planbnet.org - Felix Rotthowe

Static HTML rocks

2012-02-10T00:00:00+01:00

I’m not a blogger. I’m not a good writer and I don’t have any interesting stories or insights to share. But I always wanted to have a blog. Maybe just because everyone has one. So, even though I didn’t write much, I have tried numerous blogging engines over the years. I even wrote my own (which was used by me and my friends partly like those kids today are using facebook: to post funny pictures and party invitations). I also tried wordpress and several hosted solutions such as posterous and tumblr. But it never felt right.

This time it feels better: I’m using Jekyll to build static HTML pages out of HTML templates and markdown formatted text. I can use vim to edit posts and quickly update the whole page. This fits my simple needs perfectly, unlike those huge piles of php scripts. And I have full control over the HTML code that gets generated, so if you look into the source of this page, you’ll see nicely indented code (except for whatever the markdown parser generates).

This whole system feels so efficient and lean that I started thinking about how to further improve load times of the page. After all, there’s nothing to do on the server side but accepting requests and streaming the contents of static files to the clients. The Apache server running on my virtual server does this blazingly fast, especially (which I find a bit confusing) when turning Keepalive off in the config, but there is still something that can be improved: The page won’t be displayed by the browser before all css and javascript files have been fetched. This means a few extra requests and kb before the user sees anything on the screen. To prevent this, I’m using the modernizr load plugin, which loads the scripts asynchronously in the background. Of course, you can not use deferred loading for scripts that must run before the page has been fully loaded, but in my case it doesn’t matter if jQuery, the fancybox plugin and the google analytics script are executed a few ms after the user sees the page.

This makes the page load really fast:

Additionally, I have enabled sending cache expiry headers in .htaccess, so that each subsequent request will only need to fetch the pure HTML (which will be sent compressed for clients that support it by the way). This means that all further links the user visits will be rendered within under 100ms - given a fast enough client connection of course.

I hope you noticed this when visiting this article…

Zenburn everywhere

2011-12-06T00:00:00+01:00

I’m a big fan of the Zenburn color scheme. It’s easy on the eyes and very readable despite not providing too much contrast. Besided that, I just like the colors. That’s why I’m using it everywhere I can. To keep a reference to all the different config files and snippets, I’m creating this list:

Gnome Terminal

Based on the install script for gnome-terminal-colors-solarized I have created a simple script that modifies an existing terminal profile and sets up the zenburn color scheme. You can download it here: gnome-terminal-zenburn.sh

Mac OS (Lion) Terminal

I couldn’t find a configuration for Terminal.app on Mac OS X that replaces the default ansi colors, so I created one myself. You can find it here: Zenburn.terminal

Eclipse

I use the excellent Eclipse Color Schemes plugin which includes Zenburn. In fact that’s how I became aware of it.

Vim

Zenburn is included in the vim-colorschemes package that I installed using vundle.

My .vimrc

2011-12-02T00:00:00+01:00

Over the last weeks, I have heavily customized (and commented) my .vimrc and thought it might be useful for someone else if I publish it here. I’m not putting it under version contol with git like many others seem to do, because I would always forget to commit and push it (and I don’t need a versioned history), but I keep it synced using Dropbox. If you want to download the version posted here, you can do so here.

Maybe I will add some comments on some keymappings or why I’m using vundle for package management later, but for now I’m just dropping it here:

Pinging the Apple Store using node.js

2011-11-30T00:00:00+01:00

Every once in a while, Apple decides to take down their online store for a few hours to release some new shiny stuff. I’m not sure if they have to close the whole website because of the archaic WebObjects framework they’re still using or just to generate hype, but the yellow post-it note is well known across the Internet.

So, when there’s a new iPhone or iPad ready for sale, people keep hitting the refresh buttons of their browsers to be the first to place an order. I have to admin that I have also participated in this procedure once. But because I’m lazy (and I was at work at the time the sale started) I created a litte script that pings the Apple Store every few seconds and sends me a push notification using the great Boxcar service.

