pb's blog about life and IT

Friday, November 30, 2012

My fedora java 7 update

Today I've migrated to java 7 (Oracle version), as our development environment has been "java 7 enabled".

The transition was "rather" smooth, I just had to follow following description: http://fedorasolved.org/Members/zcat/using-sun-java-instead-of-openjdk

When I mentioned "rather", it was not due to description present on the mentioned site :). After I rebooted I could not login to my xfce (or gnome,...), login screen was shown again even after correct password input.

The problem was caused by my laziness. I didn't really read what I copy/pasted :)

After switching to text mode (Ctrl+Alt+F<X>) I've observed some EOF reported problem right after login. And my bash didn't behave :) I had to provide full path when invoking commands.

When thinking of what I've changed, file: /etc/profile.d/sunjava.sh started to be suspicious, after opening it via:

/usr/bin/sudo /usr/bin/vi /etc/profile.d/sunjava.sh

(yes, path didn't work for me somehow :)

my suspect changed to culprit when seeing the content:

export JAVA_HOME=/usr/java/default
export PATH=\$JAVA_HOME/bin:\$PATH
EOF

after fixing it to:

export JAVA_HOME=/usr/java/default
export PATH=$JAVA_HOME/bin:$PATH

and doing the reboot afterwards, via:

/usr/bin/sudo /usr/bin/reboot

all started to work as expected.

So it seems trivial things might become challenges in my hands :)
Now I see why I became a programmer rather than a doctor :)

Friday, November 23, 2012

Maven java.lang.ClassNotFoundException: org.codehaus.plexus.classworlds.launcher.Launcher

Have you ever faced exception like this?

Exception in thread "main" java.lang.NoClassDefFoundError: org/codehaus/plexus/classworlds/launcher/Launcher
Caused by: java.lang.ClassNotFoundException: org.codehaus.plexus.classworlds.launcher.Launcher
        at java.net.URLClassLoader$1.run(URLClassLoader.java:202)
        at java.security.AccessController.doPrivileged(Native Method)
        at java.net.URLClassLoader.findClass(URLClassLoader.java:190)
        at java.lang.ClassLoader.loadClass(ClassLoader.java:307)
        at sun.misc.Launcher$AppClassLoader.loadClass(Launcher.java:301)
        at java.lang.ClassLoader.loadClass(ClassLoader.java:248)
Could not find the main class: org.codehaus.plexus.classworlds.launcher.Launcher.  Program will exit.

My use case was that I had installed maven3 and afterwards tried to use maven2. All the environment variables updates have been setup in:

~/.bashrc

However once trying to use maven2 exception appeared (Even in the new terminal session).
Solution was found based on advice present in: http://cyntech.wordpress.com/2011/03/09/maven-2-error/

Running command:

which mvn

revealed the problem, as I still had in my $PATH the maven3 binary. Even if I appended the maven2, it was there on the later position => maven3 was to be used, but with changed $M2_HOME (reffering to maven 2).

Fixing the $PATH did the job for me.

This might not be your case, therefor as a general sum up of possible root causes, I recommend checking: http://stackoverflow.com/questions/6198677/java-lang-noclassdeffounderror-org-codehaus-plexus-classworlds-launcher-launche

Friday, November 16, 2012

Glassfish 3.1.1 OSE with Hibernate Validator 4.3.0.Final

Goal: Use hibernate validator 4.3.0.Final with Glassfish 3.1.1 OSE.

Lately I came across an issue, that I wasn't able to use the latest stable hibernate validator (hiberante validator 4.3.0.Final) with Glassfish 3.1.1 OSE.

Problem is that older version is beeing used as a glassfish module (glassfish/modules/bean-validator.jar). If you check the contents of the package (in MANIFEST.MF) you'd see that version 4.1.0.Final is beeing used.

