I have decided to no longer continue to blog in English on this site. I will still be blogging in Dutch.

It’s been increasingly difficult for me to maintain a certain continuity here. It’s been a great experiment, and a lot of fun, but I feel it’s time to let this one go.

I’m working on a new site with a much more narrow focus, something this site dearly lacks. It might still be a while before it sees the light of day, so I’m not revealing too much about it yet.

A big thanks to anyone who has been reading this blog, leaving comments or just following what’s happening. It’s been great!

As an early new years resolution I’ll start with sharing more of the snippets I code up to make my life easier. Here’s one : the SRT subtitle scaler. It’s a little ruby script to sync subtitles with a movie. I don’t know how often one needs this, but I got some English subtitles on-line for a Chinese movie and I just couldn’t get them to properly line up with the movie.

Mplayer has an option to set a delay (possibly negative), but that didn’t help in my case. You can also set the subtitle fps (frames per second), but that only works when the subtitle format uses frame numbers, while SRT files use timestamps.

If you have this problem this script can help. There are usage instructions in the comments.

Enjoy!

By accidentally using the right four-word combination on Google I stumbled upon a 2002 article by Mark Pilgrim, and somehow it prompted me to use that good old blockquote tag once again. These are the closing words of the article, citing himself two years before that. Spot on Mark, spot on!

As I write this, the year is 2000, and the Internet is a battleground of intellectual property disputes. Some people would like you to believe that, without proper financial incentives, music, literature, and computer software would disappear. After all, who would make music if they can’t make money on it? Who would write? Who would program? I know the answer. The answer is that musicians will make music, not because they can make money, but because musicians are the people who can’t not make music. Writers will write because they can’t not write. I’ve been programming for 16 years, writing free software for 8. I can’t imagine not doing this. If you can imagine yourself not doing what you’re doing, do something else. Do whatever it is that you can’t not do.

A while ago the fabulous folks from Ubuntu released the latest version of their popular Linux distributions. The version number of Ubuntu, Kubuntu and Edubuntu is now up to 7.10 (being the year and month it was released). More often however Ubuntu versions are referred to by their alliterating nicknames. I’ll go over my experiences upgrading from Feisty Fawn to Gutsy Gibbon.

Upgrading from Feisty

Being a Debian based distribution all software management can be easily performed by the ‘apt’ family of tools. The first step in upgrading is letting apt know that I would from now on prefer the Gutsy packages.

sudo sed s/feisty/gutsy/g -i /etc/apt/sources.list
sudo apt-get update

You might notice that there is no gutsy-commercial repository like there used to be a feisty-commercial, with for instance Opera and Realplayer. If you were using it you’ll have to comment it out or keep using the one from Feisty.

When I do these big upgrades I switch to a plain tty console because I’ve had it happen that one of the install scripts restarted X11. So press SHIFT-F1 and login to the console. To have some record of what happened afterwards I use the great little utility ‘tee’. This will record whatever passes stdout to a file. So to start upgrading do this:

sudo apt-get dist-upgrade | tee feisty-to-gutsy-`date +%Y%m%d`.log

This way you’ll get a nice date stamped log of the whole install process. If you switch back to X (Shift-F7) to do some work in the meanwhile, you can open a terminal window and do a ‘tail -f’ on that file so you know what’s happening

First hickup : device-manager

After some 1300 packages were upgraded I rebooted my computer. A new kernel (2.6.22-14) had been installed, and it seemed to be making some trouble. Instead of booting into X I saw per second several identical error messages flying over my screen, the gist of it being:

device-manager: linear: dm-linear: Device lookup failed

Booting with my previous kernel solved the problem, so I got on-line and after some googling found a fix, in /etc/evms.conf find the line that says ‘exclude=…’ and change it to:

exclude = [ sd* ]

I’m not even sure what EVMS is, I’m guessing this is related to HAL (Hardware Abstraction Layer). What this does is exclude your harddrives, from eh.. something. It solved the problem for me.

I’ve had kernel upgrades in Ubuntu create problems before, IMHO they could be a little bit more conservative with their kernels.

Smaller hickups

After booting again everything was working like a charm. They also did a great job again on the artwork. I had to tweak a few more small things.

Mozilla’s Thunderbird email program is now started simply with ‘thunderbird’ in stead of ‘mozilla-thunderbird’. I had to update a shortcut.

