Tamás Tarjányi's Blog

A bad weekend of Apple, Ubuntu and me

2011-10-15T11:22:00.000+02:00

Over the weekend I have had 2 technical problems. And a third one which was the emotional consequence of the 2 technical ones.

So the first problem was a result of iOS5 update. My iTouch 4th gen became unstable. I liked my iTouch, iPad and iXXXXX because they were always responsive and reliable. No lags no glitches no errors. But after I have installed iOS5 on my iTouch. My oh my!!! Lags are there and makes me feel like on Windows.
The other issue seems to be a bug. You can test it by yourself.
Start "App Store" - Select "Updates" - Select "Purchased" - Select "Not on this iPod" - Swipe right to left on any item and then select the "Hide" button.
In my case "App Store" always does exit.
Steve Jobs have died just one week before the release of iOS5 and here we are. Everything I really like for iOS devices have gone. Think about it.

The other problem is more serious. And to be honest more depressing to me.
I have upgraded my Ubuntu 11.04 to Ubuntu 11.10 on my "office" notebook.
And for first look I hit bug "Booting system without full network configuration..."
What does it mean you may ask? A normal user who runs into this bug can not work anymore on the machine. So this is THE experience a normal user can not tolerate and will change to something else. Good question what the next choice will be. Back to windows?

To me the conclusion was that instead of companies making products more stable - Apple and Linux like - Apple and Linux are making themselves more of a Microsoft like. And sorry to say but that is really-really depressing me.

I beg you Apple and Ubuntu to follow your own path and forget the Microsoft one.

Why ec2-ami-tools - ec2-upload-bundle - hangs

2011-06-10T00:57:00.005+02:00

Me and my colleagues had been fighting an issue for some days. Since it made me really-really f**** upset I decided to investigate it now.

The issue: is that if you are behind a proxy (not necessarily) - http or socks does not matter much - you might have problem with ec2-ami-tools. When you start ec2-upload-bundle it does nothing just looks like hanging. We had been waiting for some minutes but nothing happened. Actually it is trying to do something but can not start upload.
In contrast on my home network ec2-upload-bundle almost immediately starts uploading and showing a message to indicate it.
So what is wrong? Why this is happening from behind a proxy? Is it proxy related?

We have tested many possible workarounds without any luck.
First we have specified http_proxy and https_proxy of course but it did not work.
Then we also tried to socksify the process. Without any luck.
At that point we decided to eliminate the proxies. We got an access to a machine in DMZ and we made a dynamic port forwarding and tried to use that as a socks proxy server. And that was really surprising because even that did not work.
Everything else - Firefox, ssh - was working fine but not ec2-ami-tools.
Here we ran out of the ideas for a while.

But of course problem happens to be solved.

So I made a tcpdump, had a look and saw that the process is trying to access a special IP address 169.254.169.254 . This is a link-local IP address and this is used by amazon within the cloud to fetch some instance specific data. It is unclear why ec2-upload-bundle is trying to reach it. It is much more unclear how the program is written to be able to hang on it. I mean it is also visible that it makes a retry before the TCP timeout (0,3,9 sec) but than it gives up and hangs.

Workaround: to the problem can be to configure this 169.254.169.254 IP on the ssh server or http/socks proxy server. Also you can try somehow locally - with a firewall - reject these connections. But you have to prevent the connection to timeout since if the connection times out ec2-upload-bundle will hang. See the note for an alternative workaround.

Note: This is working on my home machine because normally my Ubuntu Linux has a 169.254.0.0/16 route. And this is enough to prevent the timeout. But only if you are not using proxy. If proxy is in between than your own routing is not used to reach this IP but the route of the proxy is used. So you have to add this route to the proxy/ssh server if you can.

route add -net 169.254.0.0 netmask 255.255.0.0 dev eth0 # Replace eth0 with your interface.

Comment: I have less motivation now to investigate it further but probably I will check what is wrong in the code and why ec2-upload-bundle hangs in such a case.

Update 1: I have checked the source code of ec2-ami-tools and it seems the error is in instance-data.rb. The initialize method is checking if the meta data is available or not and set @instance_data_accessible according. But there is no timeout defined here so you have to wait ... It seems even the TCP timeout is not realized by the open method. A Timeout::timeout(10) { } can solve this error.

Update 2: Later in instance-data.rb the read_meta_data(path) method should call the open() method only if @instance_data_accessible is true. But this is not the case. I do not know ruby at all but this seems to be a coding error. I have just added some print command to see what was going there but do not understand the code really. Besides the open() methods here also hanging indefinitely. So this is a complete hang here.

