Im develop branch laufen alle aktuellen Entwicklungen zusammen, z. B. wenn ein Entwickler meint, er habe sein feature fertig: es wird zurück in den develop branch gebracht. Vorher hat der Entwickler (weil wir ja gern TDD machen) sichergestellt, dass das feature eine ausreichende Unit Test Abdeckung hat. Für ihn persönlich sollte es also ausreichend sein auf diese zu achten. Eine persönliche CI für sein feature wäre sicherlich ein adminitrativer overkill: ist das feature klein wird es sicher schnell fertig, dafür extra einen neuen CI job aufsetzen und nach Beendigung der feature Entwicklung wieder löschen?! Ne!

Der release branch benötigt aus anderen Gründen keine CI: er unterliegt nur einer kurzen Zeit wirklicher Entwicklungsarbeit: während der Erstellung eines Release.

Wichtiger ist dann schon der master branch. Er enthält den Source Code der aktuell in der Produktion verwendet wird. Aber: gegen ihn werden auch hotfixes durchgeführt! Also Änderungen am Produktions Kode. Diese sollten frühzeitig mit CI abgedeckt werden, jeder hotfix branch sollte also einen eigenen CI job haben. Ist der hotfix gut und es werden keine Integrationsfehler (mit dem Produktions Kode) mehr festgestellt, dann ist der hotfix fertig in den master branch überzugehen (und in develop).

The weapon of choice for near real time information delivery is obviously XMPP. I will use publish-subscribe to provide information, by doing so any of my peers may or may not subscribe to a node onto which I publish information. And I will set up a hierarchy of nodes onto which I will publish different kinds of information. So here we go with the root node: home/b4mad.net/goern, this is mainly due to the fact that I use ejabberd2 to run my XMPP Server. There are a few subnodes:

• bookmark – for bookmarks that I send out, I need to figure out the format of published information
• micropublish – for twitter like things, that is basically Atom over XMPP
• media/stillimage – for a photostream…

As a good semantic web citizen I will note down what nodes exist at which URI and what they are used for. The first rough scratch is available.

For the moment there is a shell like Java application which accepts two commands: exit and purge, to exit the shell and purge off all items from the micropublish node. Any other words typed in the shell will be treated a an micropublish update and published as a Atom entry to my micropublish node.  The second application is a gateway to twitter.com: it is subscribed to my pubsub micropublish node and will pick up any atom entry and post it as a twitter update.

The next thing it write up is a process and format to publish my bookmarks and to have some nice GUI clients that can handle my updates… Or should it be integrated into WorkPress?!

I like to reuse, so that’s the first policy.
The second policy is, that I dont want to deal with the email archives, but will use RSS or Atom representation of the archive as the source of information.

How to understand the information provided by mailing list archives in a SIOC context?

First of all, the mailing list archive is obviously a sioc:Forum (archive itself represented by a RSS or Atom feed) with sioc:Posts (individual mails, RSS/Atom items). The sender of the mail is a sioc:User, which is the sioc:has_creator of the sioc:Post. The website hosting the mailing list archive is defined as the sioc:Site, the initialy mentioned sioc:Forum sioc:has_host this sioc:Site. Got it?

As a test subject I took the SIOC-dev mailing list at Google Groups, development is supported by RDFReactor.

Results of my work can be found 1. as Java code and 2. as the resultung SIOC data.

Comments and any other feedback is welcome!

ikvmc commons-logging.jar log4j-1.2.12.jar  -out:log4j.dll -target:library
ikvmc stax-api-1.0.jar wstx-asl-2.8.jar xercesImpl.jar xml-apis.jar \
 resolver.jar -out:xml.dll -target:library
ikvmc antlr-2.7.5.jar arq.jar concurrent.jar icu4j_3_4.jar \
jakarta-oro-2.0.8.jar jena.jar -reference:xml.dll -reference:log4j.dll \
 -out:arq.dll -target:library

A short test if the Jena classes are working was derived from an example of the Jena documentatio and worked out just fine… replacing import with using and omitting the trailing .* gave quick working results.

The hard part starts if it comes to persistent Jena Models. I am using MySQL, so I ikvmced the Java MySQL 3.2alpha driver and used the following code segment as learned from the docs:

  System.Activator.CreateInstance("mysql32alpha", 
     "com.mysql.jdbc.Driver");

  IDBConnection idbcon = new  DBConnection("jdbc:mysql://db-server-2/jena23", 
     "jdbc", "password", "MySQL");

mysql32alpha is the name of the assembly where the class com.mysql.jdbc.Driver is to be found. Compiling the code went well, running it resulted in Exception: Failure to instantiate DB Driver:MySQL java.lang.NullPointerException. So, I droppt the ikvm developers a line… awaiting an answer. Hang on for updates

UPDATE 1: With the devine intervention of “reinstall and configure your tool chain” I did that, currently running:

goern@node-236:~/src/mono/ARQ-1.2$ ikvm -version
CLR version: 1.1.4322.2032 (32 bit)
System: 1.0.5000.0
IKVM.Runtime: 0.22.0.0
IKVM.GNU.Classpath: 0.22.0.0
ikvm: 0.22.0.0
mscorlib: 1.0.5000.0
GNU Classpath version: 0.19
goern@node-236:~/src/mono/ARQ-1.2$ mono --version
Mono JIT compiler version 1.1.13.1, (C) 2002-2005 Novell, Inc and Contributors. www.mono-project.com
        TLS:           __thread
        GC:            Included Boehm (with typed GC)
        SIGSEGV      : normal

Redoing the build process I got the ARQ and MySql .dlls and my binary, compiled just fine and runs fine. JDBC Connection is opned up, Jena Model gets opened and all the content dumped to Console. Having all the good tools in place two point araise:

