Seriously Contented -

As Your Lawyer...

Andreas — Sun, 28 Sep 2008 17:15:31 -0400

As your lawyer, I recommend the following new TV Series:

True Blood - Vampire show, but the twist is that it's public knowledge (plus it's set in the south, instead of LA or New York).

Sons of Anarchy - Described to me as 'It's a show with the gangs from The Wire, except they're bikers.' which is fair. It's not as good as The Wire, but it's got potential.

Generation Kill - This is already done, but I'm including it because it's that good. Done by some of the same people that did The Wire, it's about a company of soldiers in Iraq.

As your doctor, I recommend that you stay away from this new TV series:

Fringe - Promising to be the new Lost/X-Files, this new show is going to be full of almost reveals and pointless intrigue. Why can't more sci-fi shows be like BSG?

In defense of gearheads

Andreas — Wed, 13 Aug 2008 18:12:17 -0400

Apparently being a gearhead is bad. At least according to 37signals, house of featureless cool. Of course, they don't really mean it. At least I hope not. Sure, buying gear for its own sake is pointless, but if you want to build something, you need tools you care about. It's not about having the Eddie van Halen guitar/pedal/amp combination. It's about having a guitar, some pedals and an amp you know and love (or hate, any strong emotion will do).

Any activity that requires a tool requires knowing your tool. Whether it's a text editor, IDE, musical instrument, woodcutter or spatula, whatever you happen to be using is part of the process of creation. That means that at some level, they are affecting how you work. How you interact with and treat your tool, therefore influence how you create. That doesn't mean you should idealize certain tools just because they've been used in other processes, but it does mean that you should care. You should find out how your tool works and how to get the best out of it. Or if not the best, then what you want.

While the 37signals post doesn't disagree with (it's attacking people who "value gear over content"), it does not acknowledge that gear is a significant part of creating content.

Picking the right (power) tools

Andreas — Fri, 04 Jul 2008 16:35:08 -0400

I picked up a bunch of (Ikea) furniture (mostly bookcases) for my new apartment the other day and realized upon delivery that I was sadly lacking in the tools department. I didn't have a single screwdriver to my name. I'd gotten by for years on other people's tools, be they friends or family. I decided it was time to do something about this and headed off to the hardware store. I recently helped a friend put together some (Ikea) furniture and his cordless drill had made the whole experience much easier.

With this in mind I went into the hardware storing pondering whether to get a simple set of screwdrivers or to get a decently sized drill. Instead I found the right tool for the job. I realized upon seeing this little beaty that power tools don't need to be ridiculously big to get the job done. Especially not when the job is assembling Ikea furniture.

The joy of the Compact is that it's tiny, powerful and gets the job done. The only two disadvantages I've found are that it's a bit unwieldy for applying full pressure on it to solidly tighten difficult screws and that it won't fit down small cylinders for deeply mounted screws. The latter is a problem with most drills I guess and should be fixable with an extension. Pricewise it wasn't too bad either, substantially less than a full blown cordless drill, but more than a cheap set of screwdrivers.

I'm very happy with it, not just because it's been a song to assemble my coffee table and bookcases, but also because it's going to be easy to store away the Compact out of sight without it taking up a lot of storage space. Overall, choosing the the right tool for the job makes a big difference, not just for efficiency and quality, but also for the experience during the execution.

Generating gibberish

qebab — Thu, 03 Jul 2008 16:52:08 -0400

Generating gibberish is not really a very useful thing to do. But it's a funny thing to do, and it's not even hard. For the purposes of this blog post, generating gibberish is defined as:

Taking some amount of real, english text as input.
Being able to generate an (in principle) infinite amount of text that looks like real english as output.