While this would have probably worked just as well in any other language, I chose node.js to implement this little script. Maybe someone else wants to use it, too, so I’m publishing it here. Please note that you may have to adjust some URLs to it works for your country (it’s currently set up for the German Apple Store)

var sys = require('util'),
fs = require('fs'),
http = require('http'),
https = require('https');

var BOXCAR_USER = 'BOXCAR_USERNAME';
var BOXCAR_PASSWORD = 'BOXCAR_PASSWORD';
var BOXCAR_AUTH = 'Basic ' + new Buffer(BOXCAR_USER + ':' + BOXCAR_PASSWORD).toString('base64');

function notify(app, title) {
  var url = "http://store.apple.com/de";
  var icon = "http://images.apple.com/euro/iphone/built-in-apps/images/nav_icon_app_store.png";
  var uniqueid = "notify"+(new Date().getTime());
  var params = {
      "notification[message]": title,
      "notification[from_screen_name]": app,
      "notification[from_remote_service_id]": uniqueid,
      "notification[source_url]": url,
      "notification[icon_url]": icon  
  };

  var dataArr = [];
  for (obj in params){
    if (params[obj]){
      dataArr.push(obj + '=' + params[obj]);  
    }
  }
  var postdata = dataArr.join('&');

  var headers = {'Host': 'boxcar.io',
    'Authorization': BOXCAR_AUTH,
      'Content-Length': postdata.length};

  var options = {
    host: 'boxcar.io',
    port: 80,
    path: '/notifications',
    method: 'POST',
    headers: headers
  };
  
  console.log("Sending request to boxcar: " + title);
  var req = http.request(options, function(res) {
      console.log("Sent notification for " + title);
  });
  req.write(postdata);
  req.end();
}

function refreshPage() {

  var headers = {
    'Accept':'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8',
    'Accept-Charset':'ISO-8859-1,utf-8;q=0.7,*;q=0.3',
    'Accept-Language':'de-DE,de;q=0.8,en-US;q=0.6,en;q=0.4',
    'Cache-Control':'max-age=0',
    'Connection':'close',
    'Host':'store.apple.com',
    'User-Agent':'Mozilla/5.0 (X11; Linux i686) AppleWebKit/535.1 (KHTML, like Gecko) Chrome/14.0.835.186 Safari/535.1'
  };

  console.log("Refresh " + new Date());
  var client = http.createClient(80, 'store.apple.com'); 
  var request = client.request('GET', '/de/', headers);
  request.end();
  request.on('response', function (response) {

    if (response.statusCode == 200) {
      var pagedata = "";
      response.setEncoding('utf8');

      response.on('data', function (chunk) {
        pagedata += chunk;
      });

      response.on('end', function() {
        if (!pagedata.match("title_backsoon1.gif") ) {
          console.log("success!");
          notify("Apple Store online!", "GOGOGO!!!!");
        } else {
          console.log("nothing...");
        }
      });

    } else {
      console.log('STATUS: ' + response.statusCode);
    }

  });
}

refreshPage();
var polling = setInterval ( refreshPage, 30000 );

That’s it, without much comment. It’s not very efficient and has to manually stopped when the store is back online, but it did help me to order my iPhone 4S without any lags (a minute later, the site went slow as hell). If you have questions, don’t hestitate to contact me.

Replacing the optical drive in a 2011 MacBook Pro 13" with a SSD

2011-03-11T00:00:00+01:00

After five long years with my original white MacBook and given the fact that Lion won't support its 32 bit Core Duo processor, I decided it was time for a new laptop. As I was really pleased with the form factor of my old 13" MacBook (after all, I have a big screen iMac upstairs) I ordered the new 13" MacBook Pro. Now there was one thing that really bothered me: I had this nifty fast 64GB SSD (Supertalent Ultradrive ME) in my old MacBook that I really didn't want to give up and I couldn't bear anything in my new laptop being slower than in the old one. On the other hand, 64GB on a new machine seemed far to little (and larger SSDs are still not really affordable). So the only solution was to order an adapter for the optical drive bay to keep the 320GB HDD while adding the SSD as a fast drive for the system and applications. Here's how I did it.