How to create a custom RedHat / CentOS Amazon EC2 / Cloud image

2011-04-30T20:35:00.000+02:00

Note: Cloud computing is currently a very dynamic segment of IT and relevant changes - regarding technologies, available features and licensing - might happen anytime.

First of all you may ask why to build your own AMI when you can find many existing - CentOS and other free - Amazon EC2 images.

The following sentences are taken from an official amazon document.

"You launch AMIs at your own risk. Amazon cannot vouch for the integrity or security of AMIs shared by other EC2 users. Therefore, you should treat shared AMIs as you would any foreign code that you might consider deploying in your own data center and perform the appropriate due diligence.

Ideally, you should get the AMI ID from a trusted source (a web site, another EC2 user, etc). If you do not know the source of an AMI, we recommend that you search the forums for comments on the AMI before launching it. Conversely, if you have questions or observations about a shared AMI, feel free to use the AWS forums to ask or comment."

So what if you do not trust those public AMIs? And where are the official RedHat images? (Old document is here)

Update: When I started to write this post - 2011.04 - RedHat did not have any available public AMIs in amazon. Besides that RedHat licensing was really-really restrictive and using your own image required special RedHat subscription and transferring 25 subscriptions to the cloud for at least half a year (at least as much as I understood the license).

Some historical documents:
~~RedHat Cloud Computin~~(Had been removed)
RedHat Cloud FAQ (Archived)

Licensing in the cloud is another general problem. Amazon released EC2 beta in 2006 and went public in Q4 2008. Although 3-5 years have gone OEM licensing is still a very big problem in most cases. OEMs are trying - or seeing the results probably not really - to adapt licensing and changing the cloud (amazon) licensing very frequently. Sometimes releasing paid AMIs and covering license fees such a way, sometimes requires special licenses and sometimes allows you to use your existing licenses. So probably if you can not find something now or license is not available currently, it will be available in a month, in week or even in a day.

Update: RedHat has released some new payed AMIs. For some additional hourly fee you can use these instances and you will not have licensing problem. Search for the owner 309956199498 in the list of available AMIs and you can see RedHat AMIs. 

See RedHat pricing here and standard pricing here for comparison.

Other historical RedHat announcements:
RedHat 5.5 32/64
RedHat 5.6 32/64
RedHat 6.0 32/64
RedHat 6.1 32/64
Common RedHat AMI updates
RedHat on Amazon EC2

So if you still want to build your own image - because of security, licensing or because you have no time to wait till it is officially released by the OEM - amazon has a GOOD description how to build your own image.

BUILDING YOUR OWN IMAGE IS ALWAYS A RISK FROM LICENSING POINT OF VIEW!!!

MAKE SURE YOU HAVE THE PROPER LICENSES BEFORE USING YOUR OWN IMAGE!!!

OR BETTER USE FREE SOFTWARES IN THE CLOUD.
(Debian, Ubuntu, CentOS)

I can only add some probably hopefully useful information.

If you want to create CentOS image via yum you have to use a pre installed CentOS operating system as a build server. (CentOS is free to use in the cloud and have a central repo even for install and update.)

If you want to create RedHat image via yum you have to use a pre installed RedHat operating system as a build server. (Licensing of RedHat might be a problem. And upgrading from the cloud via RedHat Network [RHN] is probably not the best idea.)

I recommend to use a 64bit OS version as a build platform since you can use that to build both 32 and 64 bit images. On a 32bit OS however you can build only 32bit images.

I also recommend to use the latest version of the OS. It is possible to install older version of the OS on the latest version. (It is most probably possible to install latest version of the OS on an older version but might have some unavailable new features - for example missing file system support.)

I prefer to use original kernels from RedHat and CentOS and not using the available amazon AKIs.

You can make your life easier by disabling SELinux - or at least running in permissive mode. After you could build a running image you can try to enable SELinux again if you need. (And I would suggest to use it if possible.)

When you are installing the base OS with yum the amazon document shows how to create a yum configuration file by hand. CentOS has a so called "release" file which contains the yum config. Using that it is more automatic and version safe. RedHat also has a release file but yum config is needed there.

Eucalyptus problem on Ubuntu 10.04 (Lucid)

2010-05-10T20:16:00.000+02:00

Ubuntu 10.04 (Lucid) is out for a while and I wanted to try to install a private UEC on a previously installed Ubuntu 10.04 system only for testing something. I did not want to use the CDInstall method since I like to do the magic not just watching that. So I have decided to install from packages. (Ok you are right I should have tried to do it from Eucalyptus source but I did not have enough time.)
Unfortunately it was not straightforward. During installation of eucalyptus-cloud package I got an error.

