Recent blog entries for dannu

The current Samba and free software discussion are interesting. But I wished some more people knew of the award-winning theory of Christoph Spehr "Eine Grundlegung der freien Kooperation" german for "a foundation of free cooperations" or "collaboration".

One of the most fascinating things about this 60 page-paper is that he implicitly (and by the time unknowingly!) describes free software principles. Such as:

you can leave the group and receive no penalty

instead every "leaver" should be able to take something with him

it's fundamentally important that the negotiating partners in a free cooperation have equal negotiation power.

In a sentence: partners are not equal and they can even make strange "contracts" but it must be based on as much as equal negotiation power as possible.

A crucial question is whether freely collaborating groups can refuse to let an applicant join. The answer: yes they can! but it shouldn't lock out the person from satisfaction of basic needs.

it remains undiscussed how these principles apply to e.g. the current samba case.

Basically got an offer to work for a company where i quit working two years ago. They were spending insane amounts of money while refusing to come up with a reasonable concept how to succeed (i am not going into details here).

Now the points is: i could earn lots of money for a four-week job but i planned to work for xots, my m:n-collobaration websoftware and wanted to work on wandering network agents with jum. While the money is tempting these other three projects are far more interesting. hmmm.

Refactored some helper scripts for xots. Used QT-Designer to make a gui and pyuic to generate the python code (PyQT). works well. I am going to dive into this more and come up with a nice distributed tracing tool (for CORBA) and utilities for analyzing them in python/QT.

back to writing some nice python scripts! Discovered last night the very nice python-cookbook and started to improve some python scripts inspired by some snippets from there. Also tested the python-qt binding (compiled from the source for qt3.0). It seems great. Especially that you can use the QTDesigner and pyuic to generate python-qt templates. I liked the clean API of QT even in c++. Is there any better combination than python/qt? not swing/java you bet.

i had some discussions with my co-developers about the XA-Resource-DB-Implementation in our xots. Many Databases don't support so called thread-migration which is very ugly. it means that every thread which gets involved in a database transaction has to somehow call xa_end before the "prepare" of the 2PC-Protocol can be started. This makes it hard to choose the thread-model of an application freely. I really don't like this XA-stuff...

the last days have been occupied by playing starcraft with some friends. still one of the best strategy games.

now i am starting to do some work on the xots project again. I am still trying to reserve two month of time (beginning on february) to develop a scoop-like engine with zope and python. The sections will be called "channels" and arbitrary groups will be able to create their own channel and organize their communication. Also you can subscribe to several channels and get them all on your personalized page.

Also i am going to programm some "agents" together with jum to research the possibilities. Agent systems are networks where small (scripted or compiled) programms can move like a worm or virus and execute within the respective environments. E.g. you could use an agent to look throughout your network to find files including mime/type stuff. Program once/ run anywhere :-) this is going to be a very nice hacking session. every language with bytecode like java and python should be usable.

Just got back from the first CCC-congress day in berlin. Some nice talks (especially from Felix von Leitner with "diet libc") and interesting people to meet.

What i liked too was the political consensus on the current "anti-terrorist" laws passed in many countries. I wonder whether technical advances could outlaw these depressing developments. What i mean is that networks could evolve faster (with anonymity features) than it can be controlled. Maybe there is hope in that such technical advances can "outlaw" these direct actions for gaining control.

Christmas is great. Good time to do some programming if you don't have a family and many of your friends are gone :-)

The CORBA Transaction Service is slowly developing towards release time. Especially the newly introduced template CORBA helper library (xcorba) get's nicer every day. The biggest problem still is the Database XA-Adapters.