No File Names!

From time to time I believe it is important to re-evaluate fundamental assumptions. In computing, one of those assumptions is the way that files should be organised in a file-system. The commonly implemented POSIX standard organises files into a tree structure and (ignoring sybolic links for the moment) gives files a single name.

Whilst this seems like a great model, and works well for small numbers of files, it becomes problematic on a larger scale. Not problematic for the computer: this system suits the computer just fine. It's the human beings who suffer.

Henry Newman's article File System Management Is Headed for Trouble (Feb 2009) suggests how and why more information should be added to a file. (Thanks Oly for pointing me at that.) He includes in this the addition of metadata that the users and the programs manipulating the file can add. What I am suggesting is that this data more or less replaces the 'file name' as we currently recognise it.

Please see the presentation and try out the proof of concept software. (See below.) Then start to agitate for change! Let's wake up and realise that file names are not what we want.


For the February meeting of the Kent Linux User Group I prepared a presentation intended to provoke thought and discussion on this issue. View presentation...

proof of concept

As an exercise in learning how to write software in C for the Linux environment I run at home I wrote a proof of concept. This is not intended to be a solution, but to be a starting point to play with which might trigger some ideas. I've never composed a source code TAR file for distribution before so I have no idea if this is right, but you are welcome to download the TAR file if you want to experiment with it. See the README file it contains for information on how to set up the MySQL database.