...
Setting up eucalyptus-cloud (1.6.2-0ubuntu30.2) ...
restart: Unknown instance:
(Than waiting long....~60 sec)
eucalyptus-cloud stop/post-start, (post-start) process 26868
eucalyptus-cloud-publication start/running, process 27475
...

Hmmm. I have checked whether it is an install issue only or also a runtime problem. Checked the configuration files but they were fine. Eucalyptus start is slow and finally failed without visible error message. The cluster controller web interfaces is not working and service can not be started successfully.
Also stopping the service dropped an error.

root@ubuntu:~# stop eucalyptus-cloud
stop: Unknown instance:

I wanted to check the init scripts but here I faced the problem that SysVinit had been replaced by UpStart which I did not know. So made a trip in that area to get to understand. To be honest it is not a well documented thing and the concept is a bit different than SysVinit. But for a reason.
But back to the track I have realized that the the eucalyptus startup died very early and the command failing is executing eucalyptus-cloud binary. So started it manually and saw that it drops the below error.

root@ubuntu:~# /usr/sbin/eucalyptus-cloud
[error:0185] User 'eucalyptus' validated
[error:0077] Path /usr/lib/jvm/java-6-openjdk is not a directory
[error:0323] Cannot locate Java Home

After I have seen this it was pretty straightforward to solve. I did not have openjdk but I had sun-java6-jdk. I had it from the partner repository anyway. ;) The problem here is that although eucalyptus-cloud depends on some java package the proper one was not installed. So I have installed openjdk6 by hand and than everything was working fine.

root@ubuntu:~# aptitude install openjdk-6-jre-headless

If eucalyptus does not work even after openjdk6 install you can reinstall eucalyptus-cloud after or together with openjdk6.

root@ubuntu:~# aptitude purge eucalyptus-cloud eucalyptus-cc eucalyptus-sc eucalyptus-walrus
root@ubuntu:~# aptitude install openjdk-6-jre-headless eucalyptus-cloud eucalyptus-cc eucalyptus-sc
eucalyptus-walrus

In theory this works with sun's java also but than you have to define JAVA_HOME in advance starting eucalyptus-cloud and than can try to use sun (oracle) java version but openjdk is working fine without any problem.
Some useful links if cloud is up:

HybridFox (My prefered one)

ElasticFox config (Note: Since version 1.7 it does not work with Eucalyptus. But hope it will be corrected soon. In Eucalyptus.)

NetBeans Maven Wep Application Internationalization

2009-08-31T22:17:00.009+02:00

As you know NetBeans 6.7 supports Maven projects. And you must also know what Maven is. Both are great products. Especially together. With IceFaces.
If you know both Maven and NetBeans you have most probably already tried to create a web application - a facelets based one - in NetBeans using Maven webapp archetype. This is what I tried together with IceFaces. ( By the way here is my icefaces-webapp-archetype if you have no other. )
But that is enough of free advertisement of REALLY GREAT PRODUCTS. ;)

My point is now internationalization of a Facelets based web application under a Maven based NetBeans project. As hopefully you know the standard/one way of JSF/Facelets internationalization is using f:loadBundle - or in case of IceFaces the better ice:loadBundle tag. But the location of the properties file is a bit tricky with maven based on my experience. Previously this must be located in the same directory as the java sources. But things are changing over time? Forget the old ones and learn new ones. The new location is "Other Sources->src/main/resources" directory. If you build a war with Maven then you can find the properties file in WEB-INF/classes and this is the right root directory. So do not put properties file under "Source Packages" anymore if you are using Maven and NetBeans together since Maven does not take those from there as I have seen many time.

( Versions: NetBeans 6.7.1 , Maven 2.2.1 )

Create CVE links on Oracle CPU Page

2009-05-09T01:20:00.013+02:00

As you probably know oracle releases CPUs (Critical Patch Updates) in every 3 month now. If you have ever read any of these CPU Advisories you know it does not make too much sense to read these. But they exist. Since CPUJul2008 Oracle replaced its internal numbering (like DBnn) with CVE (Common Vulnerabilities and Exposures) numbering. Does it make any sense? In my opinion it would but it does not in this way. These CVEs does NOT contain any really useful information. The CVE database is an open database anybody can access. If they are using CVE numbers why Oracle not creating links from the Advisory page to the CVE pages? Would not it be more comfortable just to click on a link if you are interested in details instead of searching for CVE numbers manually? I think the missing link is the confession of how useless this numbering is in case of Oracle. If you read a CPU Advisory released after Jul 2008 you will find CVE numbers before every identified security bug but if you want to check the CVE you have to find it by hand and you will find no more info even on that page. So it does not make much sense because the CVE has NO useful info in it but here is a VERY SIMPLE GreaseMonkey script which inserts two links after the CVE you can click on and immediately check the "details" of the CVE.
But I repeat the CVE links are not in the CPU Advisory because the CVE does NOT contain any relevant info about the security problem.
I hope it will change in the near future and CVE or CPU will give us some useful detail.


