On 8/7/07, Colin Walters <walters(a)redhat.com> wrote:
Personally - I'm a bit skeptical of Tracker actually being better
than
Beagle. I don't have anything substantial, but my instinct on this is
that writing a desktop search engine is actually really hard - not
barfing on bad files (inherent tension with supporting lots of file
types), figuring out all the touchy details of when to do the
indexing[1], etc., and that unlike Beagle, Tracker actually hasn't seen
wide deployment so it hasn't yet been truly tested.
I don't expect any implementation of an indexer to be perfect. But
what I desperately need, is an implementation that expects to run into
problems and knows how to communicate back to the users when it runs
into a problem indexing a file... so we can attempt to actually report
something credible and reproducible. My biggest beef with beagle
wasn't that it sat there and chewed system resources. My biggest beef
with beagle was that there was no obvious feedback to a user nor a way
to query what the exact nature of the problem is while its happening.
if tracker doesn't do a better job of problem feedback, it's not
going to taste better.
-jef