We of course want the output to look silly enough for it to be funny. I think the method I have used here is what's generally seen as using markov chains to generate text. But I haven't read enough about that to be sure. In any case, it looks very much like the same thing, so I have used the term order to describe the constant that defines how the program learns, just as one would when describing markov chains. We need some datastructure to map 'beginnings' to 'possible finishes', in my case I have used a Python dictionary (Otherwise known as hashtable, hashmap, or just hash in other languages). The dictionary is the most essential part of the state of the program, but the vital part is that it is there, not that it is a dictionary. A binary-search tree would work, as would an ordered array. The following shows how the state of a program may look.

order = 3
word_based = False
state = {'the': [' ', 'y', 'm'], 'he ': ['i', 's' ...] ...}

Training

The correlation between the order, word_based and the state should be fairly obvious. The way we build up a state should also be easy to grasp. The following psedo-code illustrates this:

if word_based
    input = words(input)
for index = order; index < input.length; index = index + 1
    key = input.subsequence(index - order, index)
    if word_based
	key = unwords(key)
    if state.has(key)
        state[key].append(input[index])
    else
	state[key] = [input[index]]

Here, input.subsequence is assumed to be exclusive in the right endpoint. That is, "foobar".subsequence(0, 3) would return "foo", not "foob". Our algorithm so far, has two problems. First of all, we're adding duplicate values, that is every time 'the ' occurs in the text, we add another space to state['the']. This gives us weighting, but that might not be what we want. It is, however easy to check if there are duplicates before adding it. The second problem is that we do not keep track of 'starting points' for sentences. I shall explain why this is important in the next section.

Generating output

The generating output part is pretty easy. Make a result (It may be a string or a list, it does not really matter). Select a random key from our state, and concatenate it onto the result. Ideally, we want this random key to be one that is logical to start a sentence with so it may be a good idea to keep track of those aside from just our simple state dictionary. To see why, consider what happens if we accidentally selects 'end' as the random key. Next, it might find out that it can continue end by adding a dot ('.'), resulting in 'end.'. Then, it turns out that 'nd.' is not a key, and it can't make a longer output. This is of course an example, but it shows that there are reasons for keeping track of other things than just exactly what you need. Anyway, from state[key], select a random unit and concatenate it onto result.

Now we have result. The algorithm then, is as follows:

Select the last order units of your result, as your new key
Concatenate a random value from state[key] onto your result.
Not done yet? Go back to 1).

The banananana problem

One would think that the condition for stopping the generation of output would be to stop when your key is no longer in the state dictionary. This is obvious, we can't make more output when it happens. However, there is no guarantee that this will ever happen. Consider:

state = {'nan':['a'], 'ana': ['n']}

It should be obvious that this causes a loop. The easiest way to prevent this from putting your program into an infinite loop, is to limit the possible length of the output as well. This will occasionally give you output with repeats, but it won't freeze the program and eat all your memory.

Order

Varying order varies quality of the output (Assuming that the input is of a decent quality as well). When you use a very high order, like, say 6, you will get output that is very close to the input, often several sentences will be identical. Generally, using anything above 4 for character based training gives you output that is too similar to the input to be very fun. When order is in {3, 4} you usually get pretty funny output. When you use the word-based training, you should use {2, 3}.

Samples

I felt that I couldn't really close off this post without showing some samples, so here they are. The program that generated this output was fed this article (Except for the code), and used order-3 letter based training.

We next as for it hashtable, has two program, but it does no look silly seen there, not keeping a diction for stop when it happen. Concatenate it looks like real english is not reasons following duplicatenate it happen. Consider what's not there, input. Then, is ever, word_based a Python dictionary-search tree would return "foobar".subsequences.

Here, not keeping anything, but to map 'beginning you need some amount of 'starting, so I have useful think there adding somethink the starting, but that has two pretty easiest way to longer output.subsequences will occasionary. A binary is of reasons following some datastructure to 1)

It is, how the text, it is is what mighting, but is, "foo", not even hard. For things' to check if we want in my case I haven't make a starting, but it's a funny. I shall explain why, continue endpoint.

Conclusion

The point of this post was to be guide to my method of solving the problem of generating pseudo-english text. It is by no means the only, or best method, and it might have serious shortcomings. I don't think there are any errors in the pseudocode, but I can't guarantee that. Oh, and it is hilarious to train programs like these on Steve Yegge posts (Paul Graham is on a good second, for hilarity).