// ==UserScript==
// @name          Oracle CPU Risk Matrix CVE- Link Creator
// @namespace     http://tamastarjanyi.blogspot.com/
// @description   Replaces simple CVE text on Oracle CPU pages (Like http://www.oracle.com/technology/deploy/security/critical-patch-updates/cpuapr2009.html ) to links
// @include       http://*.oracle.com/*
// ==/UserScript==


if (!GM_xmlhttpRequest) {
alert('Please upgrade to the latest version of Greasemonkey.');
}
var  tds=document.getElementsByTagName("td");
var cve=false;
for (i in tds){
if (tds[i].innerHTML.match("CVE-[0-9][0-9][0-9][0-9]-") ){
  cve=true;
  var cvetext=tds[i].innerHTML;
  tds[i].innerHTML=cvetext+" (<a target=_blank href=http://cve.mitre.org/cgi-bin/cvename.cgi?name="+cvetext+">mitre</a> | <a target=_blank href=http://nvd.nist.gov/nvd.cfm?cvename="+cvetext+">nvd</a> )";
}
}

Hardy Intrepid Update

2008-10-21T22:13:00.004+02:00

Short note: Wanted to make a distribution update from Hardy to Intrepid.
I made a sudo bash and executed update-manager -d from this terminal but this did not show the new version.
Then I have started it with gksu - press ALT+F2 and type in "update-manager -d" - and it was working. This must be a sudo environment issue.
So update-manager -d does not show the new version if started from a sudo session but works fine via gksu or su.

Glassfish Custom Login Module and Realm Configuration

2008-10-15T13:52:00.010+02:00

I wanted to create a custom login module and realm for Glassfish to implement a new authentication and authorization. I have taken the sun document but I "missed" - in fact parts of the info was there but hidden very well - some information and decided to share those here. I am focusing only on the missing parts. Rest is written in the document.

The first thing is point 3. To me it did not work to place class files into directory appserver-domain-dir/lib/classes. I created a jar file from my classes and put that into glassfish/lib and it worked after a server restart.

The next info what I missed is what jaas-context-name should be in login.conf and how it works later. So the jaas-context-name can be anything you want BUT you have to pass it to your Realm module as a parameter. Lets say you call you jaas-context-name as "customRealm".
So you have in login.conf:


customRealm{
org.CustomLogin  required;
}

Then when you create your realm via the gui or any other way you must specify this as a parameter of the realm. The name of the parameter MUST BE jaas-context. To be precise it depends on the code. With the given code this is true. Otherwise depends on the following two lines:


       String jaasCtx = props.getProperty(IASRealm.JAAS_CONTEXT_PARAM);
     this.setProperty(IASRealm.JAAS_CONTEXT_PARAM, jaasCtx);

And JAAS_CONTEXT_PARAM inherited from super has a value "jaas-context".

Now you have a mapping between realm and login module.

But how to use the realm from NetBeans?
If you create a Web Application or Enterprise application the web part has a web.xml and a sun-web.xml file in the "Configuration Files" directory. Open web.xml and select security part. Here open the Login Configuration and if you select Basic the Realm Name: must be the name of the realm created before. This is not the jaas-context name. You have to create security roles and security constraints what I will explain later.
So now you have a realm which is using basic authentication and the Realm Name is set to custom-realm.
The result in web.xml is:


<login-config>
   <auth-method>BASIC</auth-method>
   <realm-name>custom-realm</realm-name>
</login-config>

This realm has a parameter jaas-context which specifies which login module must be used.
The realm and the login module receives the user name and password typed in and returns the list of user groups to the container.
But what to return and how to specify who has rights to do what?
In the web.xml you specified you want to use a realm but under that in Netbeans you can specify which Role has access to which url via which way. You have to create security roles. This is like a group BUT this is not necessarily the group returned by the Realm Module. This is an internal application level group. If you create something like user and admin it is a good basis.
Below that you have to specify security constraints. This is where you can specify which URL is accessible to which Role in which way. Take a look and you will understand. (When you specify the url you do not have to type the application name only the path under the context. For example if you have a jsp page as server:8080/Application/faces/user/Index.jsp you have to type in only /faces/user/* and you have to miss Application.)
Here is an example constraint:


