Wille Faler’s Buzzword Bingo

I was just made aware of a rather serious bug that may kill the hard drive of any laptop using Linux (at least recent versions of Ubuntu). Here’s a rather dramatic account of the effects.
The bug especially affects laptops when they are used from battery. Luckily, I have not used my laptop on battery too much, and having checked the number of load cycles on my hard drive as per instructions in the bug report, I’m still in the clear, the bug hasn’t affected me, but nonetheless, it means that my trust in Linux on a laptop is all but gone (I had about 30 000 load cycles after 6 months of using my laptop, which puts the lifespan of the disk at about 6 years given the current rate of cycles and use, as most drives are made to last at least 300 000 load cycles).

It’s a shame, I’ve enjoyed using Ubuntu so far, but having the default settings potentially kill the lifespan of laptops is unacceptable. Yes, sure, there may be workarounds, but it doesn’t change the fact that the default settings may actually kill your hard drive.

If you are required to hack your default settings in order not to potentially destroy your hardware, Linux still has a lot of growing up to do for mobile users..

(thanks Matt for the tip)

The Daily Telegraph reports that after a cool summer, a ski resort in Austria that was previously said to be “doomed” by global warming is now able to open up six weeks early because of ample snow.

As warm, cold, calm, stormy and every other type of weather imaginable can be blamed on global warming, the cold weather and snow is probably down to global warming..

(Now back to creating a mass hysteria by means of scare-mongering around the “second decade of the millennium bug”, which I hope will fund what I hope to be my very early retirement..) 

First off, I’m not a Mac user: I use Ubuntu Linux as my primary platform, and have an older Windows XP laptop for media use (read: iTunes).

But my experience with Windows Vista (which instigated my switch to Ubuntu) has really made me consider Mac. Ubuntu is fine, but after watching the guided tour of Mac OS X Leopard, I’m sold, why would anyone want to use anything else as their platform?

Well, I’ll tell you why: I’m a Java developer/architect. My Java knowledge is my breadwinner, and incidentally, Apple’s support and future plans for Java on Mac is somewhat obscured by a lack of tangible official news.
Mac OS X supports Java 5, but Apple has previously stated that they will no longer provide Cocoa bindings for future Java releases. Also, a Java 6 alpha on Apples Developer web site was pulled a few months ago, with no official notes as to why. All of that makes for an ominous feeling for anyone wanting to do Java development on a Mac in the future.
What makes it even more ironic is that a lot of Java developers have adopted Macs enthusiastically, and if Apple won’t support Java, they are left to hang.

Even though Java developers are a rather small community, they tend to be early adopters and enthusiastic evangelizers, leaving them to hang would seem stupid, supporting a small, but vocal community has a lot of knock on effects: they persuade others to try a platform, they create a lot of goodwill (or badwill if you turn on them). It would seem reasonable for Apple to continue their support of such a community.

So please Apple, disclose to the public what your plans for Java are, both short term and long term. If you do, and it’s positive, you have my promise that my next computer will be a Mac!

One thing about Ubuntu Linux that takes a bit of getting used to is the lack of direct root access: Instead, Ubuntu uses a program called “sudo” for anything that requires root privileges. When you invoke sudo before a command, you are simply prompted for your password (if the user has sudo privileges).

I’m definitely no expert at *nix security, but it got me thinking: isn’t this actually circumventing regular Linux security? Instead of having a separate user with a separate password, regular users with sudo access can do root-level stuff. In practice this means that for a malicious person, it is sufficient to find the password for a user with sudo access, and it is quite likely that this user will use his/hers password a whole lot more than if there was a separate root user dedicated to root tasks and only root tasks.
To me, this kind of looks like a security risk: it is not very far off having a user actually running everything as root, which is the very thing we are trying to avoid on a Linux system.

Obviously, the workaround for this is to create a further user that is dedicated to sudo access, and do all things root through him, but only that. But firstly, isn’t that just having a root-user with another name? And secondly, who will actually bother doing so?

Other peoples thoughts would be greatly appreciated..

I received my brand spanking new iPod Touch today. There are a few things that really blow me away about the device:

• The touch screen interface is so much more than any previous touch screen I’ve come across: use two fingers to zoom in and zoom out. It is probably a taste of what we’ll see in future screens everywhere in ten years time.
• The interface is actually so usable, that it is almost simpler to use the Touch interface, rather than the regular iTunes on a desktop machine.
• I thought my browser on my Blackberry was great, but the Safari browser just blows it out of the water: The ability to zoom in and out on the screen is just brilliant for viewing regular web sites easily.

The interface and ease of use actually makes me see how future iPod Touch devices could be standalone music devices: buying and downloading is so simple, you barely need desktop iTunes for anything but backups. With a slightly larger screen (but not by much) and more storage space (I have 16GB which is borderline too small for a standalone device), I can envision myself just having a dedicated iPod Touch docked in an iPod stereo/speaker system permanently and use it as a stereo.

I also viewed a few videos on my touch, and the screen is great: I’ll definitely use the touch and iTunes to watch TV shows on my travels to pass time.

All in all, the touch is a great little device, which no doubt holds a lot of potential for future development by Apple. My only little complaint is that the interface for text input is inferior to that of my Blackberry: it is much better than the input for most touch screen devices, but no match for a proper keyboard. In practice this means that while I will use my iPod Touch, I probably won’t get an iPhone unless they radically improve the input and/or create an iPhone with a dedicated keyboard.