Programming languages!

qebab — Wed, 02 Jul 2008 19:20:30 -0400

This post is not very serious, and should not be taken as such. I'm jotting down notes about my 'relations' to several programming languages, and trying to give my impression of what sort of person they'd be. Whether I am talking about the standard, an interpreter or a compiler varies from point to point, but it should be pretty (un)clear from context. I also assign genders to them. This happens pretty randomly. Anyway, what I say about them is my opinion, and my opinion only. I am being pretty nonsensical, but I think, that for instance the paragraph about Haskell will convey that I am in awe of the language, and the paragraph about Java will convey a dislike. They are just general feelings and remarks, put forward in a strange way. I have written only about programming languages in which I have written a program that *does* something. Typically, I count programs that *does* something as small Unix shell tools, solutions to project euler problems > 30 and things like that. I have done many times that much in most of the languages mentioned here.

I'll start with C. C is a pretty good friend, we're on decent terms. It took me quite a while to get to that point though, C trusts me even when I screw up, so occasionally I hurt him and he hurts me back, even though neither of us meant it. He's generally a pretty careless guy, so I need to care all the more when I'm around him. C is a slim athlete, but it's still pretty much work to get him to do anything non-trivial. I still ask him for help when I need something done really quickly though.

C++ is Cs younger brother. He's pretty much insane. He is also anything but slim, but is still considered an athlete. Sometimes, when I say something wrong, C++ will yell gibberish for *ages*. It doesn't even have to be anything particularly wrong. I mean, I could say 1983 instead of 1982, and he'd yell for minutes about things that happened in 123981 BC. Then we both get confused. He also has all these weird things you have to remember when talking to him, like 'never say the word carrot on tuesdays and mondays, and always say car before boat' and things like that. It's pretty strange... I usually keep my distance to this guy.

I didn't talk much to D, but she seems like a much more sane sibling of C++. The little while I talked to her left me pretty impressed with her capabilities as an athlete, and as a friend.

Erlang is a pothead. Ever went to a party were they smoked pot, and there was this guy in a chair, having a gazillion thoughts in the air all at once? That's Erlang. I mean, I once told him to make dinner, breakfast, do my coursework and sing a song all at once. And he did! You know, I was pretty amazed... Erlang is nothing like the C family. He won't let you tell him two different things about the same thing twice, I mean if you say 'roses are red' he will scream at you if you later tell him that roses can in fact also be blue (Can roses be blue?). So you have to think about the things you tell him up front. If I ever want a bazillion things done at once, I know I should ask Erlang to do it.

Haskell is sort of a philosopher, but also sort of a pothead. But she's simultanously very pure. Did you ever see a person so pure, so good you thought they must be from some other planet or something? That's nothing. Haskell is so pure that pure sweetwater is to her like radiactive waste is to pure sweetwater. And that's saying something. She's still well able to do things out there in the real world though, she just has to wear a protective suit she calls 'Monad'. Which looks so unlikely that I didn't really believe it until I understood how *simple* it was. I mean, it was like... Yeah. I often can't really explain my dealings with Haskell. It's just too... Strange, and mindbending. "Do not try to bend the spoon." Right... Who wouldn't? Anyway, Haskell is sort of in the same family as Erlang, but then, sort of not. They're both pretty good at the doing-many-thing-at-once-and-it's-easy thing. And they both call you a liar when you tell them that you didn't really mean it when you said roses were red. And I wikied, and found out that yes, roses can be blue. So I guess that makes me a liar. Haskell is a very fast runner (But she's lazy, so sometimes she doesn't even bother starting), and what's more, the Monad suit thing gives her superpowers. And she has other superpowers too! You should definitely get to know her.

Java used to be my babysitter, but I got rid of her for a while now. I mean, our relationship is one of mutual distrust. It's like Java says "You're trying to do this because you hate me and it's illegal and I won't allow it hah!" and I think "You won't let me do this because you hate me so you made it illegal and oh god why don't I kill myself". We occasionally get along well, but that's easily forgotten during the times we didn't get along well and wrecked my entire house before my parents got home. Well, I did. Java didn't really go out of line, but I was like "GRAAAHHH DIE DIE DIE!" occasionally. She's a member of the C family, but she's a lot less flexible than her family members, because she insists on saying things like ThisIsAnObject.withAReallyLongName().reallyIAmNotJoki
ngDude() all the time. I mean, occasionally it feels better to yell "run(hills)!" than "moveToPointIn3dSpace(hills.getLocation())". Occasionally. We once played a mathematical game with big numbers, and she made me say things like one.add(three.multiply( AND I CAN'T TAKE THIS ANYMORE!!