  <security-constraint>
      <display-name>Application User</display-name>
      <web-resource-collection>
          <web-resource-name>Application User Pages</web-resource-name>
          <description/>
          <url-pattern>/faces/user/*</url-pattern>
          <http-method>GET</http-method>
          <http-method>POST</http-method>
          <http-method>HEAD</http-method>
          <http-method>PUT</http-method>
          <http-method>OPTIONS</http-method>
          <http-method>TRACE</http-method>
          <http-method>DELETE</http-method>
      </web-resource-collection>
      <auth-constraint>
          <description/>
          <role-name>Application-user</role-name>
          </auth-constraint>
      <user-data-constraint>
          <description/>
          <transport-guarantee>CONFIDENTIAL</transport-guarantee>
      </user-data-constraint>
      </security-constraint>

One missing thing is how to map Authentication Groups to Application Roles. If the Realm returns the same name as group as an existing role this is mapped automatically but this is not the usual case. If user is for example in Administrator group and application has an Admin role you need a mapping between these. You can specify these in sun-web.xml Open and do it. It is self evident.
<security-role-mapping>
<role-name>application-admin</role-name>
<group-name>Administrator</group-name>
</security-role-mapping>

So the whole procedure is when somebody calls your application via an url container checks if there is a security-constraint on the url and which Security Role you need to have to access it. The container also know which Realm is responsible to authenticate the users in this application. Makes the authentication and send this info to your module. The login module is taken based on the jaas-context parameter of the Realm. Your module returns list of user groups if user is authenticated. If the user is in the group which is mapped to the necessary Security Role access is granted. That is all.

Firefox OCSP Server experiences an internal error ( sec_error_ocsp_server_error )

2008-09-19T23:30:00.008+02:00

I am in a Hotel now and they have free Wifi in the rooms. This would be a nice feature if it worked. The theory is that Wifi is not protected an authentication is not required but when you type in an URL the gateway forwards you to an authentication page - apc.aptilos.com, pas.aptilo.com - and you have to type in a user name and password what you can get at the reception - or via the TV.
If you type in the user name and password you can go through the gateway for a while.
I suppose even the mac address or the IP is registered in the gateway and/or in the firewall and allowed to communicate.
Nice concept except Wifi is not encrypted.

But I had a problem. When I typed in any URL in Firefox it does not work and after a while I got the error message:
"OCSP Server experiences an internal error" ( sec_error_ocsp_server_error )
I have tested this from Konqueror and it was working fine. ( I have Ubunut 8.04 now. )

I made a network traffic sniff and saw the problem is that Firefox tries to check if the servers key is revoked via OCSP. But of course this is not possible since network is not opened by this time.
The solution from aptilo would be to open the connection to the corresponding OCSP server. I do not belive this will be done.
The solution from client side is to disable OCSP in Firefox or use a client which does not have OCSP.
To disable OCSP in Firefox go to Preferences->Advanced->Encryption->Validation. Here you have a choice.

You can completely disable OCSP.

You can disable to treat the connecting server invalid if OCSP fails.

Whatever you do NEVER FORGET TO ENABLE THIS AFTER YOU LEFT THE PLACE.
This is a security feature for your safety.

Firefox Phishing ignores user' ignore.

2008-07-19T00:57:00.001+02:00

Firefox does not allow you to ignore the Phishing warning sometimes because of two stupid bug even if you know this is not a Phishing site.
This is because of bugs: 435081 and 442929
A quick BUT DANGEROUS workaround is to change two config values:
browser.safebrowsing.enabled=false
browser.safebrowsing.malware.enabled=false
which disables safebrowsing.

Please note:

- Always enable this feature again right after you finished browsing the site you wanted. Besides does not open any other site meanwhile.

- You might believe this is not dangerous to disable safebrowsing on a site you know but this is FALSE.
A cracker can crack the site and can put dangerous code snippets on the page without anybody noticing that.
Even the site owner does not know about it. The code snippet can sit silently on the server and nobody can see that but can infect machines. Of course since the Phishing protection used by Firefox is black-list based somebody - or something - has already noticed and reported the site. If the site was hacked the owner should be informed and should correct the page. If the site was build intentionally you should get out of this immediately.

Netbeans 6.0.1 Missing Ruby and Rails Problem

2008-02-21T11:19:00.003+01:00

Today decided to test Netbeans 6.0.1. I was satisfied with 6.0 but I had a stupid problem with JAX-RPC which prevented to use that feature. Instead of trying to solve that decided to move on to this version. So I downloaded the OS independent zip - this is what I like after all - but when I started... ooops.
I got a message that Ruby and Rails version is not correct. So what? Try to upgrade. Same error again. What a shame. Decided to uninstall the Ruby and Rails and restart. Did not help. Deleted ruby files from ~/.netbeans/6.0/modules also did not help.
Finally I solved this problem with starting the IDE and updating only IDE Platform first. Since that I could not reproduce the error and I have no time now to play more with this. So if you are having this make a try. But I will try to find out the real reason as soon as I will have some time.

