Everything as an amazing Metadata/Database file system

General discussion related to "Everything".
Post Reply
twinbee
Posts: 2
Joined: Wed Jan 05, 2011 12:00 pm

Everything as an amazing Metadata/Database file system

Post by twinbee »

I've just started to use Everything properly, and firstly, I'm amazed at how fast it is. It is blinding. I'm curious - what kind of techniques does Everything use to get the search so fast? I'm assuming some kind of advanced tree structure maybe? I'm guessing C/C++ was used to program Everything? Congratulations on an amazing program - I look forward to creating a writeup on it. (I wish Windows could use it by default).

Now for the main meat of my post: Does anyone use Everything as the basis for a metadata/database filesystem? The idea of such a system is to save all and any files in a single folder (or at least not worry about creating any folders), and simply use Everything to retrieve anything in future. It sounds scary, but as long as the filename contains some of the keywords you might use to search for that file in future, it's far more effective and efficient to save and locate files like this. I created an article which speaks about such a system many moons ago:
http://www.skytopia.com/project/article ... ystem.html
"Towards a single folder metadata (database) filesystem"

Because of the amazing speed of Everything and its wildcard features, it now seems practical to switch to this system completely, and if it proves as effective as I think, then I owe Everything's creator a massive debt. Once again, thankyou for this wonderful piece of software!!
jar
Posts: 2
Joined: Wed Jan 05, 2011 7:17 am

Re: Everything as an amazing Metadata/Database file system

Post by jar »

twinbee wrote:The idea of such a system is to save all and any files in a single folder (or at least not worry about creating any folders), and simply use Everything to retrieve anything in future. It sounds scary, but as long as the filename contains some of the keywords you might use to search for that file in future, it's far more effective and efficient to save and locate files like this.
Why just a single folder? As in many situations the foldername is just the metadata you need. If you keep to your naming-convention.
I use Everything already as you mention. Sometime I search on a file name and sometime on a folder name or a combi of both.

Whatever the file is I find it in seconds.

Jar
twinbee
Posts: 2
Joined: Wed Jan 05, 2011 12:00 pm

Re: Everything as an amazing Metadata/Database file system

Post by twinbee »

Yes, the general idea is that the file can be given the folder name as just another tag (potentially batch tag for a group of files). Thus saving you the need to create a folder in the first place (or navigating the hierarchy to find it later). Obviously, I'm not saying to go about removing all your current folders, as they're already there now. But rather, I'm saying using the system would reduce or potentially eliminate the need to create any further future folders.

I suppose the whole point of something like Everything is to allow us to be less careful *where* we place our files, as the penalty for finding it later is reduced with an instant search. Using a single folder for a database filesystem is taking this concept to the logical conclusion.

Pragmatically, you'd still need folders for things like programs (which often depend on a folder setup), but for our own files/data and misc files, the need is drastically reduced. Potentially completely eliminated if standards come into place.
dave5836213
Posts: 7
Joined: Tue May 31, 2011 12:16 pm

Re: Everything as an amazing Metadata/Database file system

Post by dave5836213 »

Yeah -- I'd love to get a sense of the DB structure -- it is amazingly fast and compact

via typical indexing, I'd imagine it may index things on two or three letter combinations then use a more conventional algorithm for the rest but I don't think that's indeed the case

the most interesting thing about the searching functionality is that I can search for text in the middle of a string -- most indexing doesn't support that...
Post Reply