Symfonys behaviour to rename uploaded files with a sha1-hash of the original filename and a randomnumber between 11111 and 99999 [why those numbers ?], ( done in sfValidatedFile::generateFilename ) is pretty ugly. No more seo-friendly filenames, filenames where you can actually have a clue about what's behind an url or how the creator supposed it to be named.

Nice and simple usability-boost for the filtering in the backend if you use it often and don't want to press the filter-button each time.
But it too can be annoying if you often filter based on date-fields and don't have a datepicker, because selecting a range is 6 changes and each time it get's submitted.

<script type="text/javascript">
  $(document).ready(function() {
    $('.sf_admin_filter select, .sf_admin_filter input[type="checkbox"]').attr('onchange', 'this.form.submit()');
  });
</script>

Symfony offers a fast and easy way for setting up admin backends including support for image uploads. You only need to change the fields widget and validator in the Form Class and symfony handles the rest:

  public function configure()
  {
    $this->widgetSchema['image'] = new sfWidgetFormInputFileEditable(array(
      'label' => 'Field Name',
      'file_src' => '/uploads/images/'.$this->getObject()->getImage(),
      'is_image' => true,
      'edit_mode' => !$this->isNew(),
      'template' => '%file% %input% %delete% %delete_label%'
    ));
    
    $this->validatorSchema['image'] = new sfValidatorFile(array(
      'required'   => false,
      'mime_types' => 'web_images',
      'path'       => sfConfig::get('sf_upload_dir').'/images',
    ));
    
    // delete checkbox
    $this->validatorSchema['image_delete'] = new sfValidatorPass();
  }

Combining that with the power of sfImageTransformExtraPlugin to create thumbnails of images based on profiles is really great:

When refactoring code or tests I like working until I break something. When my phpunit tests switch from green to red, I want to know it immediately without restarting the tests manually each time I change something, because that's annoying and a waste of time.
So after thinking about running the test-suite after each save in Netbeans, which I didn't find a way to do it and for unittests which run longer than some seconds, it wouldn't probably be such a good solution if after saving some changes the testsuite will take minutes to run. So I came up with another simple solution:

The new Truecrypt 7.0 release is almost 7 times faster compared to 6.0 on my i7-620M with AES-NI. It is some hundred mb/s faster now than dmcrypt (which runs my system-encryption on Debian Squeeze), but that is expected since truecrypt makes use of multiple cores AND aes-ni and dmcrypt only supports 1 thread per mounted device, so unless you create a RAID consisting of multiple dmcrypt-devices, you can only use 1 core.

Another nice thing, they ported the included benchmark-tool from windows to linux. In 6.3 there was no benchmarking when running under Linux, but now it is supported.

Truecrypt achieves between 1600 and 1700 mb/s. I ran the benchmark for 1mb, 5mb, 10mb, 50mb, 100mb and 1024mb, skipped 200mb and 500mb, because I was tooo lazy :)

Here are some images:

If you are looking for an easy way of using an up to date version of Chromium (the free open-source version of Goole's Chrome browser) on Debian Squeeze you should try one of the Ubuntu Chromium Projects PPA Repositories, they have a beta based on the version used for the current Chrome Linux-Builds and and a daily-trunk based one.

Important thing is that you use the karmic-repository, lucid didn't work for me.

deb http://ppa.launchpad.net/chromium-daily/ppa/ubuntu karmic main 
deb-src http://ppa.launchpad.net/chromium-daily/ppa/ubuntu karmic main

deb http://ppa.launchpad.net/chromium-daily/beta/ubuntu karmic main
deb-src http://ppa.launchpad.net/chromium-daily/beta/ubuntu karmic main

The package is named chromium-browser

A benchmarks showing how awesome fast dmcrypt is with AES-NI.

New notebooks are sometimes a little pain in the ass to get working with Linux. After more than 4 weeks, my new notebook finally arrived last monday. Monday this week I tried installing Debian Squeeze (still is testing) on my new Latitude and after 2 hours formatting the drive with dmcrypt, failed with a not starting X, not finding a screen. I didn't have the time to check it out because I needed a running System with OpenGL and Virtualbox to run my old Windows XP-Image using VisualStudio with the setup OpenGL-Stuff for university, so I went with Windows 7 for the first 3 days.

My Latitude E6510 specs:

• Intel i7-620M 2.66 GHZ Dualcore with AES-NI and Turbo Mode (biggest dualcore with AES-NI offered with the new Latitude-series)
• 4GB RAM (2 * 2GB) (8GB would have been 250€(+taxes) more)
• 250GB @7200RPM HDD (had 120GB in my old one, but never used it, since all big data normally goes on to external drive or the raid in my server)
• 15.6" @ 1920x1080 (After working over 4 1/2 year with 1024*768 it was time to get big)
• Nvidia Quadro NVS 3100M (well no player-card, but at least it seems some old games run on it)
• Intel Centrino Advanced-N 6200 Wifi

Today I helped somebody buy a cheap notebook as a replacement for a really old PC.
In the end we went home with a Samsung R530 and I immediately started configuring it.

After the installation I was looking on a lot of crap on the desktop and the startmenu.
I knew most notebooks today come with a lot of preinstalled crap like MS Office, ISP-software (AOL, T-Online, ... ) , virus-scanner, firewalls and more stuff (the same here, 60-day-trial for the new office, a McAfee Security-Solution (virus-scanner, anti-spam and more), ...) but I didnt see THAT coming.
Lots of games and programs from a company I already forgot the name, which which where usable for some hours but then you would have pay for them. I uninstalled them, even though they were from the same company and only some mb in size, each had it's own uninstaller. But I figured the low price for the notebook is partly due to those programs.
Then I discovered a popup for the most awesome program ever Phoenix FailSafe ... a product which helps you when your notebook gets stolen. It tracks it, deletes/downloads your data, makes a photo/video of the thief and stuff like that if he goes online or is in reach of any accessable gps-device the notebook can connect to (bluetooth or whatever) . It was a trial too and you would have to pay for it on a monthly base.

The first thing I did was searching for an uninstaller in the starmenu: nothing.
Then in Windows software-control: nothing.
Then I searched on google and found a thread about it. No real solution.

You can set it up, register, then login on their page, deactivate it and then download an "uninstaller"! Register to uninstall it ? WTF!!!

According to an entry in the forum, the support said, you can "uninstall" it manually by deleting the directory, the files and the start-menu entry by hand ...

What kind of crap is that ? A silly way to protect it from beeing deleted by a potential thief ? I would accept it if it would remove it's entry in start-menu and uninstaller after you have choosen to use it. But not even providing one ?
I will give the manual way a try next week and look if there is more data from this tool in the registry, autostart, services or anywhere else.

The HTML Validator-Extension for Firefox doesn't run on Debian Squeeze out of the box, the problem is the missing libstdc++5, but only libstdc++6 is installable from debians default-repositoy.

Well today I wasted about 12+ hours trying to compile firefox + html validator-extension from scratch, made many mistakes, first I used ff3.5-sources, then I found out that the instructions where not meant for firefox 3.5, but for firefox 2, then some typos on patching source-files that I only foundout about ... 40 minutes later, because building it took between 30 and 60minutes each time ... That's why 12 hours past so fast.

But finally I found a way to get it running. libstdc++5 wasn't the only problem :)

Using the solutions provided for other linux distributions, mixing them a bit and doing some testing myself I finally got a solution: