket.java:475)
        at sun.net.NetworkClient.doConnect(NetworkClient.java:163)
        at sun.net.www.http.HttpClient.openServer(HttpClient.java:394)
        at sun.net.www.http.HttpClient.openServer(HttpClient.java:529)
        at sun.net.www.http.HttpClient.<init>(HttpClient.java:233)
        at sun.net.www.http.HttpClient.New(HttpClient.java:306)
        at sun.net.www.http.HttpClient.New(HttpClient.java:323)
        at sun.net.www.protocol.http.HttpURLConnection.getNewHttpClient(HttpURLConnection.java:860)
        at sun.net.www.protocol.http.HttpURLConnection.plainConnect(HttpURLConnection.java:801)
        at sun.net.www.protocol.http.HttpURLConnection.connect(HttpURLConnection.java:726)
        at sun.net.www.protocol.http.HttpURLConnection.getInputStream(HttpURLConnection.java:1049)
        at java.net.URL.openStream(URL.java:1010)
        at WebReader.main(WebReader.java:10)

This is caused by one little line in /etc/sysctl.d/bindv6only.conf saying you want to explicitly bind via IPv6. But my connection (maybe yours too) communicates still over IPv4, so this method of networking of course fails. To change this behavior you have to choose between two solutions.

Solution 1: Permanent modification (needs to be root)

You can change this behavior for the whole system by editing the file /etc/sysctl.d/bindv6only.conf :

# original only IPv6
# net.ipv6.bindv6only = 1
net.ipv6.bindv6only = 0

After that just type invoke-rc.d procps restart in your terminal to let your changes take effect. Your next run should work fine.

Solution 2: Change it for this single example

If your are not allowed to change system settings, you can add -Djava.net.preferIPv4Stack=true to your execution command:

java -Djava.net.preferIPv4Stack=true  WebReader http://localhost
# instead of `java WebReader http://localhost`

This causes your actual runtime to connect the network via IPv4, no matter to system preferences. I hope this could save some time of developers like me ;-)