OK, so how to deal with that? I've found out that I'm not the only one having issue here.
http://stackoverflow.com/questions/10548931/how-to-upgrade-the-hibernate-validator-4-3-0-final-to-the-glassfish-3-1-2

Suggestion gave me some motivation, as it seems to be possible to achieve :)

As I'm using ear, the simple solution provided in answer was not feasable for me (as it was a war case). OK, let's see how far we can get. There are multiple references (finally I ended up with the following 2 links):

I read the later one, as it seemd like a good idea to learn creating modules deployable to glassfish, however even if I created maven osgi bundle it neither contained all the dependencies required nor the descriptor seemed to be OK. Sounds like a wrong way for me then.

Going back to the 1.st link, I found following svn repository and tried to build things: https://svn.java.net/svn/glassfish~svn/trunk/external/source-build
However only once it has been achieved (and the way there was quite long) I realized that this just rebuilds the hibernate validator with it's dependencies, and that is of course not needed, as these are already built and available in public maven repositories :).

OK, going via next steps in the article, I ended up with the following solution:

checked out glassfish sources

svn co https://svn.java.net/svn/hk2~svn/branches/hk2-gf-3.1.1/

updated bean-validation related pom, see my diff:

Index: pom.xml
===================================================================
--- pom.xml (revision 4105)
+++ pom.xml (working copy)
@@ -54,9 +54,10 @@
          maven-bundle-plugin and hk2-maven-plugin together -->
     <packaging>jar</packaging>
     <properties>
-      <hibernate-validator.version>4.2.0.Final</hibernate-validator.version>
+      <hibernate-validator.version>4.3.0.Final</hibernate-validator.version>
       <javax.validation.version>1.0</javax.validation.version>
       <slf4j.version>1.6.1</slf4j.version>
+      <jboss-logging.version>3.1.0.CR2</jboss-logging.version>
     </properties>
     <name>Validation API (JSR 303) version ${javax.validation.version}, Hibernate Validator version ${hibernate-validator.version} and its dependencies repackaged as OSGi bundle</name>
 
@@ -89,7 +90,7 @@
                         <Embed-Dependency>
                             <!-- Only specify root artifacts that need to be embedded, everything else
                                  will be pulled in automatically based on Private-Package settings. -->
-                            *; artifactId=hibernate-validator|slf4j-api|slf4j-jdk14; inline=true
+                            *; artifactId=hibernate-validator|jboss-logging|slf4j-api|slf4j-jdk14; inline=true
                         </Embed-Dependency>
                         <Export-Package>
                              <!-- 
@@ -105,7 +106,7 @@
 
                         <Private-Package>
                              <!-- Have a private copy of external non-standard dependencies -->
-                             org.slf4j.*; com.googlecode.jtype.*; org.joda.time.*; org.jsoup.*
+                             org.jboss.logging.*; org.slf4j.*; com.googlecode.jtype.*; org.joda.time.*; org.jsoup.*
                         </Private-Package>
 
                         <Import-Package>
@@ -117,6 +118,12 @@
                                which is a JPA 2 class.
                            -->
                            org.slf4j; org.slf4j.spi; org.slf4j.helpers; version=${slf4j.version}; resolution:=optional,
+                           org.jboss.logging; version=${jboss-logging.version}; resolution:=optional,
+                           org.apache.log4j;resolution:=optional,
+                           org.jboss.logmanager;resolution:=optional,
+                           com.ibm.uvm.tools;resolution:=optional,
+                           com.sun.jdmk.comm;resolution:=optional, 
+                           javax.jmdns;resolution:=optional,
                            javax.persistence.*; version="2.0"; resolution:=optional,
                            *
                        </Import-Package>
@@ -263,6 +270,13 @@
         </dependency>
 
 
+ <dependency>
+  <groupId>org.jboss.logging</groupId>
+  <artifactId>jboss-logging</artifactId>
+  <version>${jboss-logging.version}</version>
+            <optional>true</optional>
+ </dependency>
+
         <!-- We bundle jdk binding inside this OSGi bundle -->
         <dependency>
            <groupId>org.slf4j</groupId>

built:

cd hk2-gf-3.1.1/external/bean-validator
mvn clean install

copied to glassfish modules and removed old one:

rm ${GLASSFISH_HOME}/glassfish/modules/bean-validator.jar
cp hk2-gf-3.1.1/external/bean-validator/target/bean-validator-1.1.15-SNAPSHOT.jar ${GLASSFISH_HOME}/glassfish/modules/

done :)

Issues? well, we'll see later, during testing :) But very first try worked perfectly.