At first, the SSD needed a fresh start. To really erase all data and reset the internal block allocation tables (i.e. to bring it back to full speed), you need to to a security erase on the device. There exist a lot of ugly windows tools for that, but I went the following way: Using an Ubuntu boot CD, I booted Linux on my old MacBook (by holding c on startup) and went to the console. The following commands should have been enough to reset the SSD:

hdparm --security-set-pass NULL /dev/sda
hdparm --security-erase NULL /dev/sda

However, all I got was an Input/Output error. That's because something (probably the EFI) locks the internal drive on boot and prevents this kind of commands for it. The solution sound dangerous, but works without a problem: Just pull the SSD out while the computer is still running, then plug it back in and it won't be locked anymore so you can fire up those hdparm commands.

Next was preparing the optical drive bay. This step was quite easy: The SSD just needs to be plugged in. To fix it, I used a small strip of tape and two hair ties (don't ask).

Now to the exciting part - voiding the warranty on day one. First, I got something to collect all those little screws. I used a milk tooth jar (again, don't ask). Then I removed all the screws on the back side of the macbook.

After removing the back cover, you can directly see the optical drive and the 3 flat cables which need to be unplugged first. To prevent electrostatic discharge, during the whole procedure, I touched a grounded spot every few seconds.

To remove the optical drive, a small plastic part carrying the wifi chip and the speaker needs to be lifted a bit. Fortunately, no antenna cables have to be removed, just 3 screws. This part needs to be shifted to the right just enough to unscrew one of the 3 screws that hold the optical drive. The other two screws are easy to find and reach. After they have been removed, the drive can be carefully lifted out.

That's it for the hard part, I thought, but unfortunately the optical drive adapter I bought wasn't really suited for the 13" MacBook Pro. Using a saw and a file, I needed to remove some of the plastic on the front side of the "drive" before it fit (see the image below). I was a bit supprised that the drive bay didn't need any screws, but as it's not a moving part and the little plastic thing with the wifi chip and the speaker on it presses it into its position, it seems quite stable.

After assembling everything, I started up the MacBook and to my suprise, everything still worked. I then used Carbon Copy Cloner to clone the original drive to the SSD, formated the HDD and set up some symlinks (for example the Movies and Downloads folder) to point from my home directory (which is located on the SSD) to the HDD. Finally, I set the SSD as the startup drive and rebooted. Voilà: Super fast and not too expensive while still offering enough space. The only drawback: There's no optical drive anymore, but I can't remember when I used that the last time on my old MacBook...

PS: Sorry for the bad image quality. I only had my iPhone 3GS at hand.

The future of Java wears a suit and tie

2010-11-06T00:00:00+01:00

Not even a year after Oracle's acquisition of Sun was completed, they start doing what everyone expected: Pushing prices an killing off free products. Now they're ready for the next big step: Earn big money from Java. In a first step, there will be a "premium" JVM besides the free one.

Let's guess what comes next. Many people say Oracle is destroying Java. They're both right and wrong. Right because moves like this kill the spirit of the community. Right now, there's a Java open source library for nearly every problem that can be solved using computers. Many of those libraries are written by individuals for fun or fame, not for profit. I believe those people will soon turn their back on Java and move on to languages where "the other cool guys" already are (ruby, python, javascript or whatever will be the next big thing).

But that does not mean Java will die. Even today, Java is seen by many programmers as a boring, verbose enterprise language. Big companies move slow and only when they have to. They don't have. Even if at some distant point in the future, Oracle will stop supporting OpenJDK, someone will make a large Excel sheet and find that paying a few thousand bucks to Oracle every month is cheaper than even thinking about alternatives. There will always be enough Java programmers, at least as long as Schools and Universities teach it as the first language. And why should they stop doing so when the corporate world has great demand for Java programmers?

In the long run, Java will become exactly what the haters are calling it: Boring corporate stuff. Of course that was inevitable, but Oracle sure does the best it can to accelerate this development.