cat /dev/snapshot

Integrating Jenkins, SonarQube, and Nexus

noreply@blogger.com (Unknown) — Sun, 23 Feb 2014 23:25:00 +0000

In this post I will describe a solution to add a portal page, a common header and a dynamic panel to each of the three tools Jenkins, SonarQube, and Nexus. Those modifications lead to a better visual integration and intensifies the impression that the tools belong together.

As a pre-requisite you need an Apache HTTP server, configured as reverse proxy and the tools accessible via the following context path:

/jenkins
/sonarqube
/nexus

The folder structure which contains the portal page and the customization fragments looks like this:

index.html
js

portal-panel.js
jenkins-customization.js
sonarqube-customization.js
nexus-customization.js
require-2.1.11.js
jquery-1.11.0.js

Portal Page

The portal page index.html is available under the context root and basically contains the following links:

<a href="https://draft.blogger.com/jenkins">Jenkins</a>
<a href="https://draft.blogger.com/sonarqube">SonarQube</a>
<a href="https://draft.blogger.com/nexus">Nexus</a>
<a href="https://draft.blogger.com/wiki">Wiki</a>
<a href="mailto:xxx@yyy.zzz">Mail</a>

Portal Panel

The portal panel is a small box at the top of the page at a fixed position. When the mouse enters the box it gets maximized and displays buttons for accessing the other tools. When the mouse exits the box, the box shrinks to its initial size. When scrolling the page down, the box remains at the top of the window so that the other the tools are accessible from everywhere in the page.

The panel heavily relies on jQuery to do the animations and the styling. The following block contains the script with some information removed for clarity:

function poller()
{
    // wait until jQuery becomes ready
    if(typeof jQuery !== "undefined") {

        // settings
        var animMillis = 200;
        var collapsedWidth = 30;
        var height = 30;
        var expandedWidth = 350;
        var logo = "/img/Logo.png";
        var email = "xxx@yyy.zzz";
        var germanMailText = "Hallo Support Team,%0A%0AQuelle: " +
            window.location.href + "%0A%0A%0A%0AMit freundlichen Gruessen,%0A";
        var englishMailText = "Hello Support Team,%0A%0AOrigin: " +
            window.location.href + "%0A%0A%0A%0AWith Kind regards,%0A";

        // add center method to jQuery
        jQuery.fn.center = function () {
            this.css("position", "fixed");
            this.css("top", "0px");
            var width = jQuery(this).width();
            var left = Math.max(0, ((jQuery(window).width() - width) / 2) + 
                                   jQuery(window).scrollLeft());
            this.css("left", left + "px");
            return this;
        }
        
        // detect mail language
        var language = window.navigator.userLanguage ||
                       window.navigator.language;
        var mailText = englishMailText;
        if (language.indexOf('de') == 0) {
            mailText = germanMailText;
        }

        // add markup (removed for clarity)
        jQuery("body").append('MARKUP');

        // style markup (removed for clarity)
        jQuery("#portalPanel").css("STYLING");
        jQuery("#portalLogo").css("STYLING");
        jQuery("#portalMenu").css("STYLING");
        jQuery(".portalButton").css("STYLING");
        jQuery(".portalButton a:link").css("STYLING");
        jQuery(".portalButtonMagenta").css("STYLING");
        jQuery(".portalButtonDark").css("STYLING");
        jQuery(".portalButtonLight").css("STYLING");

        // button animation
        jQuery(".portalButtonMagenta").hover(function() {
            jQuery(this).css("background-color", "#f59fcd");
        }, function() {
            jQuery(this).css("background-color", "#d8006f");
        });
        jQuery(".portalButtonDark").hover(function() {
            jQuery(this).css("background-color", "#b1b1b1");
        }, function() {
            jQuery(this).css("background-color", "#4c4c4c");
        });
        jQuery(".portalButtonLight").hover(function() {
            jQuery(this).css("background-color", "#f2f2f2");
        }, function() {
            jQuery(this).css("background-color", "#b8b8b8");
        });

        // make box relocate when browser window is resized
        jQuery(window).bind('resize', function() {
            jQuery('#portalPanel').center();
        });

        // initial state
        jQuery("#portalPanel").center();
        jQuery("#portalLogo").show();
        jQuery("#portalMenu").hide();
        
        // animate the box
        var dirin = true;
        jQuery("#portalPanel").hover(function() {
            dirin = true;
            jQuery('#portalMenu').clearQueue().stop(true, true, true);
            jQuery('#portalPanel').clearQueue().stop(true, true, true);
            jQuery('#portalLogo').clearQueue().stop(true, true, true);

            if (dirin) {
                jQuery('#portalLogo').fadeOut(animMillis, function() {
                    if (dirin) {
                        var left = Math.max(0,
                            ((jQuery(window).width() - expandedWidth) / 2) + 
                            jQuery(window).scrollLeft());
                        jQuery('#portalPanel').animate({'width':expandedWidth +
                               'px', 'left':left + 'px'}, animMillis, 'swing',
                            function() {
                              if (dirin) {
                                jQuery('#portalMenu').fadeIn(animMillis,
                                  function() {
                                    // do nothing
                                });
                              }
                        });
                    }
                });
            }
        }, function() {
            dirin = false;
            jQuery('#portalMenu').clearQueue().stop(true, true, true);
            jQuery('#portalPanel').clearQueue().stop(true, true, true);
            jQuery('#portalLogo').clearQueue().stop(true, true, true);
            
            if (!dirin) {
                jQuery('#portalMenu').fadeOut(animMillis, function() {
                    if (!dirin) {
                        var left = Math.max(0,
                            ((jQuery(window).width() - collapsedWidth) / 2) +
                            jQuery(window).scrollLeft());
                        jQuery('#portalPanel').animate({'width':collapsedWidth +
                               'px', 'left':left + 'px'}, animMillis, 'swing',
                            function() {
                              if (!dirin) {
                                jQuery('#portalLogo').fadeIn(animMillis,
                                  function() {
                                    jQuery('#portalLogo').show();
                                });
                            }
                        });
                    }
                });
            }
        });
    }
    else
    {
        setTimeout(poller, 100);
    }    
}

poller();

Jenkins Customization

To use jQuery in our customization fragments the Jenkins jQuery plugin must be installed via the Jenkins Update Center. The plugin adds the jQuery JavaScript library to each Jenkins page without colliding with Jenkins' JavaScript. Because Jenkins uses the jQuery shortcut "$" for one of its own functions, jQuery can only be accessed by using the full name "jQuery".

Download the latest release of the page-markup plugin and copy the page-markup.hpi to the folder JENKINS_HOME/plugins.

Finally the plugins need to be activated by restarting the Jenkins server.

Add the following snippet to the text area under Manage Jenkins -> Configure System -> Additional Page HTML Markup -> Header HTML to extend the Jenkins HTML page:

<style type="text/css">
#top-panel {
display:none;
background: none;
}
.top-sticker, #top-sticker {
top: 62px;
}
</style>
<script src="/js/require-2.1.11.js"></script>
<script>
require(["/js/jenkins-customization.js"], function(util) {
// nothing to do
});
require(["/js/portal-panel.js"], function(util) {
// nothing to do
});
</script>

The small CSS part hides the Jenkins logo and the blue header so that it does not flicker when the page is reloaded. Because the jQuery code takes some time to get executed, the logo is displayed shortly if no countermeasures are taken.

