Re: Google Desktop for Linux Released

Conrad Knauer wrote:
>> How good/bad is Google desktop at this?
>
> My home folder has ~20K items taking up ~108 GB.
> The .google directory is ~135 MB.

Given the likely greater utility of the results, I can forgive a 10x
increase in the file size.

> Note that I am not using the feature that lets me keep/search
> old/deleted versions of files (that's probably why they suggest that
> you have ~1GB of space)

Neat. Versioning. OTOH, if other people can get to it, that might be a
drawback (privacy). I'm curious if it'd handle a versioning FS well (or
if it'd dupe the previous versions as in an ever increasing number of
files).

>> Is it easy to only have it index certain directories?
>
> Yes :)
>
> Here's a screenshot of the preferences:
> http://members.shaw.ca/Limulus/misc/google_desktop_options.png
>
> By default it also indexes various man page dirs, but I removed those.

I'm going to say that's a crantastic feature. It's fresh like a certain
citrus fruit, and fantastic all at once. Based on the screenshot, it
looks like the engine is an HTTP one, meaning you can (potentially)
search across your LAN w/o much fuss (which was also a touted
enhancement to Spotlight in MacOS 10.5+). Unlike Spotlight (which
really pokes my peaves), this sounds useful. I can imagine it being
very handy to be able to search man pages across (for example) a
distributed build cluster, allowing me to quickly see the differences
between BSD, Linux, and MacOS X versions of a particular non-POSIX
syscall (or their non-POSIX extensions) in maintaining portable code.

Apple's Xcode tool is internally all HTML for its documentation pack,
and uses some sort of search engine to determine relevant hits for APIs,
examples, etc. If Google Desktop for MacOS can index those, I'd have a
reason to download and install that.

>> Also, I'm curious if such an app phones home
>> or otherwise fiddles with bits not its own.
>
> http://desktop.google.com/en/linux/privacypolicy.html
>

This also sounds reasonable, but I wouldn't trust it unless I could
verify a packet dump of traffic (or peruse the source, but that's not
going to happen).

Call me paranoid, but closed-source stuff generally has to earn trust in
my book.
Received on Tue Jul 3 20:03:17 2007