1. Determining the Language you use is hard! C# just looks like Java to me at the moment:

using System;
using com.hp.hpl.jena.rdf.model;
using com.hp.hpl.jena.db;

public class hodge {
   public static void Main() {
           IDBConnection idbcon = null;
           ModelMaker maker = null;
           Model model = null;

           System.Activator.CreateInstance("mysql32alpha",
            "com.mysql.jdbc.Driver");

           idbcon = new  DBConnection("jdbc:mysql://server/jena23", 
            "user", "pass", "MySQL");
           if (idbcon == null)
                   Console.WriteLine("no IDBConnection");

           try {
                   maker = ModelFactory.createModelRDBMaker(idbcon);
                   model = maker.openModel("pim");
           } catch (Exception e) {
                   Console.WriteLine("Exception: " + e.Message);
           }

           Console.WriteLine(model); // model.write(System.out);
   }
}

2. C# and Mono world seems to be open for Jena.

PS: I have attached a distribution that contains a nant build file for the tasks above and the source code, just configure at top of build.build.

PPS: I have not tested this under some MS Windows.

The GUI is pretty simple, you may get or set the current position of a user. The username (and if available) the passowrd may be set in the Preferences dialog. It you get the current position of a user the landmark combobox is updated with the landmarks of the user. That’s it, not much but a start.

next step: find foaf file location of user and display his menow status :)

The .jar contains the full source but no documentation jet…

TODO: write documentation.

I will implement the web service using a JBoss Application Server and a variant of J2EE technologies. TODO: figure out the url of the web services…

UPDATE 20051024: First steps have been successful completed: There is a Servlet that returns all trustsRegarding statement for a given foaf:mbox_sha1sum.

The archive of witw.scpt also contains widw.py and supporting files, just set your username and password in witw.scpt and go…

1. start a gateway between WITW webservice and a jabber pubsub node, this will look for updates on the jabber node and update WITW according to the informations received.
2. use a python program to update a jabber pubsub node, to let others know that my geo positions has been updated
3. and finaly take a look at where I am

Next Step: build a GUI to set the geo position, maybe using Python and wxPython

Erster Anruf geht zum echo service, erstaunlich gute Sprachqualität ist das Resultat, besser als bei einer X-Lite Installation und genau so gut wie bei Skype. Zweiter Anruf gilt meiner sipgate voice mailbox: geht auch gut, wenn man mal ausgefunden hat was die sipgate Vorwahl aus dem sipphone Netz ist: 1777, umgekeht ist es 000747 .

Fazit ist, dass ich einen neuen Versuch starten werde SIP zu nutzen, scheinbar hat sich sowohl die Bedienbarkeit eines clients als auch die Sprachqualität verbessert.

Will der Automator aber nicht… erstens kann er nicht alle Titel einer Wiedergabeliste abfragen, zweitens kann er nichts im iTunes löschen. Hier ist noch der klassische Raum für Verbesserungen.

The source for the new releae is here (I love here-links).

Next Step: a Gnome based desktop applet that subscribes to the XMPP node and displays data received from it.

UPDATE: there are some things going on, lost boy hacked up some WITW to GoogleMap service [via: Phil Wilson].

getUserAsHTML() – will return the given user’s current location

getUserAsMeNow() – will return the given user’s current location using the MeNow RDF Schema

getLandmarksAsXML() – get all landmarks of the given user

setUserLongLat(lon, lat) – set the given user’s long and lat

Please find the .tar.bz2 and test it…

After a few webpage readings I came up with “Wo in der Welt?” aka widw… a little python script picking up the current WITW info of a given user. As worldly wisdom showed once again brute force and ignorance helps! My next idea was to fetch the atom feed and push updates of locations into jabber. Unfortunatly the atom feed doesnt seem to be a Atom 0.3 feed, I “filed this bug”… so stay tuned for another step forward to my personal vision of an information desktop…

Next to give a shot to was Blam, configured and compiled without any error and installed to /usr/local/blam-1.6.1/ Starting the binary resulted in a error message (I guess) of Mono:

nostromo:/usr/local/blam-1.6.1 goern$ ./bin/blam 

Unhandled Exception: System.DllNotFoundException: libblam.dylib
in <0x000d0> (wrapper managed-to-native) Imendio.Blam.MessageConnection:bacon_message_connection_new (string)
in <0x00028> Imendio.Blam.MessageConnection:.ctor (string)
in <0x000e4> Imendio.Blam.Application:.ctor (string[],object[])
in <0x00054> Imendio.Blam.Application:Main (string[])

nostromo:/usr/local/blam-1.6.1 goern$ _

Anyone got an idea if this is related to Mono or Blam or the MacOSX dynamic linker or …

[posted with ecto]

Have a try on it.

PowerPC G4 (1.1)
CPU-Geschwindigkeit: 1.33 GHz
L2-Cache (pro CPU): 512 KB
Speicher: 768 MB
Busgeschwindigkeit: 167 MHz
Boot-ROM-Version: 4.8.3f1
Seriennummer: 4H445LC8PLW

Actually I can’t match that to anything on the overview page, but anyhow…

PS: maybe it is a 7457/7460?!

And here is the status:

• I have some Python code communication with the pubsub component at pubsub.ik.nu, it creates new nodes, subscribes to existing nodes…
• the desktop application subscribes to generic/B4mad/mood/goern to listen for MeNow updates published there
• the desktop application pops up new notifications as soon as it receives updates
• and I have a some test program to send updates

But where is the code you are talking about? Ahm, not yet released, just drop me a line if you are interested. Especialy the Mac users!! :)

coded using: Krieg from the album “Und Lautlos Fliegt Der Kopf Weg ” by Ben Becker