SonarQube Customization

Download the sonar-pagedecoration-plugin and install it manually to the $SONARQUBE_HOME/extensions/plugins folder of your SonarQube installation. You need to restart SonarQube to get the plugin activated.

Add the following snippet to the Settings -> General Settings -> Page Decoration -> Script text area and click on Save Page Decoration Settings to add the fragment to the SonarQube HTML page:

<script src="/js/require-2.1.11.js"></script>
<script>
require(["/js/sonarqube-customization.js"], function(util) {
// nothing to do
});
require(["/js/portal-panel.js"], function(util) {
// nothing to do
});
</script>

The SonarQube customization does not need to load jQuery explicitly as SonarQube already uses jQuery in its HTML pages.

Nexus Customization

Download the nexus-pagedecoration-plugin, unzip the file nexus-pagedecoration-plugin-2.7.1-01-bundle.zip to the folder SONATYPE_WORK/nexus/plugin-repository and restart Nexus to activate the plugin.

Add the following snippet to the Administration -> Page Decoration -> Pre Head text area and click on Save to add the fragment to the Nexus HTML page:

<style type="text/css">
load.
div#branding img {
display:none;
}
div#header {
background-color: white;
background-image: none;
}
</style>
<script src="/js/require-2.1.11.js"></script>
<script>
require(["/js/jquery-1.11.0.min.js"], function(util) {
require(["/js/nexus-customization.js"], function(util) {
// nothing to do
});
require(["/js/portal-panel.js"], function(util) {
// nothing to do
});
});
</script>

In Nexus the original logo also needs to be hidden to avoid flickering upon page reload.

The Nexus UI is based upon ExtJS so that jQuery needs to pulled first. When jQuery has been loaded completely, the other JavaScript files get loaded and are executed.

The scripts can be found in the following GitHub project:

https://github.com/mrumpf/software-build-service/tree/master/portal

Page decoration plugins for Jenkins, SonarQube, and Nexus

noreply@blogger.com (Unknown) — Sun, 23 Feb 2014 17:34:00 +0000

While trying to apply a common visual layout to the tools Jenkins, SonarQube, and Nexus I learned that there are some plugins available, but none of them really fits the intended purpose.

Jenkins has no easy way of changing its appearance and the Simple Theme plugin comes only close to what is needed. In SonarQube there is a Branding plugin with which you can display an image at the top of each page, but that's basically all. The Nexus plugin to change the Sonatype Nexus logo is available for the professional version only, but buying the professional version just for being able to change the logo is not really an option.

Analyzing the APIs of the different tools revealed that similar extension points are offered in order to inject HTML fragments into the HTML page:

Jenkins: hudson.model.PageDecorator
SonarQube (since 3.3): org.sonar.api.web.PageDecoration
Nexus: org.sonatype.nexus.plugins.rest.NexusIndexHtmlCustomizer

For Jenkins and SonarQube I found plugins which use these extension points, although the plugins are not available via the official plugin repositories and therefore somewhat hidden from the public:

Jenkins: page-markup (GitHub)
The original Subversion repository has been forked to GitHub but the plugin has not made it into the Jenkins Plugin Repository yet.
SonarQube: sonar-pagedecoration-plugin (GitHub)
It seems as if there has been no attempt at getting the plugin into the SonarQube Plugin Forge yet.
Nexus: No plugin available or at least I did not find one yet.

Inspired from the Blog postings "Nexus and UI contributions" and "Extending Nexus UI" I decided to write my own plugin which matches the functionality of the SonarQube's pagedecoration and the Jenkins's page-markup plugins.

The new nexus-pagedecoration-plugin adds a menu item "Page Decoration" to the Administration section of the Nexus UI:

The corresponding Page Decoration tab provides text areas for all the four extension points:

Pre Head
Post Head
Pre Body
Post Body

You can add HTML fragments to each of the text areas and when saving the changes and reloading the page, the fragments will be rendered into the Nexus UI HTML page.

The nexus-pagedecoration-plugin is available under the Apache 2.0 license and is hosted on Github. The initial release for Nexus 2.7.1-01 can be found under GitHub releases.

For the other plugins I did a fork on GitHub and made some releases on GitHub:

page-markup release: v0.4-SNAPSHOT
sonar-pagedecoration-plugin release: v1.0.1-SNAPSHOT
nexus-pagedecoration-plugin release: v2.7.1-01

Music Player Daemon on Raspberry Pi

noreply@blogger.com (Unknown) — Mon, 10 Jun 2013 19:06:00 +0000

This is not yet another "how to configure MPD on the Raspberry Pi" article, because there are a lot of excellent MPD HOWTOs already available:

However, I experienced a bottleneck with the configuration described by those articles. The load was slightly over 1 but it was high enough to cause some stuttering every 30 seconds.

The first optimization is to avoid using a separate streaming solution like icecast2. MPD has its own, built-in streaming plugin which can be configured like this:
audio_output {

type "httpd"
name "My HTTP Stream"
encoder "flac"
port "8000"
bitrate "96"
format "44100:16:2"
}

But before adding this you need to check your sound card default output parameters:
sudo cat /proc/asound/card0/pcm0p/sub0/hw_params

which returns something like this:
access: RW_INTERLEAVED
format: S16_LE
subformat: STD
channels: 2
rate: 44100 (44100/1)
period_size: 5513
buffer_size: 22050

Then note the rate and channels above and make sure they are the same as in mpd.conf:
audio_output {
type "alsa"
name "My ALSA Device"
device "hw:0,0" # optional
format "44100:16:2" # CHECK rate and channels
mixer_device "default" # optional
mixer_control "PCM" # optional
mixer_index "0" # optional
}

audio_output {
type "httpd"
name "My HTTP Stream"
# we are using a faster encoder note that some player might not supprt Flac
encoder "flac"
port "8000"
# quality "5.0" # do not define if bitrate is defined
# leave at 128 and check cpu usage...
bitrate "96" # do not define if quality is defined
format "44100:16:2"
}

# This setting will change all decoded audio to be converted to the specified
# format before being passed to the audio outputs. By default, this setting is
# disabled.
#
audio_output_format "44100:16:2"
#
# If MPD has been compiled with libsamplerate support, this setting specifies
# the sample rate converter to use. Possible values can be found in the
# mpd.conf man page or the libsamplerate documentation. By default, this is
# setting is disabled.
#
samplerate_converter "Fastest Sinc Interpolator"

The important hints for those optimizations came from the Raspberry Pi Forum here.

Debugging UPnP/DLNA traffic to a Sony LCD TV

noreply@blogger.com (Unknown) — Mon, 08 Oct 2012 12:27:00 +0000

The issue I have is that my Sony LCD TV seems to be very picky regarding the UPnP/DLNA protocol when playing media from a control point to the television. The Android App BubbleUPnP does a good job here as playing music works out of the box. Initial tests with the Cling Java library to play music does not work out of the box. Therefore I wanted to debug the HTTP SOAP protocol in order to see where the differences are.

The UPnP inspector tools discover all UPnP devices on the network, but they cannot monitor the traffic between the devices as the communication is happening directly (service discovery is via multicast):

Because of a switched network the traffic does not go through my OpenWRT router either, so sniffing on the WLAN-LAN bridge br-lan does not work either. That means the only chance to get a hook into the communication channel was to use arpspoof, part of the dsniff networking tool-chain, together with tcpdump and wireshark for sniffing and visualization.