Netbeans Missing New Project Categories

2008-01-29T15:51:00.000+01:00

Jump to solution. (Skip stupid story.)

It was raining outside and it was a dark and said day and there was nothing to work on when I decided to give a try to Netbeans in J2EE development. If you have read my blog before you must know I am an Ubuntu user. Actually not a fan. I am not loving anything easily. But I am in general very satisfied with Ubuntu. I like it very much. Am I a fan? Together with its problem too. But from time to time I am running into problems which can make life harder than .... than is should be. (Life is hard anyway. It does not matter if you are Linux Windows or Mac OS user. So hold on your arguments about it. :) )

Back to the point I have Ubuntu Gutsy which currently contains Netbeans 5.5. So I started it and wanted to start a new project. What a surprise I have seen I could not. There were no menu items besides import existing ant project what I could reach. Holy s.... something! What is this? I was asking.
Tried to purge package and reinstall but no luck. So finally I concluded that the problem is around Netbeans 5.5. What can people do than? I have taken the lattes version Netbeans 6.0 and tried. Much more surprise I had the same situation. No menu to create new project. I started to think about some kind of global conspiracy and decided to unveil it so the day just turned darker.

So first started to Google around. It almost always helps. And I found ideas to delete the ~/.netbeans directory and restart IDE. It did not help me too much. Also find some idea to use different java version. Which also did not help me.
What helped me was deleting ~/.netbeans AND using a Sun java version together. Even java 6 or 5 from Sun. (Anyway you can specify the java version as a netbeans command line parameter like: $ netbeans --jdkome /usr/lib/jvm/java-6-sun )
So I would have sat back and enjoy it. But I wanted to know what was wrong. So first checked that the java version which prevented me to use netbeans was icedtea-java-7. The other thing I had to realize - and primarily made me to investigate it - was that if you ever started netbeans with icedtea-java-7 you can not start it after even with the sun java without deleting ~/.netbeans . This is very uncomfortable since you can loose your settings I decided to find out what was the problem. I do not think that you are interested how I did it but here is the result and how to correct it without loosing your settings.

Solution:
$ rm -rf ~/.netbeans/6.0/config/Modules

Simple right?
If you are missing New Project Templates in Netbeans and you have started netbeans ever with IcedTea Java 7 - or any other "unsupported one" - first use only SUN Java as long as this error is not corrected to run netbeans. I do not know if this is an error in Iced Tea Java or in Netbeans and do not care too much anymore. To correct the problem completely delete ~/.netbeans/6.0/config/Modules and restart Netbenas with a Sun Java. This directory will be recreated and you can enjoy the newly available menus without loosing your settings.

Hope it helped you too.

FlightGear Keyboard Shortcut Reference

2008-01-06T23:53:00.000+01:00

I just wanted to have a list of keys can be used in FlightGear but found only the short Reference.
I realized that key shortcuts is an xml file so easy to change and not good to publish a static one on the net while anybody can change it anytime easily. So here is a command to get it under linux anytime on your own machine.

First please change directory to the one contains keyboard.xml. Gutsy have it under /usr/share/games/FlightGear


$ awk 'BEGIN {name="";desc="";spec="";} \
/.*<\/name>/{ name=$0; spec="";desc="";} \
/.*<\/desc>/ { desc=$0 ; line=spec " " name " = " desc; print gensub("<[^>]*>([^<>]*)<\/[^>]*>","\\1","g",gensub("\ \ +","\ ","g", line)) ;} \
/<mod-.*>/ {  spec=gensub("<mod-(.*)>","\\1 + ","g",$0); } \
' keyboard.xml

Output is something like:
Ctrl-A = Toggle autopilot altitude lock.
Ctrl-B = Toggle speedbrake.
Ctrl-C = Toggle clickable panel hotspots
Ctrl-G = Toggle autopilot glide slope lock.
Ctrl-H = Toggle autopilot heading lock.
Ctrl-I = Show instrument setting dialog.
...

Than format print, learn and enjoy the game. :)

Insufficient Disk Space reported under wine

2007-11-03T19:30:00.000+01:00

Did you try to install/setup any Windows Application - actually a Game what else could be necessary - and got a message that you do not have enough free space on your drive meanwhile you had lot of free space on the chosen mounted partition?
You will learn the problem and hopefully the solution too. (Of course I suppose it is not the real situation you have no enough space. If so do not read ahead.)

