ivy’s page

Over 10 years ago, Dr. Dobbs Journal was my favorite computer-related magazine. The copies would cost at least twice as much as other magazines in the store because there was only small demand, they were flown into our country and were paid in (back then) expensive US dollars. I considered a copy worth its money however, because it contained well-written, in-depth articles of a broad range of issues of the software engineering field. I could spend hours reading it. If I didn’t buy them myself, I always had a nice gift idea at hand…

For at least five years now, I have seen it’s quality declining. The content just didn’t seem so appealing anymore and the magazine became also physically thinner. I still had a subscription; partly because of old-times-sake, and partly because as a working guy and the current €/$ rate it felt well-affordable.

In the beginning of this year I received a notice of a change; Dr. Dobbs Journal would merge with Information week, and in order to keep receiving issues under my subscription, I had to fill out some web page. At first I wasn’t aware of what was happening, but having received two issues of Information Week with a small part reserved for Dr. Dobbs, I’m completely convinced I will not continue my subscription. It’s sad, but I don’t feel part of the target audience anymore

Google told me I’m not the only one:

• Farewell to Dr. Dobbs…
• The demise of Dr Dobbs Journal – hopefully exaggerated
• R.I.P. Dr. Dobbs
• The 2008 Media Inflection: Meet Dr. Web, the New Gorilla
• Dr. Dobbs Journal Goes Away

In the beginning of the internet (now I sound like an old guy), its users behaved in ways prescribed by netiquette; a term that Wikipedia defines as a set of social conventions that facilitate interaction over networks. You would for instance be flamed for sending unsolicited mail to a bunch people, and if you did it again, your sysadmin would hear about it and block your account for a while…

Since internet became common and commercial, netiquette faded into oblivion and gave way to the dark side . Ah well, common sense get’s you a long way and the internet became a lot more fun too, so you don’t hear me complaining (or at least not that much though).

One thing that annoys me though is the continuing spam in my mailbox; an unthinkable thing if the guys sending this stuff had ever glanced over the netiquette guidelines. Today we have to block and filter instead of relying on the naive behavioral constraints posed by netiquette. It’s a rat race where spammers come up with new ways of slipping through spam filters every day. Spam filters ‘read’ the text? The spammers put theirs ads in pdf attachments! These mails get blocked too? Let’s hide our message in images… And so the story continues.

One method I hadn’t seen before is HTML table-based characters, where a table cell represents a pixel in the message. I have to give them credit; it’s an original avoidance strategy. The spamfilter just sees HTML codes like:

<table border="1" cellpadding="2" cellspacing="2" width="100%">
  <tbody>
    <tr>
      <td bgcolor="navy" valign="top"><br>
      </td>
      <td valign="top"><br>
      </td>
      <td valign="top"><br>
      </td>

But the browser renders:

Anyway, if you ever feel like finding out how ‘the elders of the internet’ intended us all to behave, read the officially defined Netiquette Guidelines. Be warned though; the latest update was in 1995, which is light years in internet time

I once heard someone claim that XML has the efficiency of a plain text file combined with the readability of a binary. Every time I have to type XML by hand, I tend to agree! XSLT for instance seems nice… until you actually have to program and maintain logic/functionality encoded in it for a while: it is verbose, it forces you to express constructs in hierarchical XML structures and it becomes a bit frustrating….

XML is great for tools and communication between programs, but it’s verboseness and forced hierarchical structure makes most XML documents not as readable as promised. The problem I observe is caused by the accidental complexity (the difference between intention and result) inherent to the language. You have to close each tag (type it twice), parameters are passed by attributes or subnodes (a lot of text) and there’s only hierarchical structures instead of more shortly expressed ways. There’s a lot ‘ceremony’ to express simple intent.

Therefore, I’d advice people to use XML for it’s strong points, but not overestimate it’s power for purposes it’s just not intended for. For many purposes I’d rather create an (e.g. oaw-based) Domain Specific language that let’s me express more things more easily and gives me context support in my favorite editor while doing so.

With some music

A load-balanced web application I’ve been working on suffered a -quite annoying- problem of people being logged out within only minutes of inactivity. It was related to lost sessions. After some tests for common session time-out problems I discovered quite an interesting feature of Internet Explorer 6 (and higher) worth to be aware of.

We tested the application ourselves with the mentioned browsers and never had problems, but as various clients offered access to our application through a frame on their site we were confronted with an untested and problematic set-up in which IE6 (and higher) refuses cookies.

To sketch the situation; http://clientx.com/ contains a frame linking to http://oursite.com/app/clientx . Because the host name in the frame (oursite.com) is different from the toplevel host (clientx.com), Internet Explor the frame as 3rd party content. From IE6 and upwards, the default privacy settings lets the browser silently ignore cookies from 3rd party content. This is done by intent, see the Microsoft KB issue.

A simple solution is to add a P3P-response header defining a Compact Policy (which you may or may not respect), for example:

P3P: CP="CAO PSA OUR"

As a developer you’re not likely to run into the problem during development because you:

1. use firefox
2. have changed the default IE privacy settings to something more workable
3. access the site directly instead of through a frame

java.lang.System.getProperty() is a Java method to access a system property. The properties you can lookup are partly predefined, and others are set by tools or trough command-line parameters (-Dmyproperty=true). Because System is a class which deals with lots of system related things, Object-Orientation-wise it makes sense to put this responsibility (method) in this class.

But who was sleeping at the Java design team when someone sneaked in Integer.getInteger(String), Boolean.getBoolean(String) and alike? Completely counter-intuitive these method return a System property in Integer/Boolean form. I would have felt comfortable with something like System.getProperties().getBoolean(String), or System.getBooleanProperty(String), and I even could have lived with the legacy of a System.getBoolean(String) method, but not this. I’ve tripped in this myself (again) a short while ago, thinking that it was a method which would parse the String for me. I’ve seen other people make the same mistakes. From experience Java programmers know to use Integer.valueOf(String), but not because the method name makes so much more sense…

Isn’t this a great deprecation candidate for Java 5, 6 7?