So, I’ve been rejected for credit for the second time this year, this time for an American Express Business gold card, that I thought would be useful for separating company expenses from personal expenses.

I’m beginning to suspect that either someone has stolen my identity, or the credit raters actually punish people for being affluent and debt free. It is funny how I was approved for a £400K mortgage earlier this year (which I eventually never took out), but I can’t even get the simplest of credit/charge cards.
Reason seems to have nothing to do with it: I make a comfortable upper middle class income, I have a semi-considerable amount of assets in investments, and I have absolutely no debt, zero, nada, zip. The only thing that could be perceived as “debt” is possibly my monthly credit card bill, which runs at around £1000 a month, but which I pay in full each month on time.
Furthermore I have never-ever paid a single bill late since I moved to the UK three years ago, in fact, I am kind of anal about bills and probably act a bit stupid by always paying bills the second I receive them.
From a creditors point of view, I should be a dream: I always pay, and I always pay on time with time to spare.

I can only conclude that I am either subject to identity theft, or I am being punished by not having any debt. Being punished for being debt-free is actually a possibility: I have heard that a lot of creditors and credit agencies feel more comfortable lending to people who are already in debt.
The herd logic of lending tells them “if others where comfortable enough to lend someone money, it must be a vote of confidence in them”. In other words: it makes more sense to mindless pencil-pushers to lend money to someone who is already halfway to bankrupcy, than someone who has an abundance of assets and cash but no debt. Brain dead..

A comment on my previous post: Embedded JBoss doesn’t seem too stable or dependable yet. My problem was that that EJB’s would not seem to be bound when running the embedded server from a maven script, thus throwing a NamingException.

Problem “solved”: using Bootstrap.scanClasspath() to deploy ejb’s seems convenient. But it doesn’t always work: works like a charm in Eclipse, fails every time in Maven. The solution? Use the virtual file system and assemble “virtual jars”. More work and tedium, but at least it works under both circumstances.

There are few things I hate more than half arsed build scripts, especially if they are provided by a third party provider.

I’m not mentioning any names (JBoss), but certain companies are notorious for only doing half-hearted, semi useless build scripts. Today I’ve struggled with getting Embedded JBoss working nicely with Maven 2: It works like a charm in my IDE, but when I run it in the Maven-build, there is always some test failing, because the Embedded JBoss behaves slightly differently from a build script compared to in the IDE (EJB’s that are bound in the IDE, are all of a sudden “not bound” according to the failing test).

Before this, I had to chase around for dependencies manually, because since one of the big selling points of Maven 2 in particular is its dependency management, JBoss decided to not create proper dependency trees that retrieve all the dependencies. Not a first for JBoss by the way: I had the same problem, but even worse when it came to Hibernate and standalone JPA.

Now, if you are intent on not doing your build scripts properly, or annoying the hell out of your user base, why does someone bother with build scripts to begin with?
If there where no public maven repositories, I’d have no expectation of them working. But now that they have provided the maven repositories, I have been naive enough to think they would actually work, and it annoys me to no end that they don’t.

Apparently this blog post about setting up Ubuntu got a mention at The Register (one of the UK’s most popular IT web sites) with regards to common network problems users of Ubuntu 7.10 have experienced. In practice it means that I’ve almost had as many visitors today, as I usually get in a good month.

..best let my easily inflatable ego enjoy the shine of my 15 minutes of pseudo-fame among fellow nerds..

EJB3 is pretty much what Enterprise Java Beans should have been all along: Just annotating a class with @Session and having a related interface is a lot simpler than having two interfaces, one implementation class, one or more xml configurations and in most servers having to run the EJB through a pre-processing application.

I just have a feeling that EJB3 is too little too late when it comes to J2EE/JEE in general, and EJB’s in particular..
Firstly, the complexity and overhead of EJB 1.x and 2.x has burnt so many developers and organizations that they may not give EJB’s a second (or probably fourth) chance anymore, they will just tune out at any suggestion of EJB’s.

But my second thought is probably more relevant for practical reasons: the part of EJB3 that most people will want is probably having the container inject JPA EntityManager’s and manage transactions transparently for stateless session beans, it’s what 95% of applications need and use (even “enterprisy” ones).
Only caveat for JEE and JEE vendors is, that this is dead simple to implement without the need for a JEE container: all it really consists of is managed service objects with annotations that are invoked via dynamic proxies and looked up via some sort of factory (that probably caches annotation and class metadata). Any reasonably talented developer could develop and test such a solution within a day or two (and quite a few probably have).

Why would you want to bring on the full complexity (and risk of bugs and problems) that a 100mb (JBoss 5beta2 is 93mb) behemoth of an app server potentially brings, when you can achieve the same result with a 20kb jar-file deployed in Tomcat or Jetty (or even embedded Jetty)?
I just don’t see a rationale for it.

In my opinion, EJB3 contains a bunch of great ideas and patterns, but they are actually so brilliantly simple that you no longer need a full JEE app server to use them, in fact, doing so even seems a little counterintuitive. It is ironic that the simplicity of EJB3 probably means that J2EE/JEE as we know it is and should be dead.
The big and monolithic application server should be sent to where it belongs: the scrapheap of history.
Bring on the tiny, modular “pick your own components”-app servers!