Neil Turner's Blog

Blogging about technology and randomness since 2002

Google Desktop Search – observations

I’ve been using Google Desktop Search for a day now, and have a couple of observations to make:

  • Integration with the browser is seamless; when you do a search from the quicksearch box in Firefox, it will also show results from your computer in the results. All Google pages also gain a ‘Desktop’ tab to run the same search on your own computer.
  • It’s very fast, although the index of my HD is around 120MB. Each user has a separate index.
  • There is a slight incompatibility with internationalised builds of Firefox. Every time the search app starts, it checks for a file called google.src in the searchplugins subfolder of the Firefox application folder, and if it doesn’t exist, it is recreated. This file exists by default in the en-us builds, but not necessarily in others; in the en-gb build there’s a google-uk.src file instead which means Google searches are sent to google.co.uk instaed of google.com. This means that you end up with two Google entries on the quicksearch menu in the browser.
  • The Trillian Pro plugin only indexes conversations made from when the plugin was installed – old conversations are not explicitly indexed. They do, however, appear as the contents of text files and may be included in search result pages but their display isn’t as intuitive as it could be.
  • Viewing cached copies of files strips out all formatting.

4 Comments

  1. Thank god you can open the containing folder, which was my biggest complaint of the beta. Still, it would be nice if you could view the extended file properties/meta data from the retults page itself.
    The data files for me add up to 535MB! haha

  2. 332 MB for me. I have the all text, help, OpenOffice, and Trillian plugins installed. I agree with your statements. Copernik (or however it is spelled) is okay, but didn’t like the interface. I like this new version much more.

  3. what’s the memory footprint and your impressions on performance?

  4. Memory footprint is about 10MB idle. Performance-wise it’s very, very quick.