Lisp is a very simple guy. He has a magnificent white beard, and he speaks in very simple terms, though often like Yoda speaks he does. He's very meta, I mean, he's the sort of guy who can reproduce without a female. He can literally change himself. He's like "Hrm, I sort of want another ear" and then he makes himself another ear and suddenly he can hear better. But he doesn't *look* any different (So where the fuck did that ear go, eh??). There used to be disturbances on the line when we spoke, but I can hear him pretty clearly these days. In the past, it was like there were some seagulls between us, constantly squawking around. Come to think of it, they are still there (He lives under a pier. Not the most popular of these guys) but I don't notice them anymore. Not unless I'm looking anyway. We don't talk together very often, but the talks we do have are usually very pleasant. Sort of like:
me: Give me a beer
Lisp: Tell me how to make a beer, then.
me: Wait, give me a brewery instead.
Lisp: Alright, this is an awesome-beer-brewery-for-you.
Very convenient in other words. But you can literally hear the hyphens when he speaks. Which bothers me sometimes.

Matlab is very single-minded. I heard tales that he can actually do many things, but after having spoken with him, I think that I have been lied to. He also looks weird. He's rich though, but he doesn't have a family. Well, actually I lie, Octave is sort of like a bastard son to him, but no one wants to admit that. Matlab thinks he coded the Matrix.

Ocaml is another crazy fellow. I don't know him well, I started talking with him, but I decided that I liked Haskell better. But Ocaml has a reputation for being very strange, and I am somehow attracted to strangeness so I may give it another go. He's also rumored to be a fast runner, and a powerful lifter.

Python is the person in this bunch whom I have known the longest time. We get along very well, we actually speak pretty much every day. The coolest thing about Python is that he pretty much understands everything I say (As long as I start some sentences with INDENT!! and some with DEDENT!!, that is). He's a pretty awesome fellow. Java would kill me for many of the things that Python says is entirely okay with him. I mean, I can even tell Python what + means when the left hand side of it is something that I made up. I mean, I could say 1 Hobbit + 13 Dwarves + 1 Dragon = 1 DeadDragon, and Python is fine with that. Python is also very clear when he has something to tell me. It's like:
Python: Dude, you ran over a squirrel.
Python: Turn to the left, about 20 meters back.
Python: In the middle of the road.
Python: It's dead.
I mean, don't shoot the messenger. That's a pretty good way to give out messages, even if they're as sad as this one (On the plus side, no squirrels were killed in the making of this blog post (We used a stunt squirrel (It knew how to get away in time))). Java would tell me something like:
OhMyGodYouRanOverASquirrelException("Long text"):
InTheLeftForkOfTheRoad20MetersBack("Even longer text!"):
And then never really getting to the point (The fact that the squirrel is actually almost certainly dead!)

Ruby isn't quite the brother of Python. They're more like cousins. Ruby mother is probably Smalltalk, but I don't know for sure. Ruby is awesome in the sense that I can say anything to him and he will do it, unfortunately sometimes I say things I don't really mean. Python will often protest and say like "Wait, you wanted to *kill* the dead squirrel?" and Ruby will say "Alright" if he can make sense of it somehow. Ruby is very understanding, so most of the time, he can. Ruby is also in close family with Perl, but fortunately he made off with the weird looks. I mean, who wants a green beard and tattoos all over anyway. Some say that Ruby is the best of the young programming languages. I say he's one of the most easygoing ones, and a good alternative for the younger programmers to get to know. I think me and Ruby will become good friends in time.

Smalltalk is sort of like Lisp. She's very meta. I mean, most languages have been told my some outsider that 'True and False are different things and this is what they mean' and so on, but Smalltalk figured that out by herself. She's so clever that she bred a dog that breeds squirrels that she can ask about things like ifTrue and such. Smalltalk is also like lisp in the sense that she can add bodyparts and still not look different! Very remarkable indeed. She doesn't live under a pier like lisp, she's even less known. So no one really knows where she lives. She hasn't decided if she likes it this way yet. We don't know each other very well at all.

Right, I don't *think* I have forgotten any languages that I have actually used. Scheme falls under the Lisp umbrella.

What programming languages do you have good relations with?

What is whitespace sensitivity?

Andreas — Wed, 02 Jul 2008 03:10:59 -0400

This post raises the question of what is whitespace sensitivity. This is a somewhat contentious issue, as it's generally used as a reason to not use Python (presumably by people who have written Makefiles), so pythonistas are usually a bit defensive about it. I've always found it a bit unfair on Python as a language and found the Ruby example amusing.

The average student does not exist.

qebab — Sat, 07 Jun 2008 10:03:10 -0400

I'm 21 years old now, and I have spent two thirds of my life being educated. My brother is 17, he's spent 11 years being educated. There was a reform between us, so he's been in public school one more year than I had when I was his age.

We have wildly different interests, but we still agree that most of what we did in school was a waste. The same goes for practically everyone I know. I think that for every Norwegian youngster you find that liked school (The educational part of it), you will find five, or ten who didn't.

I can see a lot of problems, but hardly any solutions. I wish to make it clear that I don't have any, and that I don't think there are any good ones, but I still think it's my right to point out problems, and hope that someone smarter than me can see solutions. Or alternatively, convince me that there are no problems; It doesn't seem very likely, but never let it be said that I was not open for it.

The main problem with Norwegian education the way I see it, is that it is tailored for the average student. This student does not exist. In nearly every class, you will find both bright and slow students, you will find people who average in one subject and aces another, you will find people who average in one, and dumps in another. I don't think you'll find anyone who averages every subject, and even if you did, that student would probably not be happy about his or her results, or the curriculum in every subject.

The bright students don't get the challenges they need to enjoy school, and the slow students don't get the help they need to enjoy school. This is a consequence of the fact that school is perfectly tailored to suit the person you get when you add these people together, and divide them by the number you have of them.

I think it is very important that we give everyone the chance to enjoy education, we give them the satisfaction of being good enough, but only after having worked for it. My little sister is 11, once in a while I give her a puzzle I consider to be hard for her, but not so hard that she gives up. She will work with it, and the joy she gets from solving it is more reward than I need to have for thinking it up. The amount of problems like that I encountered in school was pretty minimal.

When I started at university, I was pretty shocked to find out that basically everything I had learned up until that point was just disregarded. They started with things I had learned three years prior, and taught it all over again (In six months). I learned it a lot better the second time around, not only because I had done most of it before, but mostly because I was treated as a consenting adult rather than a child to be spoon fed. I was allowed to prioritize the things I found interesting, and that alone goes a long way towards making university cooler. Of course, there are still mandatory subjects if I want a degree, and not all of them are interesting at all.

My theory is that if basically all knowledge I had from earlier education (Except really basic things like spelling and fundamental arithmetic and so on) will be disregarded at university anyway, then the time in earlier education could be spent more wisely. A fine idea would be to help students discover what they are good at, and what they are interested in (These two things often overlap, as far as I can tell). I would have loved a school where I had the option to choose most of the time. I would certainly not have had to spend a year in university taking a degree that was not the right one for me, and then switch, like I did.

The fundamental curriculum in school is just outright boring and uninteresting when seen as a whole. Parts of it are cool, though. Is this the best we can do? Tailoring it into fitting a student that does not exist?

Moving to git-svn

Andreas — Sun, 01 Jun 2008 15:38:08 -0400

I've decided to start using Git for source control at work. Unfortunately, the project I'm on is stuck on Subversion. I've raised the idea of putting up the necessary infrastructure for optionally using Git for source control, but so far it's still in the idea phase. I'm probably going to have to come up with that infrastructure on my own to get it done. I have a couple different reasons to prefer Git to Subversion:

My current project at work occasionally requires me to commute to our customer's offices. This means about an hour and a half spent on a train each way without proper internet access. Since our project uses Subversion for source control, this means coping without any source history.
Just this Friday I had to spend 45 minutes cherry picking revisions to revert after a failed experiment on trunk. We created a branch immediately after, but this was hindsight and we'd already figured out how to do our experiment right by then. Two commits later we merged back into trunk. I regularly find myself wanting to branch for entirely trivial reasons and occasionally not so trivial reasons, but as long as the bar is raised by having to muck with everyone's repository whenever I want to do that, I won't.
I want to be able to share my branches without committing to the project repository. This will only work if I get other people to use Git, but given the utility I think I should manage to get a couple people to, and that'll be enough for me.

