Wednesday, June 16, 2010

A day in the life in diorama

Let me tell you how I work.

I keep all my documents in a hierarchy of folders.
The folders are all named tersely, but human-readable and obvious.
I put files together that are in related projects.

I maintain this hierarchy of folders in CVS, which allows me to archive it, along with dated versions of everything I add or change in the hierarchy.
I buy time on a cloud server and run several processes, one of which stores my hierarchy, in sync, archived.

My data are protected by industry standard security which would be good enough protection against any except the highest level hackers with access to large compute clouds (e.g. the government).

OK. Everything above is factual, extant, and uses existing open source technology. Everything that follows in this post is a dream, combining how I work, and new, open source software to model and automate what I now do manually, and what I cannot currently do, but wish I could do.

In addition to this hierarchical, archiving, versioning, distributed file system, I maintain my metadata in a system called Diorama. Diorama stores all the information about my files, and maintains this as a context that can be searched, executed in, added to. I can choose to add new items to my own Diorama-run file system, or I can add new items that are stored out in the cloud, in which case Diorama simply stores the URL and any metadata that get associated with that document. I use the Diorama-run file system for text and small images, and remote storage for larger, binary files. You could use any file system you want, and just have Diorama maintain metadata and indices to documents only.

In addition to tags and categories, where you were in the context at the time of document status change, can be saved in the metadata as well (if you allow your context history to be saved in metadata or metadata links to your context history database). Diorama uses hierarchies of tags as categories, with paths, in "tagonomies", or, tag-based taxonomies. The taxonomies are only suggested, and there are so many aliases, mispellings, case-changes, that they cannot be said to be purely concrete taxonomies, since they don't root together in a particularly organized or distinct way, and in many ways are really graphs. But Diorama soldiers on through the aliasing process, until small set of commands and small set of paths are what I work with most of the time, and those paths and commands become familiar to me and to Diorama, so they become quicker to execute in the command shell.

Since I work mostly through Diorama, everything I do is stored in my personal command history database. It is indexed by time. Documents and remote processes can link to the exact time I executed something, and for API calls inbound, any Diorama context I was logged into can always return its local timestamp, so that the history data are kept local, but can be indexed by request in the future.

Any time I am using Diorama, I get to keep history. So this works regardless of where I log in from. The history data are kept in my Diorama distributed database.

The Diorama command line is simply a command line environment that has a display area that can show menus of commands, pictures, web pages, etc. Diorama has a context area that shows information about the context you are in. Menus, commands, resources, shortcuts, notes, etc. You execute commands to open documents, take actions like send mail, create content, publish, and subscribe.

The context area shows you the tagonomy of where you are, and also the tagonomies of any document you have open, etc. So for this page I might see
"docs/dioramaBlog" or "projects/diorama/blogs/InThroughTheDioram". You can tag anything with these path tags (the character I use is a / but you could use anything, such as "projects :: blogs :: In through the diorama"). The system treats things in the path as parent contexts, and learns about how you categorize things this way. It starts suggesting context paths, for example. The system also does quite a bit of aliasing, phonic checking, and case-insensitivity.

Diorama also has a context area that shows you where you are in the web of contexts. Most of the contexts you will spend time in are yours, like your home directory, or directories of projects you are working on. But you stay in these contexts in multiple windows, and can switch contexts with the keyboard, while still in command mode on the command line. It's very much like being in emacs, sending mail. But it is very menu driven and simple, obvious, and fast. But you will be able to visit contexts of others. You will be invited to join these contexts, because a person on the other end knows you somehow. The system is based, just like human interaction in the offline world, on trust. Most people don't trust pimps, hookers, or car salesmen, but in fact these people are at least honest about their game. Still, when you go to hang out in the park with your best girl, you don't want pimps there, and when you go have a drink with your friends, you don't want a used car salesman standing over your table like a drunken mariachi. So Diorama has no ads. And everyone you run into, or whom you allow access to, is known to you.

No comments:

Post a Comment