"ARP spoofing[1] is a computer hacking technique whereby an attacker sends fake ("spoofed") Address Resolution Protocol (ARP) messages onto a Local Area Network. Generally, the aim is to associate the attacker's MAC address with the IP address of another host (such as the default gateway), causing any traffic meant for that IP address to be sent to the attacker instead." (See http://en.wikipedia.org/wiki/ARP_spoofing)

On my Ubuntu desktop the following packages need to be installed to get started:

sudo apt-get install dsniff wireshark tcpdump

Three devices are part of the show:

Android Phone - 192.168.0.107
Sony LCD TV - 192.168.0.77
Desktop PC - 192.168.0.17

Before spoofing can be set up we need to make sure that IP packages are forwarded on the desktop machine:

sudo "echo 1 > /proc/sys/net/ipv4/ip_forward"

We need to tell the Android Phone that we are the Sony LCD TV:

sudo arpspoof -i eth0 -t 192.168.0.107 192.168.0.77

and the Sony LCD TV that we are the Android Phone:

sudo arpspoof -i eth0 -t 192.168.0.77 192.168.0.107

Then the sniffing can be started with the following command

sudo "tcpdump not arp -s0 -w - | wireshark -k -i -"

In Wireshark you can simply search for the String "AVTransportURI" (Edit / Find packet... / Find By: String "AVTransportURI" / Search In: Packet Bytes) to find the HTTP request that informs the LCD TV which song to play.
Right click on the selected line and choose "Decode as..." from the context menu. Then swith to the "Transport" tab and select "HTTP" from the list. Finally click on "OK" to decode the packet as HTTP.

POST /upnp/control/AVTransport HTTP/1.1
Content-type: text/xml;charset="utf-8"
Soapaction: "urn:schemas-upnp-org:service:AVTransport:1#SetAVTransportURI"
Content-Length: 1493
Host: 192.168.0.77:52323
Connection: Keep-Alive
User-Agent: Android/4.1.1 UPnP/1.0 BubbleUPnP/1.5.4

<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<s:Envelope s:encodingStyle="http://schemas.xmlsoap.org/soap/encoding/" xmlns:s="http://schemas.xmlsoap.org/soap/envelope/">
<s:Body>
<u:SetAVTransportURI xmlns:u="urn:schemas-upnp-org:service:AVTransport:1">
<InstanceID>0</InstanceID>
<CurrentURI>http://192.168.0.107:57645/external/audio/media/10839.mp3</CurrentURI>
<CurrentURIMetaData>
<DIDL-Lite xmlns="urn:schemas-upnp-org:metadata-1-0/DIDL-Lite/" xmlns:upnp="urn:schemas-upnp-org:metadata-1-0/upnp/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:dlna="urn:schemas-dlna-org:metadata-1-0/">
<item id="/external/audio/albums/6/10839" parentID="/external/audio/albums/6" restricted="1">
<upnp:class>object.item.audioItem.musicTrack</upnp:class>
<dc:title>Blind Willie</dc:title>
<dc:creator>Rob Towns</dc:creator>
<upnp:artist>Rob Towns</upnp:artist>
<upnp:albumArtURI>http://192.168.0.107:57645/external/audio/albums/6.jpg</upnp:albumArtURI>
<upnp:album>media</upnp:album><dc:date>2001-01-01</dc:date>
<res protocolInfo="http-get:*:audio/mpeg:DLNA.ORG_PN=MP3;DLNA.ORG_OP=01;DLNA.ORG_CI=0" size="2703360" duration="0:02:15.000">http://192.168.0.107:57645/external/audio/media/10839.mp3</res></item>
</DIDL-Lite>
</CurrentURIMetaData>
</u:SetAVTransportURI>
</s:Body>
</s:Envelope>

With this method it will be easily possible to tweak a Cling Java client to speak to the Sony LCD TV.

Use Java 7 on Buildhive

noreply@blogger.com (Unknown) — Tue, 14 Aug 2012 12:09:00 +0000

Changing source and target version to 1.7 is not sufficient to enable Java 7 on Buildhive at the moment.
Adding the following profile to your Maven POM makes sure that your local build is not affected by the special configuration and when running on Buildhive, JDK 7 is used for both build and test-run.

<profiles>
<profile>
<id>buildhive</id>
<build>
<plugins>
<plugin>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-compiler-plugin</artifactId>
<version>2.5.1</version>
<configuration>
<source>1.7</source>
<target>1.7</target>
<showDeprecation>true</showDeprecation>
<showWarnings>true</showWarnings>
<encoding>UTF-8</encoding>
<executable>/opt/jdk/jdk1.7.0/bin/javac</executable>
<fork>true</fork>
</configuration>
</plugin>
<plugin>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-surefire-plugin</artifactId>
<configuration>
<jvm>/opt/jdk/jdk1.7.0/bin/java</jvm>
<forkMode>once</forkMode>
</configuration>
</plugin>
</plugins>
</build>
</profile>
</profiles>

The profile can be selected by adding the following line to the configuration of your Buildhive job:

Maven Goals: -Pbuildhive clean install

Raspberry Pi

noreply@blogger.com (Unknown) — Sat, 30 Jun 2012 10:26:00 +0000

My Raspberry Pi finally arrived after 4 months:

The stable Debian distribution "Squeeze" does not contain support for all of the Raspberry Pi features. That means there is no way around the the latest Debian "Wheezy" image: http://www.raspberrypi.org/archives/1435

But even with the latest image version there is still some configuration effort left...

WLAN Stick

For network connectivity I have chosen the Conrad WLAN Stick N150 Nano, which is based on the following chipset: USB: 0BDA:8176 Realtek Semiconductor Corp. RTL8188CUS 802.11n WLAN Adapter.
Unfortunately this chipset has some issues with the current drivers from both Debian Linux images as discussed in the forum thread Automatic setup for RTL8188CUS based wifi adapters. By executing the script install-rtl8188cus-latest.sh from the forum thread, the adapter started working immediately after the script was finished.
The general WLAN configuration is described in detail in the Debian Wiki: http://wiki.debian.org/WiFi/HowToUse#Command_Line

Analog Audio

The following commands must be issued to activate the analog output of the Raspberry Pi:


sudo apt-get install alsa-utils gstreamer-tools \\


   madplay gstreamer0.10-alsa


sudo modprobe snd_bcm2835

sudo amixer cset numid=3 1

Now you can play a mp3 audio stream via the ALSA device like this:


gst-launch souphttpsrc \\ location=http://dradio_mp3_dlf_m.akacast.akamaistream.net/7/249/142684/v1/gnl.akacast.akamaistream.net/dradio_mp3_dlf_m \\


   ! mad \\

   ! audioconvert \\

   ! audioresample \\

   ! alsasink device=hw:0

The fact that the device needs to be specified explicitly "device=hw:0" indicates another issue here as this is not necessary on my desktop Linux machine.

For testing the output without an internet connection you can use the following commands:


cd /opt/vs/src/hello_pi

make

./hello_audio

The audio output is not as clear as I expected it to be. I still hear some clicking noises when playing a stream. I hope this will be fixed in upcoming releases of the analog audio driver.

GPIO Ports

In a first test I connected a button to the GPIO port 0 and tried to read the switch state via a simple script:

sudo -s
echo "0" > /sys/class/gpio/export
echo "in" > /sys/class/gpio/gpio0/direction
watch cat /sys/class/gpio/gpio0/value

When pressing and holding the button pressed for at least 2 seconds you will see that the value turns from 1 (not pressed) to 0 (pressed).