Comments? Feel free to share, as I'm not the osgi expert (yet :)), so some of my updates might ... let's say won't make too much sense.

Monday, October 22, 2012

Solving Eclipse 4.2 unresponsive UI problems

Q: How to fix Eclipse 4.2 unresponsive UI problems?
A: Reserving more memory for it.

Now the long answer comes:
While using Eclipse 4.2 on Fedora 17 I got into problems, that opening Find dialog within the file took around 5-10 seconds.
That is crazy and not the way the things can work to be productive.

However following forum entry gave some hope to me:
http://www.eclipse.org/forums/index.php/mv/msg/367243/896690/#msg_896690
And after applying the suggestion ecilpse came "back to life" :)

So these are the args that worked for me:

-vmargs
-Xms512m
-Xmx1024m

Sunday, September 16, 2012

Generating graphical representation of XSD (XML schema).

Question: Have you ever seen the graphical representaiton of XSD in a way XMLSpy is capable of generating?

Did you ever need to generate similar output via free/open source tools?

Answer: XSD Diagram (http://regis.cosnier.free.fr/?page=XSDDiagram).

When searching for the alternative couple years ago, that was my choice and I didn't regret since then,

Usage is very simple, let's see how works. I'll go step by step in a usage. For a real world example I use as a sample for the xsd the following one (standard web application descriptor schema): http://java.sun.com/xml/ns/javaee/web-app_3_0.xsd

Start XSDDiagram.exe and choose File -> Open (open the downloaded xsd file)
Error is thrown: "Could not find file 'dir/web-common_3_0.xsd'."
OK, so we need reffered xsds as well. Let's get the rest then (web-common_3_0.xsd, javaee_6.xsd, jsp_2_2.xsd and javaee_web_services_client_1_3.xsd).
Copy them to the same dir and reopen the original xsd again.
Click the button in toolbar "Add all toplevel elements"

or just choose the element you are interested in and click "Add selected toplevel element"
Click multiple times (till all the levels are expanded) the button in toolbar "Expand one level"

or just click the + sign for those elements you're interested in.
Choose File -> Export Diagram -> (choose the file type feasable for your use case, I chose svg) and save.
You're done!

Output could look like this (please note that schema graphical representation in this case is huge):

Monday, July 30, 2012

fixing incorrect dates in JPEGs on (x)ubuntu

Have you ever faced the problem of incorrectly set date in your camera?
I have, and realized it only after copying data to my laptop.

In my case year was set to previous one. For the solution description, please keep this in mind as I fix the date shift only, however the tools used could fix any other incorrect date setup as well.

Fixing it for the future pics is easy, just update your settings in camera, however how to fix it for the already created pics?
There are 2 places where date update is required:
- fix file properties as well as
- fix EXIF data.

Fixing file date
In linux it's easy. My case was I set up last year instead of this one => after having all the jpegs in one dir, I just ran:

touch -d '+0 year' *

(solution ideas were inspired by: http://superuser.com/questions/122863/linux-shell-change-a-files-modify-timestamp-relatively)

How to fix EXIF data?
Exiftool (http://www.sno.phy.queensu.ca/~phil/exiftool/) seems to do the job easily. Date shift feature using exiftool is described in more detail on: http://www.sno.phy.queensu.ca/~phil/exiftool/Shift.html

