Google have released their Desktop search application thingummy for Linux, and like a good little sheep, I’ve been and gone and installed it. All in the interests of science, you understand.
One swift download and install later, and it’s churning away merrily in the background, indexing my voluminous home folder and XP partition slowly, but with pleasingly low CPU usage. I’m quite happy with it taking a few days to index the whole thing if it means there’s no appreciable slowdown while it’s doing it, so that’s good.
In theory, Google Desktop is a Good Thing. Hit the Control key twice and up pops a search box. Enter whatever you’re looking for and immediately see a list of all the relevant documents, emails, etc. Click one, and it opens.
There’s only three problems to Google Desktop, as I see it. First off, it doesn’t work. Secondly, it’s not customizable enough. And third, Beagle does it better. Some people would add in privacy concerns regarding the kind of information being sent back to Google about your hard drive’s contents. That doesn’t bother me in the slightest, though I can understand those concerns in the piracy rife world of Windows.
I say it doesn’t work because of two problems I’d found with the system so far. Firstly, the search box doesn’t support regular expressions by default. In the Linux world, that’s an unforgivable omission. Also, under Ubuntu GNOME at least, clicking on a text file opens up gedit, but doesn’t actually open up the text file itself if the filename contains a space. I’m left staring at a blank window. This latter problem is a bug, so should be fixed asap; I’d hope that regex support comes in Real Soon too, or Google Desktop isn’t going to spend much time on the hard drives of hardened Linux users.
When it comes to customization, the choices are limited. I can set how many search items to display, which folder to search (or not), select from a short list of filetypes, and choose whether to send information back to Google. What I can’t do is set preferred applications. It can’t index using image EXIF data making it all but useless for photo searches. I can only sort by relevance or date. If you want to search by date last accessed, filesize or even by alphabet order, you’re out of luck.
Beagle handles all of this, and more. It’s the de facto standard on Linux for this kind of localized intelligent search systems, and for good reason. Google Desktop just doesn’t come close in terms of usability, flexibility or integration. Google aren’t going to win many fans by putting their Desktop search app right at the top level of the menu system either. I’ll lay big bets that’s going to be removed in 99% of user cases.
One problem with both Google Desktop and Beagle is the size of their indexes. 20Gb of files generates around a 2Gb index. That’s 10% the size of source, a far too large percentage and a significant storage hit for something that’s little more than a useful utility. This needs to drop to 1-3% of source size to be acceptable.
Where Google Desktop wins out though is in it’s extremely low CPU utilization. Beagle could learn a trick or two from that, for sure. Other than that, Google Desktop has some way to go.