> The idea is a "desktop database" that maintains indexes of all of the
> information on your desktop file hierarchy for advanced queries. If the
> maintainence isn't done in realtime, it's at least done, say, at a
> user-defined interval like once a day.
The Remembrance Agent is an Emacs function that displays other files with
similar context as your current buffer, working as you type; part of the RA
package is a db tool that creates a queriable database of all the
files/directories you specify, and you can search it outside of Emacs. You
could run a cron job to update the db.
> Come to think of it, I could also use an "advanced desktop caching
> agent" that stores copies of EVERYTHING that enters and exits my machine
> as I receive and send stuff in email, news, chat, Web... I guess it's
> not so much a cache as a lifestream.
>
> Add the lifestream to the desktop database, and suddenly everything
> that's taken brain cycles away from me is indexed and queryable. A
> "history" of all computer activity that is never deleted.
I've been wanting such a thing for years. I would even want to take it
farther than just creating lifestreams of your files, but would want it to
save timestamp info on text that you enter, so you could go back to any text
and determine the exact time (according to the system clock) that it was
entered -- as well as include the ability to play back this text at the
exact _speed_ it was entered, backspace characters and all. A realtime os
would probably be necessary.
> I wonder how much information on the Web is currently unsearchable
> because it hides embedded in Powerpoint files or jpeg's or gif's or
> PostScript files or compressed files.
Yup, you can't grep PostScript. Anyone been following search
tools/algorithms for searching non-text data?