Because the polling approach consumes too much resources I started looking after an interrupt solution.
The Debian "Wheezy" Beta version does not support interrupts for input ports. However, patches exist and have been included into Linux kernel version 3.2.21 from here: http://www.bootc.net/archives/2012/07/01/new-revision-of-3-2-21/
In order to download the kernel and activate it, run the following commands (NOTE: the kernel version changes quite frequently, so the 3.2.23 might be outdated):


wget http://apt.bootc.net/debian/pool/main/l/linux-source-3.2.23-rpi1+/linux-image-3.2.23-rpi1+_1_armel.deb
sudo dpkg -i linux-image-3.2.23-rpi1+_1_armel.deb

sudo cp /boot/vmlinuz-3.2.23-rpi1+ /boot/kernel.img

After a reboot the kernel provides interrupts on GPIO input ports and a simple C application published on the following thread demonstrates how to handle interrupts: http://www.raspberrypi.org/phpBB3/viewtopic.php?t=7509&p=92074

Wake On LAN

noreply@blogger.com (Unknown) — Sat, 03 Mar 2012 21:50:00 +0000

Under Ubuntu I chose the following way to activate wake on LAN on the eth0 device:

# Install ethtool
sudo apt-get install ethtool

# Create init script to set the wol bit on the device
sudo vi /etc/init.d/wakeonlan

#!/bin/sh
ethtool -s eth0 wol g
exit

# Make the script executable
sudo chmod a+x /etc/init.d/wakeonlan

# Create the init links
sudo update-rc.d wakeonlan defaults

Under Debian the following approach has been chosen:

# Get root
sudo su -

# Edit the network interface configuration
vi /etc/network/interfaces

# Add the lines post-up/pre-down to your network device section
iface eth0 inet dhcp
post-up /usr/sbin/ethtool -s $IFACE wol g
pre-down /usr/sbin/ethtool -s $IFACE wol g

The post-up/post-down commands execute the ethtool each time the eth0 device gets started or stopped. This happens also during boot (post-up) and shutdown (pre-down).

Both approaches activate WOL on the network device so that the machine can be started by sending a WOL package.

Deploying weld-osgi to jcoderz.org repository

noreply@blogger.com (Unknown) — Sat, 31 Dec 2011 15:36:00 +0000

For deploying weld-osgi to the jcoderz.org repository the following modifications to the pom.xml files are necessary (after applying the wagon-ssh extensions to the your Maven installation, see previous post):

1. Deploy the WELD API packages

git clone -b WELD-963 http://github.com/mathieuancelin/api.git WELD-963

cd WELD-963

vi pom.xml

# Add the following XMl fragment to the maven-deploy-plugin

   <build>
      <plugins>
         <plugin>
            <groupId>org.apache.maven.plugins</groupId>
            <artifactId>maven-deploy-plugin</artifactId>
            <configuration>
               <skip>true</skip>
            </configuration>
            <dependencies>
              <dependency>
                <groupId>org.apache.maven.wagon</groupId>
                <artifactId>wagon-ssh</artifactId>
                <version>2.2</version>
              </dependency>
            </dependencies>         </plugin>
      </plugins>
   </build>

mvn -e clean install -Dmaven.test.skip=true

mvn
 -e deploy 
-DaltDeploymentRepository=jcoderz-snapshots::default::scp://jcoderz.org/var/www/repos/maven2-snapshot/
 -Dmaven.test.skip=true

cd ..

2. Deploy the WELD OSGi packages

git clone -b weld-osgi http://github.com/mathieuancelin/core.git weld-osgi

cd weld-osgi

vi pom.xml 

 # Add the following XML into the <build>/<plugins> tag section

         <plugin>
           <groupId>org.apache.maven.plugins</groupId>
           <artifactId>maven-deploy-plugin</artifactId>
           <dependencies>
             <dependency>
               <groupId>org.apache.maven.wagon</groupId>
               <artifactId>wagon-ssh</artifactId>
               <version>2.2</version>
             </dependency>
           </dependencies>
         </plugin>

mvn -e clean install -Dmaven.test.skip=true

mvn
 -e deploy 
-DaltDeploymentRepository=jcoderz-snapshots::default::scp://jcoderz.org/var/www/repos/maven2-snapshot/
 -Dmaven.test.skip=true

cd environments/osgi



# Deploy weld-osgi-parent pom.xml to the repository (Is there a way to include the parent pom.xml in the previous run?)

mvn
 -e -N deploy 
-DaltDeploymentRepository=jcoderz-snapshots::default::scp://jcoderz.org/var/www/repos/maven2-snapshot/
 -Dmaven.test.skip=true

With those deployments the repository is prepared to compile the m3server project...

Maven 3: SSH deployment to foreign repository

noreply@blogger.com (Unknown) — Sat, 31 Dec 2011 15:08:00 +0000

Using the maven-deploy-plugin to deploy artifacts to a repository that is not the original target repository of a project is getting complicated with Maven 3 when the wagon SSH provider is not part of the Maven distribution.

The following JAR files must be added to the Maven distribution into the lib/ext extension folder:

jsch-0.1.44-1.jar
plexus-utils-3.0.jar
plexus-interactivity-api-1.0-alpha-6.jar
wagon-ssh-2.2.jar
wagon-ssh-external-2.2.jar
wagon-ssh-common-2.2.jar

Then the parent pom.xml of the project must be extended by the following lines:

<build>
<plugins>
<plugin>
<groupid>org.apache.maven.plugins</groupid>
<artifactid>maven-deploy-plugin</artifactid>
<dependencies>
<dependency>
<groupid>org.apache.maven.wagon</groupid>
<artifactid>wagon-ssh</artifactid>
      <version>2.2</version>
      </dependency>
</dependencies>
    </plugin>
</plugins>
</build>


The settings.xml contains the user credentials and looks like this:

   <servers>
      <server>
         <id>myrepo-snapshots</id>
         <username>username</username>
         <privatekey>${user.home}/.ssh/id_rsa</privatekey>
         <passphrase>key-passphrase</passphrase>
         <filepermissions>666</filepermissions>
         <directorypermissions>777</directorypermissions>
      </server>
   </servers>

Then the deploy phase can be executed like this:

mvn deploy -DaltDeploymentRepository=myrepo-snapshots::default::scp://host:port/var/www/repos/maven2-snapshot/

Building weld-osgi

noreply@blogger.com (Unknown) — Tue, 06 Dec 2011 22:11:00 +0000

In case you want to use Weld inside your OSGi application you first need to compile a few branches on your own:

WELD-963 branch of the weld/api project
weld-osgi branch from the weld/core project

Those branches will soon be merged into the official Weld codebase. But in the meantime you need to follow the steps below in order to get the packages compiled.
A prerequisite is that git and Maven 3 is available

git clone -b WELD-963 http://github.com/mathieuancelin/api.git WELD-963

cd WELD-963

mvn clean install

cd ..

git clone -b weld-osgi http://github.com/mathieuancelin/core.git weld-osgi

cd weld-osgi

mvn clean install -Dmaven.test.skip=true

find . -name "weld-osgi-core*SNAPSHOT.jar" -exec cp {} . \;

The last command above copies the following files into your current directory:

weld-osgi-core-api-1.1.5-SNAPSHOT.jar
weld-osgi-core-extension-1.1.5-SNAPSHOT.jar
weld-osgi-core-integration-1.1.5-SNAPSHOT.jar
weld-osgi-core-mandatory-1.1.5-SNAPSHOT.jar
weld-osgi-core-spi-1.1.5-SNAPSHOT.jar