Unfortunately, git-svn is a bit confusing to use, since it requires understanding enough of both Git and Subversion to see how the translations are going to happen. I'm still not into the comfort zone on this, but this simple guide on using git-svn with Google Code is very good. It covers the most used commands without going into enough detail to be confusing.

Missing the point on buying UGC companies

Andreas — Sun, 01 Jun 2008 15:12:58 -0400

Google App Engine is pretty neat, but it's not an argument against Google buying Twitter. The fact that Google can build a multicast messaging system in a few days is orthogonal to why they'd buy Twitter. The same way having Google Video didn't preclude buying Youtube. Buying Twitter would be to gain the userbase, not to gain the source code.

This is why I don't think Twitter is going to be bought any time soon. Sure, it has a lot of users, but Twitter is basically like a modern version of IRC. Instead of requiring people to congregate in non-scaling (no pun intended) channels, they describe the exact communication ties they wish to follow. The people who are ubiquitously on Twitter definitely seem to think they're in some sort of importosphere, but the same went for IRC and Usenet, neither of which ever made anyone (bar a few hosting services) any money or broke into anything remotely nearing the mainstream.

So I guess they're not all insane

Andreas — Thu, 08 May 2008 04:51:03 -0400

It's good to see that not all Java developers are complete raving lunatics when it comes to static typing. I can't believe anyone would seriously believe that this is good idea:

Map<@NonNull String, @NonEmpty List<@Readonly Document>> files;

The mind boggles . . .

If GWT doesn't scale, what does?

Andreas — Mon, 28 Apr 2008 13:37:54 -0400

At work I'm currently building a pretty big webapp using Google Web Toolkit. Often when I tell people this they respond by asking if it scales. Usually they will back up this question by qouting some company that did an analysis and concluded that it didn't. Some people on my project could very well end being quoted like that.

I wonder what these people consider the alternative to be when building rich internet applications. Realistically it can only be Javascript or Flash and its embedded brethren. Setting aside the embedded alternatives for a moment, Javascript has been used quite successfully to build some quite large and successful webapps. The various Google apps come to mind.

But if this is the solution, doesn't it imply that Java does not scale? Some of the problems with GWT are language independent (compile-time), but that doesn't seem enough on its own to write off GWT.

The idea that Javascript scales better than Java must seem heretical to Java purists, who firmly believe that strong, static typing is the only way a project can scale beyond a tiny handful of developers.

I wonder what the serverside implications of this are. Perhaps that Java should be reevaluated in favor of other languages, like Python or Ruby.

Heroes Happen Here - Review

Andreas — Fri, 25 Apr 2008 17:20:45 -0400

When I first read about Heroes Happen {Here}, I thought it sounded like a bad idea gone wrong. I still have no idea what the actual Heroes Happen {Here} thing is, but it has something to do with celebrating IT professionals. I think the curly braces in {Here} are supposed to denote a hug, apparently because IT professionals work somewhere lonely.

I was further discouraged when I read about the accompanying comic. Funny sponsored by Microsoft and Seagate. To top that up, it's written to take ideas from IT professionals's experiences. Of course, I work with funny IT professionals and use Microsoft and Seagate products every day, so I figured I'd give it a shot. I subscribed to the RSS feed and read the funny about Heroes Happening... {Here}.

It seems to me like a bad try at some sort of Web 2.0 comic written by the target audience, but it's drawn and written by pros. The story so far has followed an anti-virus company hot on the trail of a rootkit. And that's it. Nothing amusing has happened, nothing worth writing home about. It's some people doing their job and making bad puns.

Of course, I'm giving them free advertisement. Make up your own mind. Perhaps this this gem will convince you. Or perhaps you wish to read about audacious employees demanding wedding cake donuts. You read it, I can't anymore.

