This week there is barely a day, where I am not bored with some trivia from the “embassy cables”. Now my first problem is that most of this is bloody obvious. Who would have thought that Mrs Merkel is risk-averse and rarely creative? Obviously the hacks from the Gruniad didn’t know or at least feel the need to tell me. Also it seems the American foreign policy is quite arrogant, ignorant, and at times stupid, but we even knew that they tortured Iraqi’s before Wikileaks imparted that knowledge upon us. So it’s not really news, let alone new information.

The second thing the annoys me, are the naïve justifications I hear for publishing rather private information. Freedom of Speech and the Press respectively are being cited as the legal basis the publications. Now as it happens there is more than one human right. Obviously there is a lot to be said for privacy on the other side. The Grundgesetz (arguably the best constitution in the world) comes with Artikel 10, which guarantees the privacy of mail and telecommunications. And this is not just a German quirk, Article 12 of the Universal Declaration of Human Rights talks about privacy of correspondence. People might argue that the US government shouldn’t enjoy human rights, but I think there is a point that it’s agents are human beings and should not have there correspondence published. The second problem is, that this sets a nasty precedence and the next time it could be someone’s private mail or documents.

The important here is that different rights need to be balanced, because exercising your rights might compromise other people’s rights. Funnily enough the lack of that insight strikes me as a very American trait, where Freedom of Speech is regularly being used to justify fascist hate speech and the Right to bear arms has cost many a man’s and woman’s life.

To sum up I don’t like the idea of wholesale publication of correspondence and I can’t see how any of the stuff I have read so far would justify this. So I would like to see Assange go behind bars, not for treason, because I don’t think an abstract entity like a government or state can have these rights, but for spreading people’s correspondence for no good reason.

This sequence diagram degenerated into a christmas tree shape. What is the problem with that? Well these methods operate on different layers. In the case of the SimpleFormController there is methods dealing with Requests and then there is methods dealing with FormBackingObjects. Different levels of abstraction get mixed into a single object (even if they are structured by the inheritance hierarchy). Also this is a point where I struggled a lot with idiomatic smalltalk code. This kind of code might be fine if you are a genius. If you are just a simple developer like I am, probably somethings like this might be much more intelligible to you:

Here each object lives only in a single layer. The same argument can also be used in the long standing debate over private methods, because the receiver of private methods is also self. Now I think there is definitely a case for calling some methods on self, but having a single instance more than twice or three times on your call stack is probably a bad thing.

What do I mean with that? Well being all eagerly object orientated our domain is nicely modeled in classes, whatever language we choose. In theory (that is what we have to tell ourselves) these are enforcing all kinds of contracts on our business objects. However taking a close look how all those frameworks work, we quickly realise, that the instances of these class are highly transient and live for about one request. If we look even closer we will also realise that people wrote loads of SQL scripts for migrating data, doing reporting and the likes, because doing it in the language of choice “doesn’t perform”. There goes your data integrity…

Contrast this with the way a lively object framework like seaside works. There the actual object instance exists throughout the runtime of the application. If you are changing the same conceptual customer, you are also changing the same instance. This comes very handy when debugging. You can actually inspect (watch) a customer object and see what happens over time and there is a lot less surprises, due to the way objects get externalized and then marshaled back. The lifespan of the object mirros
the lifespan of the conceptual object (yes, I do believe in the value of OOA as a way of thinking about the real world).

If however you are stuck with a relational database, OO is not your friend. At least not as in modelling your business object classes as classes in your programming language. The task of getting data from the database and then throwing it at a browser seems very functional. So you might be better of using functional programming techniques. Meta programming can also be applied successfully. To make it a complete heresy: Delphi is still unbeaten, when it comes to writing applications against relational databases quickly.