You can now deploy those packages into your local repository so that they can be used in your own project's pom.xml.

mvn deploy:deploy-file -DgroupId=org.jboss.weld.osgi \

                       -DartifactId=weld-osgi-core-api \

                       -Dversion=1.1.5-SNAPSHOT \

                       -Dpackaging=jar \

                       -Dfile=weld-osgi-core-api-1.1.5-SNAPSHOT.jar

mvn install:install-file -DgroupId=org.jboss.weld.osgi  \

                         -DartifactId=weld-osgi-core-extension  \

                         -Dversion=1.1.5-SNAPSHOT \

                         -Dpackaging=jar \

                         -Dfile=weld-osgi-core-extension-1.1.5-SNAPSHOT.jar
mvn install:install-file -DgroupId=org.jboss.weld.osgi  \

                         -DartifactId=weld-osgi-core-integration  \

                         -Dversion=1.1.5-SNAPSHOT \

                         -Dpackaging=jar \

                         -Dfile=weld-osgi-core-integration-1.1.5-SNAPSHOT.jar

mvn install:install-file -DgroupId=org.jboss.weld.osgi  \

                         -DartifactId=weld-osgi-core-mandatory  \

                         -Dversion=1.1.5-SNAPSHOT \

                         -Dpackaging=jar \

                         -Dfile=weld-osgi-core-mandatory-1.1.5-SNAPSHOT.jar

mvn install:install-file -DgroupId=org.jboss.weld.osgi  \

                         -DartifactId=weld-osgi-core-spi  \

                         -Dversion=1.1.5-SNAPSHOT \

                         -Dpackaging=jar \

                         -Dfile=weld-osgi-core-spi-1.1.5-SNAPSHOT.jar

The pom.xml of your assembly package might look like this:

        <dependency>
            <groupId>org.jboss.weld.osgi</groupId>
            <artifactId>weld-osgi-core-api</artifactId>
            <version>1.1.5-SNAPSHOT</version>
            <scope>runtime</scope>
        </dependency>
        <dependency>
            <groupId>org.jboss.weld.osgi</groupId>
            <artifactId>weld-osgi-core-extension</artifactId>
            <version>1.1.5-SNAPSHOT</version>
            <scope>runtime</scope>
        </dependency>
        <dependency>
            <groupId>org.jboss.weld.osgi</groupId>
            <artifactId>weld-osgi-core-integration</artifactId>
            <version>1.1.5-SNAPSHOT</version>
            <scope>runtime</scope>
        </dependency>
        <dependency>
            <groupId>org.jboss.weld.osgi</groupId>
            <artifactId>weld-osgi-core-mandatory</artifactId>
            <version>1.1.5-SNAPSHOT</version>
            <scope>runtime</scope>
        </dependency>
        <dependency>
            <groupId>org.jboss.weld.osgi</groupId>
            <artifactId>weld-osgi-core-spi</artifactId>
            <version>1.1.5-SNAPSHOT</version>
            <scope>runtime</scope>
        </dependency>

m3player version 0.1.3 released

noreply@blogger.com (Unknown) — Fri, 19 Aug 2011 23:45:00 +0000

m3player version 0.1.3 has the following new features:

m3player can be run as a daemon with -d or --daemonise
The name of the UPnP Media Renderer can be configured by -n or --name. This is done by copying the XML files from /usr/share/m3player to /var/lib/m3player and filtering place-holders in some of the files.
The version of the MediaRenderer device can be configured by -r or --root. The default is MediaRendererV2.xml
The UPnP server can be bound to a network interface, specified by -i or --interface. Example: eth0

The help output shows an overview of all the parameters available:

$ ./m3player -h

Usage:

  m3player [OPTION...]

Help Options:
-h, --help          Show help options

Application Options:
-c, --config        Path to the config file
                      (Default: /etc/m3player/m3player.ini)
-p, --pid           Path to the pid file
                      (Default: /var/run/m3player.pid)
-s, --xmlsource     Path to the XML source folder
                      (Default: /usr/share/m3player)
-t, --xmltarget     Path to the XML target folder
                      (Default: /var/lib/m3player)
-n, --name          The name of the player instance
                      (Default: HOSTNAME)
-r, --root          The name of the root device file
                      (Default: MediaRendererV2.xml)
-l, --log           The name of the log file
                      (Default: /var/log/m3player.log)
-i, --interface     The name of the interface the process
                      is bound to (e.g. eth0)
-d, --daemonise     Fork the player as daemon

Turning m3player into a daemon

noreply@blogger.com (Unknown) — Thu, 11 Aug 2011 19:29:00 +0000

As most of the public HOWTOs seem to be either incomplete or a bit outdated and because I have no copy of the book Advanced Programming in the UNIX Environment, I was happy to find the following microHOWTOs:

By following the steps, described there I was able to turn m3player into a daemon. The daemon behavior can be turned on by passing the -d or --daemon switches to the command line. When the daemon switch is specified, all output (stdout and stderr) gets redirected into the log file at the default location /var/log/m3player.log or to the location, specified by the -l or --logfile switch.

m3player version 0.1.2 released

noreply@blogger.com (Unknown) — Sun, 31 Jul 2011 23:22:00 +0000

I just released version 0.1.2 of m3player. The new version includes:

Compilation and packaging for OpenWRT
Installation of the ipkg package on the TL-WR1043ND device
Support for command-line parameters:

config - Path to the config file (/etc/m3player/m3player.ini)
pidfile - Location of the PID file (/var/run/m3player.pid)
xmlfolder - Location of the descriptor XML files (/usr/share/m3player)
name - Set the name of the player instance (Hostname or m3player)
filename - The name of the root device file (MediaRenderedV1.xml)

Support for an INI file (/etc/m3player/m3player.ini)
Fixes to the GStreamer pipeline to work in the stripped down OpenWRT environment with only a few plugins

When launching the m3player it starts playing the first entry from the preset list.

The next version will include support for preset switching via the QSS button.

Building a TP-Link TL-WR1043ND OpenWRT 10.03 image for MP3 audio playback

noreply@blogger.com (Unknown) — Sun, 31 Jul 2011 21:48:00 +0000

This article gives a step-by-step description on how to get the m3ddity-player working on the TP-Link TL-WR1043ND device, using the stable version of OpenWRT 10.03 (backfire).

First you need to checkout the OpenWRT 10.03 (backfire) branch:

svn co svn://svn.openwrt.org/openwrt/branches/backfire

cd backfire 

Then the feed for the m3ddity-player must be added:

cp feeds.conf.default feeds.conf

echo "src-svn jcoderz http://www.jcoderz.org/svn/m3dditiez/branches/openwrt-10.3/packages" >> feeds.conf

./scripts/feeds update ./scripts/feeds install -a  

Now run menuconfig and select the necessary modules:
make menuconfig

The following list shows only menuconfig entries where settings have been changed. Settings not shown here are left untouched.