The problem is that wine does not check the amount of free space on the mounted partition corresponds to the selected directory but reports the free on the root of the directory the partition mounted to . ;( Probably it is not clean so here is an example:
Let say you have / only and something is mounted as /mnt/part1
If you directly select /mnt/part1 during installation wine will check free space in fact on / and does not calculate free on the partition mounted under /mnt/part1.

How to solve it you may ask? It is easy. Start winecfg and create a new drive with the directory you want to use.

Keywords: Wine Insufficient free space not enough free space on drive

Ubuntu Gutsy 64bit Wine misses ssl support so Google Picasa can not sign in.

2007-09-27T02:07:00.000+02:00

Today I have realized that Picasa - I am using "windows version" of picasa under wine - can not sign me in - means no web upload is possible - from my home machine (Ubuntu Gutsy 64bit) and wine showed the following error:

fixme:win:FlashWindowEx 0x33afd4
fixme:wininet:InternetSetOptionW Option INTERNET_OPTION_CONNECT_TIMEOUT (10000): STUB
fixme:wininet:InternetSetOptionW INTERNET_OPTION_SEND/RECEIVE_TIMEOUT not supported on protocol 1
fixme:wininet:NETCON_init can't use SSL, not compiled in.

I was a bit surprised since Picasa on my office notebook (Ubuntu Feisty 32bit) worked fine. So checked the error and decided to recompile wine and see if ssl is really not compiled in.

First I downloaded the source and dependencies needed to compile.

# apt-get build-dep wine

I would mention that it will download a lot.
Then downloaded wine source too into a temporary directory and tried to compile.

# mkdir /tmp/wine_build
# cd /tmp/wine_build
# apt-get source wine
# cd wine-0.9.45
# dpkg-buildpackage -rfakeroot -b

And part of the result is:

checking for -lcapi20... not found [ CAPI support ISDN Cards ]
checking for -lcups... not found [ CUPS printing support ]
checking for -ljack... not found [ Jack audio server support ]
checking for -lssl... not found
checking for -lcrypto... not found
checking for -ljpeg... not found
checking for -lpng... not found

So configure could not find ssl library. After some check it turned out that this is a 32 bit compilation therefore it needs 32 bit libraries. And of course I did not have some 32 bit lib on my system. My current problem is the access of internet because of the missing ssl support so now focus on this.

I have downloaded libssl0.9.8_0.9.8e-5ubuntu1_i386.deb and extracted this and copied the missing files to the right place and created required soft links.

# mkdir /tmp/libssl
# dpkg -x libssl0.9.8_0.9.8e-5ubuntu1_i386.deb /tmp/libssl
# cp /tmp/libssl/usr/lib/libssl.so.0.9.8 /usr/lib32/
# cp /tmp/libssl/usr/lib/libcrypto.so.0.9.8 /usr/lib32/
# ln -s /usr/lib32/libssl.so.0.9.8 /usr/lib32/libssl.so
# ln -s /usr/lib32/libcrypto.so.0.9.8 /usr/lib32/libcrypto.so

If you try to compile wine again with dpkg-buildpackage -rfakeroot -b this should work now.

[ Update: It took a long time on my machine. :( But finally it is working. :) ]

[ Update: I saw sadly that some fonts are not visible under Picasa. Hmmm. Not easy to get it to work. But it can be. If you have font problem install msttcorefonts. But that is not all you have to do. After that you have to remove your ~/.wine directory and reinstall Picasa. The important thing is not Picasa reinstallation. But when wine does not find a ~/.wine it recreates and this is the point after all. I'll try to find another way to do this but this is working. ]

[ Update: Tried to start EVE Online under wine and found two problems. Both were discussed on wine web page but here are details.
Main error is below but first solve the easier one:

First
$ cp /usr/share/fonts/truetype/msttcorefonts/arial.ttf ~/.wine/drive_c/windows/fonts/

Main error message:
fixme:d3d_draw:drawPrimitive Using software emulation because not all material properties could be tracked
Still do not know really how to solve this.]

P.S. :

As you could see some other libraries are missing too. So if you have a similar problem just do a procedure like this and you can correct it.