Dvorak, part II

qebab — Sun, 13 Apr 2008 10:23:45 -0400

So, I've been using dvorak for about a week now, and I'm finally starting to get some semblance of control. I can tell for sure though, that this is the last time I will ever change my keyboard layout significantly. I am currently typing about 30 words a minute, which is a good deal more than I was typing a few days ago, but also less than I was typing before I switched, and by quite a margin. I am unsure if I will regain the speed I had, but it's not that important. I suspect that I will not get entirely back up, but I type fast enough, and it feels more comfortable than before, which was indeed the goal from the start.

The one thing that bothers me is that my login screen reverts back to qwerty on reboots. I *think* I fixed it by editing /etc/X11/xorg.conf, but xorg.conf has blown up in my face before and I have not yet done a reboot, so it's quite possible that it is just a ticking bomb at the moment. Time will tell.

I still find it odd to code. I can write english fairly fast again, I am also comfortable with norwegian, but coding is still hard. I guess it will work itself out in time though. The longest program I wrote after the switch was a meagre 100 lines, so I certainly hope so. At least I can chat again. And blog.

In other news, I managed to do an rm -rf on the wrong folder the other day, and am still trying to recover what I lost. The lesson to be learnt from this is that tab-completion can not read minds.

And that's it for now.

Dvorak, part I.

qebab — Wed, 09 Apr 2008 16:00:36 -0400

Changing your keyboard layout turns out (surprisingly enough!) to be a hard thing to do.

In keeping with the spirit that hard things are rewarding in the end, I decided that I should blog about it. As a bonus to the people who bother reading it, I will note down the time occasionally, starting now. It's now 21:15 CEST.

I have been using the dvorak simplified keyboard layout for the last 26ish hours, of which I have spent about 8 in front of my laptop (Which really is quite scary). I am not fighting my hands nearly as much as yesterday, and it's getting somewhat easier to type by now. I am amazed at how much less I actually move my hands, and also at how crippling it feels to be a novice typist. I think that I'm typing at somewhere around 16 words a minute, which is barely anything at all to me, though I make surprisingly few typoes at this point. This was cited as one of the strengths of dvorak, and at the minute it does look promising.

This is going to be a funnily small text, the time is now 21:30, and it feels like I have been writing this line since forever and a day already. I just erased a line, and I nearly killed myself over it. Anyway, I am feeling less like I'm made out of wood than I usually do at this time of the day, whether this is the decreased speed at which I type or dvorak is hard to tell at the moment, but I am fervently hoping for the latter. The supposed effect against RSI was the main reason that made me decide to give this crazy idea a shot. I imagine that I will be working with computers for the better part of the rest of my life, so I figured it was worth a shot.

I have no idea what the best approach to something like this is (It's now 21:46, by the way), but I decided that since I did not want to relearn touch I should not remap my keyboard. Thanks for the advice, Andreas and #dvorak. I did find something that helped me learn the keys relatively fast though, a program named dvorak7min. I do wonder what the 7min part is all about, it sure as hell can not (Is not even allowed) to mean 7 minutes. Seriously.

I'm going to call it a day now, and try writing another post in a few days to check if things have improved. So, for the record, the time is 21:56.

Disabling the Dashboard in Leopard

Andreas — Tue, 08 Apr 2008 15:59:47 -0400

I recently picked up a new hard drive and a copy of Mac OS X 10.5 (Leopard). I've been reconfiguring it to match my old Tiger setup, and improve where I can. One of my pet peeves with Mac OS X is the Dashboard. Unless you have a lot of widgets you use actively, there's no real point in having it hang around eating up memory. Especially not when some of the common widgets have memory leaks. The peeve, however, is that there's no GUI configuration to disable the Dashboard.

Fortunately, there's here still works in Leopard. So to disable the Dashboard

defaults write com.apple.dashboard mcx-disabled -boolean YES

To enable it again, simply undisable it.

defaults write com.apple.dashboard mcx-disabled -boolean NO

Finally, to make the changes take effect, restart the Dock. In a terminal, type:

killall Dock

And now the Dashboard is no longer leeching precious memory from my scrawny macbook.