Target System (Atheros AR71xx/AR7240/AR913x)
Target Profile (TP-Link TL-WR1043ND v1)

Global build settings --->
[*] Remove ipkg/opkg status/data files in final images
[ ] Enable ipv6 support in packages

[*] Build the OpenWRT based toolchain
[*] Image configuration --->
    (192.168.0.5) LAN DNS server
    (192.168.0.5) LAN Gateway
    (255.255.255.0) LAN Network Mask
    (192.168.0.111) LAN IP Address

Base system
<*> block-mount

< > dnsmasq

< > firewall

Kernel Modules

Filesystems

<*> kmod-fs-ext3 (for USB drive)

<*> kmod-nls-utf8

Other modules

    <*> kmod-ledtrig-netdev

Sound Support
    <*> kmod-sound-core

USB Support
    <*> kmod-usb-core
    <*>   kmod-usb-audio
    <*> kmod-usb-ohci
    <*> kmod-usb-storage (for USB drive)

Network

< > ppp

< > iptables

Time Synchronization
<*> ntpdate

Multimedia

<*> gst-mod-alsa

<*> gst-mod-mad

<*> gst-mod-souphttpsrc

<*> gstreamer

<*> m3ddity-player (only with jCoderZ.org modules)

Utilities

<*> alsa-utils (for alsamixer/alsactl)

After the configuration step is completed, the image can be compiled with the following command:
make

The next steps assume that the router is already running a stable version of OpenWRT (flashed initially via the original vendor UI). If that is the case you can now connect to the router by running:
telnet 192.168.0.111 # The IP configured during menuconfig

Now copy the new image to a location on the router with enough space, for example a USB stick (/dev/sda1):
mkdir /mnt/usb
mount -t ext2 /dev/sda1 /mnt/usb
cd /mnt/usb
scp USERNAME@HOST:HOME/backfire/bin/ar71xx/openwrt-ar71xx-generic-tl-wr1043nd-v1-squashfs-sysupgrade.bin .

Now drop the caches:
echo 3 > /proc/sys/vm/drop_caches

Then reflash the image with the following command:
mtd -r write openwrt-ar71xx-generic-tl-wr1043nd-v1-squashfs-sysupgrade.bin firmware

The router will reboot and you can connect again after reboot:
telnet 192.168.0.111

Replace the USB stick with the USB audio device and run the m3ddity-player with the following command:
m3player

Now you should hear a sound as the m3player starts playing one of its presets from the config file /etc/m3player/m3player.ini.

Music player fully integrated into the OpenWRT build process

noreply@blogger.com (Unknown) — Sun, 24 Jul 2011 22:44:00 +0000

The m3ddity-player and the GUPnP libraries, on which the player depends upon, can now easily be build for OpenWRT, just follow the steps below:

1. Checkout the trunk version of OpenWRT

svn co svn://svn.openwrt.org/openwrt/trunk

cd trunk

2. Add the jCoderZ package repository to the list of feeds

cp feeds.conf.default feeds.conf
echo "src-svn jcoderz " \
"http://www.jcoderz.org/svn/m3dditiez/trunk/" \
"openwrt/packages" >> feeds.conf
./scripts/feeds update
./scripts/feeds install -a

3. Run menuconfig and select the packages as described here

make menuconfig

4. Additionally select the package Multimedia/m3ddity-player. This automatically selects the necessary GUPnP libraries.

5. Finally, you can build the image

make -j 5

But be warned, the trunk version of OpenWRT might be broken and flashing the image might result in a bricked device that can only be restored via the serial console. For the TL-WR1043ND, this even involves soldering!

jCoderZ.org

noreply@blogger.com (Unknown) — Sun, 24 Jul 2011 11:31:00 +0000

The OpenWRT music player, which I have blogged about recently, has found a home at https://github.com/jCoderZ/m3player

Quest for a USB audio device part 2...

noreply@blogger.com (Unknown) — Tue, 19 Jul 2011 18:21:00 +0000

The new audio devices arrived but unfortunately they show the same glitches as the devices I tested before:

1. USB Soundkarte mit Virtual 3D Soundeffekt LogiLink®

This device is also based on the chipset Tenx Technology, Inc. TP6911 Audio Headset and has the same issues as the previous device (USB Sound Audio), i.e. the output is far too loud.

2. USB Soundkarte mit Virtual 7.1 Soundeffekt LogiLink®

The sound is good but as with the previous device a clicking noise is heard during reboot, power-off, power-on, plug-in/out. The chipset is C-Media Electronics, Inc. Audio Adapter

So both adapters use the same chipsets and show the same issues. It seems as if there are not too many different cheap USB audio devices on the market, just rebranded stuff from the same manufacturer.

... to be continued!

Seeking a cheap USB audio device that works with the TP-Link TL-WR1043ND under OpenWRT

noreply@blogger.com (Unknown) — Wed, 13 Jul 2011 20:48:00 +0000

I'm seeking for a USB audio device that works with the TP-Link TL-WR1043ND under OpenWRT. Until now I tried 2 different devices with unacceptable issues.

1. USB Sound Audio (Amazon)

The device gets detected under Linux as Tenx Technology, Inc. TP6911 Audio Headset. The following warnings are reported by dmesg when connecting the device:
usb_audio: Warning! Unlikely big volume range (=13568), cval->res is probably wrong.
usb_audio: [2] FU [PCM Playback Volume] ch = 2, val = 2560/16128/1
usb_audio: Warning! Unlikely big volume range (=6096), cval->res is probably wrong.
usb_audio: [5] FU [Mic Capture Volume] ch = 1, val = 0/6096/1

The issue with this device is that the volume level is so high that it cannot be controlled. Between mute and 10% the output sounds distorted. At 20% the volume level is at its maximum, with no audible change between 20% and 100%.

2. auvisio Visual Sound Technologies (pearl.de)

The device gets detected as C-Media Electronics, Inc. Audio Adapter.

The issue with the device is that a very loud clicking sound gets played when the OpenWRT router starts-up or during shutdown/power-off. The sound is quite loud and it will be very annoying when using the router as a music-player.

Next

Both devices are unsuitable for using them with the router. The next try will be with the following devices...

Make ALSA settings persistent on OpenWRT

noreply@blogger.com (Unknown) — Wed, 13 Jul 2011 19:36:00 +0000

The command-line tool alsamixer can be used to control the volume of the USB audio output:

Unfortunately the setting does not survive a reboot. Another tool from the alsa-utils package can be used for this purpose: alsactl

After setting the volume with alsamixer, the following command persists the current setting in the file-system for device 0:
alsactl -f /etc/alsa/alsa0.state store 0

During startup, the state must be restored again. This can be achieved by adding the following command to /etc/rc.local:
alsactl -f /etc/alsa/alsa0.state restore 0

Building an OpenWRT image optimized for MP3 audio playback

noreply@blogger.com (Unknown) — Wed, 13 Jul 2011 18:36:00 +0000

For the audio-renderer, see revious posts, I decided to switch to the trunk version of OpenWRT.
I experienced some issues with the USB audio hardware and the trunk version should prove that there are no issues with the USB audio drivers which are already fixed.
The following configuration is for the bleeding-edge version, some module names have changed so it cannot be applied to the Kamikaze (10.3) version without modification (e.g. block-mount contains the overlay functionality and kmod-fs-ext4 has support for all ext file-system versions).
NOTE: This configuration has stripped all modules which are not necessary for USB audio and Wireless LAN.