[Remark]
The following libraries were also missing:
checking for -lcapi20... not found
checking for -lcups... not found
checking for -ljack... not found
checking for -lssl... not found
checking for -lcrypto... not found
checking for -ljpeg... not found
checking for -lpng... not found

The followings are also missing so in case of problem with wine on 64 bit linux you should consider this list:

checking for IceConnectionNumber in -lICE... no
checking for x86_64-linux-gnu-ar... no
checking for x86_64-linux-gnu-ranlib... no
checking for x86_64-linux-gnu-strip... no
checking for x86_64-linux-gnu-windres... no
checking for windres... no
checking for lclint... no
checking for lint... no
checking for i386_set_ldt in -li386... no
checking for _oss_ioctl in -lossaudio... no
checking AudioUnit/AudioUnit.h usability... no
checking AudioUnit/AudioUnit.h presence... no
checking for AudioUnit/AudioUnit.h... no
checking Carbon/Carbon.h usability... no
checking Carbon/Carbon.h presence... no
checking for Carbon/Carbon.h... no
checking CoreAudio/CoreAudio.h usability... no
checking CoreAudio/CoreAudio.h presence... no
checking for CoreAudio/CoreAudio.h... no
checking DiskArbitration/DiskArbitration.h usability... no
checking DiskArbitration/DiskArbitration.h presence... no

Line6 PodXT Linux Driver 0.7.2 is incompatible with kernel >= 2.6.22

2007-09-24T15:10:00.002+02:00

I have updated my Ubuntu Feisty to Gutsy some weeks ago because of a problem booting up from raid. Upgrade solved the issue but brought some others. One of the most serious to me was that the Linux Line6 PodXt driver
could not be compiled anymore. The error I got was:

make default
make -C /lib/modules/2.6.22-12-generic/build
SUBDIRS=/home/user1/workspace/line6usb-0.7.2 modules
make[1]: Entering directory `/usr/src/linux-headers-2.6.22-12-generic'
CC [M] /home/user1/workspace/line6usb-0.7.2/pcm.o
/home/user1/workspace/line6usb-0.7.2/pcm.c: In function
‘snd_pod_trigger’:
/home/user1/workspace/line6usb-0.7.2/pcm.c:40: warning: implicit
declaration of function ‘snd_pcm_group_for_each’
/home/user1/workspace/line6usb-0.7.2/pcm.c:40: error: expected ‘;’
before ‘{’ token
/home/user1/workspace/line6usb-0.7.2/pcm.c:33: warning: unused variable
‘err’
/home/user1/workspace/line6usb-0.7.2/pcm.c:32: warning: unused variable
‘s’
make[2]: *** [/home/user1/workspace/line6usb-0.7.2/pcm.o] Error 1
make[1]: Leaving directory `/usr/src/linux-headers-2.6.22-12-generic'
make[1]: *** [_module_/home/user1/workspace/line6usb-0.7.2] Error 2
make: *** [default] Error 2

So I investigated this a bit and found that some macro calls were obsoleted which is why pcm.c can not be compiled on 2.6.22 - actually neither on 2.6.21 and 2.6.20 is the latest working.
Fortunately this was very easy to correct this error. OK first I had to check kernel changelog and also Linux Cross Reference helped me a lot.
So here is the required modification in pcm.c of the driver to be able to compile the source again.

/* trigger callback */
int snd_pod_trigger(struct snd_pcm_substream *substream, int cmd)
{
struct snd_pod_pcm *chip = snd_pcm_substream_chip(substream);
struct list_head *pos;
struct snd_pcm_substream *s;
int err;
unsigned long flags;

printk(KERN_INFO "*** Entering snd_pod_trigger\n");
spin_lock_irqsave(&chip->lock_trigger, flags);
clear_bit(BIT_PREPARED, &chip->flags);

//snd_pcm_group_for_each(pos, substream) {
list_for_each(pos,&substream->group->substreams) {
//s = snd_pcm_group_substream_entry(pos);
s = list_entry(pos,struct snd_pcm_substream,link_list);

switch(s->stream) {
case SNDRV_PCM_STREAM_PLAYBACK:

Just change the lines so old ones are commented out and below are the new ones:

//snd_pcm_group_for_each(pos, substream) {
list_for_each(pos,&substream->group->substreams) {
//s = snd_pcm_group_substream_entry(pos);
s = list_entry(pos,struct snd_pcm_substream,link_list);

As soon as I have a bit more free time I will make a patch - a diff - to apply.

This error was reported via sourceforge and I am pretty sure we can see a corrected version soon officially from Marcus Grabner. But till then you can do it on your own easily.

P.S. : I would like to say thanks to Marcus Grabner for the driver.

[Update] The patch can be downloaded from here or from here. First save it into your home directory. Download the driver also into your home than execute the following commands (I suppose that line6usb-0.7.2.tar.bz2 and line6usb-0.7.2_patch.diff are saved into your home) :
$ cd ~
$ bzip2 -dc line6usb-0.7.2.tar.bz2 | tar xfv -
$ cd ~/line6usb-0.7.2
$ patch -p1 <~/line6usb-0.7.2_patch.diff That is all. Hope it works.