I Instaled it first:

sudo apt-get install libimage-exiftool-perl

Moreover there are multiple dates in EXIF, to list them I ran this on the particular file (using exiftool):

exiftool -k IMGP0184.JPG | grep Date
File Modification Date/Time     : 2012:07:30 22:27:57+02:00
Modify Date                     : 2011:07:22 12:48:12
Date/Time Original              : 2011:07:22 12:48:12
Create Date                     : 2011:07:22 12:48:12
Date                            : 2011:07:22
Manufacture Date                : 2011:10:07
-- press RETURN --

To fix EXIF dates on all the relevant files I ran:

exiftool "-AllDates+=1:0:0 0:0:0" *.JPG

That should be it. Let's check the result rerunning the EXIF data listing command:

exiftool -k IMGP0184.JPG | grep Date
File Modification Date/Time     : 2012:07:30 22:28:16+02:00
Modify Date                     : 2012:07:22 12:48:12
Date/Time Original              : 2012:07:22 12:48:12
Create Date                     : 2012:07:22 12:48:12
Date                            : 2011:07:22
Manufacture Date                : 2011:10:07
-- press RETURN --

OK, that worked for me :) Hope it helps you as well.

One more thing, to cleanup backup files created by exiftool I ran following:

rm *.JPG_original

Monday, July 2, 2012

Panorama photo creation in an easy (lazy) way

Q: Have you ever faced the challenge of creating panorama from the set of photos? Did you try to create it manually using your favorite graphics editor? Did you face problem of spending too much time on that?

A: There might be a solution for you. It's called Hugin.

As the homepage (http://hugin.sourceforge.net/) says: "With Hugin you can assemble a mosaic of photographs into a complete immersive panorama, stitch any series of overlapping pictures and much more."

As I'm used to run my favorite programs on Linux as well as Windows, I like the fact that it's capable of running on both, moreover it's open source (which I'm really excited for).

OK, so how it works for me? Maybe some motivating/demotivating pics in the beginning.
Shots show the views of Ramadevara Betta (Hill) in India I took

Still interested? Let's go then...

Adventure part
Take photos of your interest in a sequence.
My rules follow:
- take at least 2 rows one above each other (make the left-to-right / right-to-left sequence)
- if required, reorder pics after downloading them from camera
- make sure the pics overlap (to be able to identify conneciton points there), rule might be to overlap like 1/4 or up to 1/3 or so

Boring part
1. make sure you have Hugin installed and start it
2. click "1. Load Images..." button
3. click "2. Align..." button,
4. then be veeery patient, depending on the pics quality and count it might take quite some time
5. after done, "Fast panorama preview is shown" where you can do some final addaptions on the panorama (I usually don't need to do here anything at all)
6. when done close it and click "3. Create panorama..." button
7. when done, you should have your panorama present in tif file
8. that's it

Problem solving
If you're unlucky you might experience the problem that process fails during point 4. No controll points are found and you get error in the aligning console saying: "An error happened while loading image : caught exception: bad allocation"
It seems I was not the only one facing the issue and finally solution has been found:
https://bugs.launchpad.net/hugin/+bug/717944
https://groups.google.com/forum/#!msg/hugin-ptx/HPt0lIFFRr0/3fwBzt-EUJMJ

Problem, based on the discussions, seems to be that cpfind process, one that tries to find the controll points crashes due to problems with memory management.
OK, for me however the fix was, to slow the process a bit down (as by default it seems to be running on 8 parallel threads) and use less threads (also less memory) for analysis.
I anyway preffer the idea of someone else doing the boring job (connecting the pics in this case) rather than me, regardless how long it will take him :)

All right, but how to decrease thread count?
1. go to menu "File" -> "Preferences"
2. go to "Control points detector" tab
3. choose "Hugin CPFind (Default)" and click "Edit..." button
4. update "Arguments" from "--multirow -o %o %s" to "--ncores=1 --multirow -o %o %s"
5. click "OK" buttons
6. I also restarted Hugin (not sure if required)

I tried with switch "--ncores=4" first, but that didn't work for me so I changed to "--ncores=1" and it finally solved/workarounded the problem

What's next?
If interested, give it a try and see.
Feel free to link your panororamas/experiences on creating these in comments :)