Target System: Atheros AR71xx/AR7240/AR913x/AR934x)
Target Profile: TP-Link TL-WR1043ND v1

Global build settings --->
[*] Remove ipkg/opkg status/data files in final images
[ ] Enable ipv6 support in packages

LATER: Strip unnecessary exports from kernel image
LATER: Strip unnecessary functions from libraries

[*] Build the OpenWRT Image Builder
[*] Build the OpenWRT SDK
[*] Build the OpenWRT based toolchain
[*] Image configuration --->
    LAN Gateway: 192.168.0.5
    LAN Network Mask: 255.255.255.0
    LAN IP Address: 192.168.0.111

Base system
<*> block-mount

< > dnsmasq

< > firewall

Kernel Modules

Filesystems

<*> kmod-fs-ext4 (for USB drive)

Native Language Support
<*> kmod-nls-utf8

Sound Support
    <*> kmod-sound-core
    <*> kmod-usb-audio

USB Support
    <*> kmod-usb-ohci
    <*> kmod-usb-storage (for USB drive)

Other Modules (not sure yet whether they are needed)
??? *gpio* ??? (Buttons for preset-switching ???)

Network

< > ppp

Firewall

< > iptables

Time Synchronization
<*> ntpclient

Multimedia

<*> gst-mod-alsa

<*> gst-mod-mad

<*> gst-mod-souphttpsrc

<*> gstreamer

<*> gstreamer-utils (for gst-* tools)

Utilities

<*> alsa-utils (for alsamixer/alsactl)

After the configuration step is completed, the image can be compiled with the following command:

make -j 5 # on a quad-core machine

With the configuration above the compiled image is 5248kB large, which means 2944kB are free for modules and configuration in the /overlay file-system.
Then you can connect to the router with the following command (assuming that OpenWRT is already installed):

telnet 192.168.0.111

Copy the new image to a location on the router with enough space (I use a 32MB USB stick with ext2 file-system which I mount under /mnt/usb):

scp USERNAME@HOST:HOME/openwrt-trunk/bin/ar71xx/openwrt-ar71xx-generic-tl-wr1043nd-v1-squashfs-sysupgrade.bin .

Now drop the caches:

echo 3 > /proc/sys/vm/drop_caches

Then reflash the image with the following command:

mtd -r write openwrt-ar71xx-generic-tl-wr1043nd-v1-squashfs-sysupgrade.bin firmware

The router will reboot and when connecting via telnet again, the setup can be tested with the following command:

gst-launch souphttpsrc location=http://87.248.219.60/stream/swr_mp3_m_swr1bwa ! mad ! alsasink

Et viola, the USB audio device will play the radio station! Hopefully..... ;)

Using a password safe on multiple platforms

noreply@blogger.com (Unknown) — Sat, 09 Jul 2011 13:40:00 +0000

After the Playstation Network Hack I decided to change to secure passwords for all my accounts. The drawback is that nobody can remember secure passwords anymore, so there is no way around using a password safe.

The only application that is available on all the different platforms I use is KeePass (V1.x on Windows, KeePassX under Linux, and KeePassDroid on Android). Under Windows the the 1.x version must be used in order to stick to the same file format as the KeePassX and the KeePassDroid applications.

Using one database file on many different devices has proven to be a challenge as the file must be located where it can be accessed from all the different devices.

My first approach was to use a WebDAV folder which is available from hosters like web.de or 1&1 in Germany.
The Linux support for WebDAV is good, dav2fs does what I was looking for and I found absolutely no issues while using it.
Unfortunately Windows 7 64Bit support for WebDAV is severly broken. Access to the online folder was working a few days but it stopped working, being unable to connect anymore (Error 0x80070043). All the hints and workarounds I found did not work for my business notebook. I also tried the different WebDAV client tools like NetDrive for example. None of them was working reliably in a corporate environment behind a proxy and I did not want to find workarounds for the issues of the WebDAV clients.
There is also no free WebDAV support available on Android, making the protocol unusable on my phone and tablet. The consequence would be either to copy the file manually or to extend KeePassDroid in order to access databases via HTTPS/WebDAV.

After a few weeks of pain when trying to log into an account with a secure password I decided to give DropBox a try. Clients are available for all the devices I have and adding an account is very easy and can be setup in under a minute.

DropBox has been configured to create a new folder and not to sync the whole user's home folder:

Windows: C:\Users\(USERNAME)\Dropbox
Linux: /home/(USERNAME)/Dropbox
Android: /mnt/sdcard/dropbox

On all platforms the extension "kdb" has been associated with the KeePass application, so opening the DropBox folder and double-clicking/pressing on the kdb file opens KeePass/KeePassX/KeePassDroid which asks for the database password.

So far DropBox seems to be the solution to my problem. I can access the database with only 2 passwords to remember, the one for the password safe file and the other one for the DropBox account.

However, one issue with concurrent access to the same KeePass file remains. The KeePass application writes a lock file to signal that the database is being used by an instance of the application. Different instances create and delete the lock file so that the file is completely useless in the scenario, described above. The message that a lock file has been found and whether the database should be opened read-only must either be ignored or followed. When followed new entries can be added on one device only.

When you add an entry to the database on one platform this change is propagated to the other devices because of the DropBox sync. Unfortunately running instancs of KeePass do not recognize that a file has changed in the DropBox folder. Only when you add an account there and you try to save the database file you get a message that the file in the file-system has changed.
The

The solution to this issue would be that KeePass reloads the file on a regular basis as long as no entries have been added. In case a user has added an entry KeePass could do a merge of the two database files by comparing entries one by one and adding new ones to the loaded database before persisting it again.

As long as this has not been implemented the only work-around is to do some configuration on the clients so that the application/database is likely to be closed each time you switch the device.
One setting is not minimize the application window into the system tray. That keeps the window open and reminds you, when you, for example, cleanup the desktop of your business notebook in evening, to close the application. The next morning you need to reopen the database and load all entries that have been added last evening on your other devices.

A small step into the merge direction would be to notify the user when the database in the file-system has changed. If such a notification pops-up the database can be closed and re-opened again.

I have created feature requests for each of the three projects:

KeePass: 3362189 (Closed, Rejected: http://keepass.info/help/base/multiuser.html)
KeePassX: 3362184
KeePassDroid: 206

Radio presets

noreply@blogger.com (Unknown) — Mon, 27 Jun 2011 21:26:00 +0000

One important feature for the audio renderer is the support for presets.

I do not want to search for my mobile phone in order to select a music source when I enter the kitchen in the morning. The use-case where you turn on your analog radio and the tuning wheel is right at the position of your favorite radio station is unbeatable simple.

Therefore the audio-renderer has a list of MP3 streaming URLs where the first one will be played upon startup of the process. The channels can be switched by sending a signal to the server process:

kill -USR1 <PID of audio-renderer>

This is a good way to hijack an unused button of your router (for example the QSS button of the TP-Link WR1043ND) and write a simple script as shown here: http://wiki.openwrt.org/doc/howto/hardware.button

There is currently no support for M3U files as this is not supported by Gstreamer. The MP3 streams must be extracted manually from the M3U files.

UPnP audio renderer released

noreply@blogger.com (Unknown) — Mon, 27 Jun 2011 21:05:00 +0000

A very early version of my UPnP audio-renderer had been released on Gitorius https://gitorious.org/gupnp-audio-renderer (Update 2011-08-01: It has been renamed to m3player and been moved to here). It implements the necessary parts of the UPnP MediaRenderer specification.

The implementation is based on excellent frameworks like:

The initial target for the renderer is the OpenWRT platform (see previous articles about a WiFi music player). I decided to use C instead of Vala (that was a decision against Rygel) as the memory constraints are very limited and I need to save memory whereever possible. Although my C language skills are a bit rusty, I did not want to learn another language. Additionally, the usage of Vala would increase the effort to port the whole vala stuff to the OpenWRT build-system. Luckily GStreamer and GLib are already available on OpenWRT.