I use Emacs for most of my editing tasks, and somehow the version with XFont backend that I previously had installed got replaced by another one. The version with XFont support is not available in the Ubuntu repositories so this is understandable. There’s a guy that packages this pretty version, and apparently he has since I last checked made an actual apt repository available. See this article : Pretty Emacs by Alexandre Vassalotti.

Looking closer I found to have three different Emacs flavors on my system. I just removed them all, added the repositories from Alexandre and installed my trusted XFont version. Nice!

There seems to also be something going on with StarDict. The little windows that pop up with translations jump to the upper left corner when you hover over them, and then immediatly disappear. I might have to revert to the old version.

Conclusion

I’ve been upgrading my system since Dapper, and although there are often small hickups a little bit of google will solve most of your woes. I’ve only used my new system for a day now, so there’s probably lots of new little features that I still have to discover. All in all this seems to be another rock solid distribution which I’d wholeheartedly recommend to both newbies and seasoned hackers.

Last week saw the launch of the first version of ZhongWiki. A wiki for people learning Mandarin Chinese. It’s still in it’s infancy, but I have a feeling it might become a great site and possibly a vibrant community.

A wiki is a site that can be edited and expanded by basically anyone. The best known example is Wikipedia. This site however does not try to be an encyclopedia, but rather a resource and a meeting ground for students of Mandarin. This means that not everything has to be properly sourced and you don’t need to have a neutral point of view. Opinions are appreciated and encouraged.

The name ZhongWiki is a wordplay on 中文 which means Chinese language and is pronounced “zhong wen” in Mandarin. It is written in CamelCase with the “Z” and the “W” capitalized as a homage to early wikis which used CamelCase to create links in pages.

So head over to ZhongWiki if Mandarin is your cup of cha, and keep a lookout for the edit button.

Last week the Smiley turned 25. Happy birthday and congratulations! It was proposed by Scott E Fahlman in a post to a “Bulletin board” at Carnegie Mellon University the 19th of September 1982. Scott proposed its use to mark humorous or sarcastic remarks so they wouldn’t be misunderstood.

These two pages have more info on the origin of the Smiley

• The First Smiley
• Smiley Lore

I have somewhat neglected my English speaking audience, I must apologise. I’ve recently moved to Taiwan to study Mandarin, and my travellers journal will mostly be published on the Dutch side of this blog.

Before I left to Taiwan my grandmother gave me a beautiful little camera, and I’ve been taking pictures like a madman and posting them to my Flickr account. I would like to include these pictures into my blogposts, but I would like to use a specific HTML template, so I can easily style them with CSS, like so :

Giraf schenkt David en Arne thee

Because of this the Flickr generated code and most other solutions out there are no good. Instead I decided to code up my own solution, scratch the old itch so to say. In the meantime you guys can enjoy a little example of a Nitro app in action. It’s a wee little thingy, but it shows that with Nitro you don’t get much overhead in your code. If you make a small Rails application, the noise to signal ratio is huge with so much pregenerated stuff you’re not gonna need.

So the target is just a single page with my most recent pictures, and next to that a text area that contains copy-paste ready HTML in my desired format. I coded this up in a small hour, with most time spent reading the Flickr API docs. I tried the flickr gem, but it’s no longer maintained and I couldn’t get it to work quickly, so I started my own wrapper just for the API calls I need. (YAGNI, right.)

The “app” is called Flikker (which is a not-so-polite word in Dutch BTW, wonder how many search engine hits this will bring me). Download it here Flikker v0.0 as a tarball.

To run:

• in settings.rb fill in your Flickr API key and user name
• adapt ‘launch’ to point to your Nitro repository
• change conf/debug.rb if you want to use something other than mongrel
• ./launch
• visit localhost:9000 in your browser

Actually this is such an embarassingly small app I wasn’t going to put it out there, but it’ll be damn handy for me, so maybe somebody else has a use for it also.

Soon my first professional contract comes to an end, and I will have time again to develop my newsfeed addiction.

Previously I used Bloglines, which is certainly one of the better on-line feed readers. I’m switching to a desktop feed reader however, I’m trying out Liferea.

I’d like to add all the Planet Ugent blogs to my list of feeds, but the available OPML contains no URL’s. Pretty useless this way. Nothing is lost however, since they still have a FOAF export. The FOAF however contains only the blog URLs, not the actual feeds.