Update 2011-07-31: Porting Vala packages to OpenWRT seems to be less complex as many source tarballs already contain the C source, generated by the valac pre-compiler, so no valac must be available during the OpenWRT build process.

The requirements for the implementation are:

audio-only renderer (usually routers do not have displays) with a focus on MP3
small memory footprint (my router has only 32MB available)
easy porting to different platforms (not dependent to where Vala has been ported)
support for presets (which can be switched by pressing a button on the router device)

The renderer is being tested with the UPnP Control Point UPnPlay on my HTC Legend.

The next steps are:

Rounding the rough edges
Packaging for Debian/Ubuntu
Makefiles for OpenWRT
Cleaning up the codebase

Update 2011-07-31: The gupnp-audio-renderer has been renamed to m3ddity-player and it found a home here.

Compile a development version of GUPnP from HEAD

noreply@blogger.com (Unknown) — Mon, 13 Jun 2011 15:53:00 +0000

Follow the commands below in order to setup a local environment for working with the latest development version of the GUPNP framework.

mkdir $(pwd)/base

git clone http://git.gnome.org/browse/gssdp
cd gssdp
./autogen.sh --prefix=$(pwd)/../base --exec-prefix=$(pwd)/../base
make && make install
cd ..

git clone http://git.gnome.org/browse/gupnp

cd gupnp

PKG_CONFIG_PATH=$(pwd)/../base/lib/pkgconfig ./autogen.sh --prefix=$(pwd)/../base --exec-prefix=$(pwd)/../base

cd ..

git clone http://git.gnome.org/browse/gupnp-tools
cd gupnp-tools

PKG_CONFIG_PATH=$(pwd)/../base/lib/pkgconfig ./autogen.sh --prefix=$(pwd)/../base --exec-prefix=$(pwd)/../base

cd ..

Run the following commands for setting the environment to use the fresh compiled libraries instead of the OS provided ones (Empathy under Ubuntu 11.04 uses an older version of the GUPNP libraries. I ran into segmentation faults because the libraries did not match):

export PATH=$(pwd)/base/bin:$(pwd)/gupnp/examples:$PATH

export LD_LIBRARY_PATH=$(pwd)/base/lib:$LD_LIBRARY_PATH

export PKG_CONFIG_PATH=$(pwd)/base/lib/pkgconfig:$PKG_CONFIG_PATH

Now you can execute several examples or tests to check whether everything works as expected:

light-server &

gupnp-universal-cp &

The GUI application GUPnP Universal Control Point shows an entry "Kitchen Lights", which is provided by the light-server application started above.

Building a custom OpenWRT image for the TP-Link TL-WR1043ND

noreply@blogger.com (Unknown) — Sun, 12 Jun 2011 14:36:00 +0000

For my low-budget wireless music player I had to tweak OpenWRT in order to support my setup. For initial tests I decided to use the Music Player Daemon and madplay.

The following steps describe how to setup OpenWRT on the TP-Link WR1043ND. It assumes that the OpenWRT 10.03.1-rc4 firmware has already been installed, that the IP address of the router has been set to 192.168.0.111, and that an USB audio, as well as that an USB stick has been connected to the device via an USB hub.

Checkout the 10.03 branch (currently 10.03.1-rc5)
$ svn co svn://svn.openwrt.org/openwrt/branches/backfire
$ cd backfire

Download and install all packages:
$ ./scripts/feeds update -a
$ ./scripts/feeds install -a

Run menuconfig and select a few more items so that they are available without the need to install packages first:
$ make menuconfig
-> Target Profile: TP-Link TL-WR1043ND
-> [*] Build the OpenWrt Image Builder
-> [*] Build the OpenWrt SDK
-> [*] Build the OpenWrt based Toolchain
-> [*] Image Configuration
() LAN Gateway:
(192.168.1.1) LAN Network Address: 192.168.0.111
Exit
-> Base System
<*> block-hotplug
<*> block-extroot
Exit
-> Kernel Modules
-> Filesystems
<*> kmod-fs-ext2
Exit
-> USB Support
<*> kmod-usb-core
<*> kmod-usb-audio
<*> kmod-usb-ohci
<*> kmod-usb-storage
<*> kmod-usb-storage-extras
<*> kmod-usb-usb2
Exit
-> Sound
mpd
mpc
madplay
Exit
Exit

On quad core machines you can run make with 5 parallel jobs:
$ make -j 5

Install the new image by running the following commands:
$ telnet 192.168.0.111
$ scp USER@HOST:HOME/workspaces/openwrt/backfire/bin/ar71xx/openwrt-ar71xx-tl-wr1043nd-v1-squashfs-sysupgrade.bin .
$ cd /tmp
$ mtd -r write openwrt-ar71xx-tl-wr1043nd-v1-squashfs-sysupgrade.bin firmware

Because WLAN will not be configured after flashing a new image, you need a cable connection.
After the reboot is finished you can connect to the device like this:
$ telnet 192.168.0.111

Because mpd and madplay consume more space than the router has available on its root partition, the extroot feature must be enabled. Enable extroot for the USB stick by modifiying fstab:
$ vi /etc/config/fstab
================================
config mount
option target /overlay
option device /dev/sda1
option fstype ext2
option options rw,sync
option enabled 1
option enabled_fsck 0
================================

Reboot the device again so that the extroot overlay becomes available:
$ reboot

Connect to the device again:
$ telnet 192.168.0.111

I have setup an Apache on my desktop machine in order to easily provide the packages to the device. For using the local HTTP server, opkg must be configured like this (my desktop machine has IP 192.168.0.17):
$ vi /etc/opkg.conf
================================
src/gz packages http://192.168.0.17/packages
...
================================

When this is done you can update the package list:

$ opkg update

Now the packages mpd, mpc, and madplay can be installed:
$ opkg install mpd mpc madplay

The mpd daemon needs some special folder setup:
$ mkdir -p ~/.mpd/playlists

Make sure mpd starts when the device boots up by adding the follownig line to rc.local:
$ vi /etc/rc.local
================================

mpd /etc/mpd.conf

================================

In oder to disconnect the wired LAN connection, you need to setup WLAN:
$ vi /etc/config/wireless

Change the following entries:

================================

config wifi-device radio0
...
# REMOVE THIS LINE TO ENABLE WIFI:
#option disabled 1

config wifi-iface
...
option mode sta # Client mode
option ssid mySSID # Your SSID
option encryption psk2 # WPA-PSK2
option key 12345678901234567890 # Your network key
================================

Configure DHCP so that the device asks the DHCP server for an IP address:
$ vi /etc/config/network

Change the lan interface (eth0.1) like this:

================================

config 'interface' 'lan'
option 'ifname' 'eth0.1'
option 'proto' 'dhcp'
================================

Now you can start WiFi like this:
$ wifi up

After that you can unplug the LAN cable. The router should be connected to your WLAN and it should acquire an IP address from your local network DHCP server. Make sure that you do not gave a MAC Filter enabled in case it does not work ;)

For an intial test you can updload a MP3 file to the device:
$ scp USER@HOST:HOME/test.mp3 /tmp

With madplay the MP3 can be played like this:
$ madplay /tmp/test.mp3

With the following script running on your desktop machine you can make the Music Play Daemon play a Shoutcast audio stream:
$ export MPD_HOST=192.168.0.111
$ mpc add http://89.179.179.5:8040
$ mpc play