Ruby to the rescue!

require 'rubygems'
require 'open-uri'
require 'hpricot'

doc = Hpricot open('http://planet-ugent.be/foafroll.xml')

def extract_feed(url)
  return 'error://no url' unless url && url != ''
  begin
    page = Hpricot(open(url))
    %w(application/atom+xml application/rss+xml).each do |t|
      link = page.at("link[@type=#{t}]")
      if link
        link = link.attributes['href']
        if link =~ /^\//
          link = url[/^[w]+:\/\/[^\/]+/] + link
        elsif link !~ /^[^\/]+:\/\//
          link = url[/^.*\//] + link
        end
        return link
      end
    end
    'error://not found'
  rescue Timeout::Error
    'error://timeout'
  rescue SocketError
    'error://socket error'
  end
end

feeds = doc.search('foaf:member').map do |m|
  name = m.at('foaf:name').inner_html
  url = m.at('foaf:document').attributes['rdf:about']
  [name, extract_feed(url)]
end

puts %(<opml version="1.1">
  <head>
    <title>Planet UGent
    <dateCreated>#{Time.now.rfc822}
    <dateModified>#{Time.now.rfc822}
    <ownerName>Ikke
    <ownerEmail>eikke at eikke dot commercial
  </head>
  <body>
)

feeds.each do |name, url|
  puts %(    <outline text=”#{name}” xmlUrl=”#{url}”/>\n)
end

puts ”  </body>\n</opml>”

This script will extract the FOAF names and urls, load each page and extract the feed. Atom feeds get precedence over RSS feeds. It should be able to handle relative URLs, but this is not thoroughly tested. The OPML is written on standard out.

Find the result here.

Today I released my very first public Ruby project!

Pinyin is a pure Ruby library with a specific focus : converting between various transliteration systems for Mandarin Chinese.

Over time various systems have been devised to capture the phonetics of the Chinese language, often as a teaching aid.

This is the first public release. Pinyin for Ruby can currently convert between five different systems, with more to come.

- Hanyu Pinyin
- Wade-Giles
- Tongyong Pinyin
- Zhuyin Fuhao (BoPoMoFo)
- IPA (International Phonetic Alphabet)

It does this by decomposing each syllable into its three components (initial, final and tone) and then recomposing the syllable in a different system. This approach was chosen over a table based approach because it better captures the way each system works.

Conversion tables are included to test the package.

There’s still plenty to be done, but since it already has some nice capabilities I’m putting this out in the open, to see if there is any interest. Next step is to move from my private SVN repository to the one on Rubyforge. The project also needs a homepage.

Give it a whirl! Simply type

gem install pinyin

(Adding ’sudo’ where necessary.)

Important 2007-12-01 : This article is deprecated. The location of the repository changed and I haven’t had the time to update my scripts to reflect this. With version 0.50 approaching an easier way of trying out Nitro is on the horizon. I’ll write a new article soon that will supersede this one. If you need help getting Nitro up and running please ask advice on the Nitro mailing list.

Update: Changed the RUBYOPT line to load rubygems and to be compatible with the RSpec specs.

As I’ve mentioned I’m becoming involved in the Nitro community, a web framework for Ruby.

The latest release is 0.42, but it’s an old, outdated release and is no longer recommended. In stead we urge everyone to use the repository version, wich in time will become 0.50. The version control system used is Darcs, a powerful distributed VCS written in Haskell. However since darcs is less popular than say Subversion or CVS, I decided to make a daily tarball. This way you can try the repo version without having darcs installed.

It can be found at http://arnebrasseur.net/glycerin/nitro-latest.tar.gz. This is a symlink to a timestamped file in the same directory. A new one should be created daily around 9.30 CET (8.30 GMT).

To get started with it you need to set up your ruby load path. The easiest way to do this is to load the glycerin script, which will set up the load path to include the various Nitro components (Nitro, Raw, Og and Glue).

I have this in my ~/.bashrc to do this behind the scenes.

export NITRO_REPO=~/work/nitro
export RUBYOPT="-r$NITRO_REPO/script/lib/glycerin -rubygems"

You will also need Facets, Xmlsimple and RedCloth. These can be installed as gems. To run the blog example you also need the uuidtools gem.

There’s also a nitro command to start a Nitro app with various options. It can be found under nitro/nitro/bin, so I added this to .bashrc:

export PATH=~/work/nitro/nitro/bin:"${PATH}"

An overview of the command line options (and other assorted Nitro articles) are available on my wiki.

To get a feel for a typical nitro application, take a look at examples/blog.

Please report any difficulties you had with getting the repo version up and running.