Instead of one table for each of the three data types you described, build one “denormalized” table as follows:

DATA
id
type (of text, integer, float, etc.)
text
integer
float

The “type” describes what data type is expected. But maybe a field for each kind of data is not necessary? Maybe you could use a BLOB field (and appropriate validation/truncation for numeric fields) instead?

DATA
id
data (of type BLOB)

Additional points:

* Does your revision tracking system incorporate soft deletes in any way? In my experience, this is the most useful kind of revision tracking when it comes to working with real users.

* Maybe you could include a little bit of documentation in your github project to show the exact db structure that results from setting up the project? I’m having trouble picking that out just reading your python sources.

Trying to think of something I miss about the experience … I like constantly refining something, whether it be the company website content or code, but I already get that joy from my other ventures. I did like being able to devote a whole day to whatever problems I felt like solving, but that’s an unrealistic thing to hope for once customers come into the picture with needs other than what I’ve planned for. That’s all I’ve got at the moment. Kinda sad, actually.

Background:
I am working on a replacing the entire data path in Linux, from filesystem to raw storage. The idea is to create a logical system that runs orthogonal (fancy for saying it can be accessed by any layer) to the current data path. There are two main components (1) logical relations and (2) storage and layout. Currently, the logical relation layer is built and working on fudging the storage and layout (save/load state). The system is similar to a database. Only two abstractions are needed to build and understand the system: types, nodes, and edges.

How it differs from a database:
Control over how nodes are stored. This translates to being able to control how your data is stored. The hope here is that you match the storage pattern with your access pattern.
The storage layer has intimate knowledge of the relationships created at the logical layer.

Example:
Hopefully this picture will clear up a lot (http://i46.tinypic.com/2u9h7rl.jpg). Imagine building a dependency graph for the CSS. You can add arbitrary tags to other tags.

Queries such as “Give me all the font css classes” would be very easy to run on such a graph. Efficiency of such queries can be increased because the user provides the code to traverse the graph (think Berkeley DB but easier).

User Interface:
It seems that what you are talking about would be best integrated into something like Eclipse or Netbeans. I have no used these two products very much but I am sure they have a something like a drop down list for object oriented languages to suggest functions belonging to the object you are operating on. This would be similar.

Maybe I can use this to showcase the usefulness of my research The hard part of quantifying the usefulness, time to develop/change a webpage maybe? Also, the time to build the feature itself using Nodes/Edges.

Sorry about the confusion regarding my Remote Help topic. I’ve clarified now.

About decentralized social services … one wouldn’t need to spread usernames or passwords or emails. The system would involve an RSS/XML-like protocol for notifications. My server would notify your server if I “liked” a piece of content you wrote on X website. Similarly, if I decided to follow all of your content on X website, my server would tell yours, which would then ping mine when you publish new content.

It would definitely be open, and the beauty of it is, there could be one open social dashboard for all of this. A person doesn’t need their own web server for this to work. If I wanted to provide a proof-of-concept, I could provide this notification system as a service to anyone that wanted to sign up for it, using my server as their central hub. Of course, blogs and existing content networks and websites would need to add support for this API… but hey, pingbacks and trackbacks are widely adopted now. This isn’t much different.

Think I need to create a flowchart to make things more clear.

You’re right, the help topic is misleading … I’ve now reworked some of it to make it clearer what I’m getting at.

And man, linking straight into videos would be awesome! I bet most users would rather be shown something through video, rather than having to decipher a vague help doc.

Another excellent suggestion about embedding the video into the app itself so the user can follow-along. That would really be a killer feature if the help system didn’t lose your place as you worked with the application.

Your feedback is making me to want to develop a working prototype … crap, now I’m going to have to find some more time, pitch to a VC, get some funding, find another programmer and a designer. Ahhhh!

About the CSS editing tool … maybe I need to explain this more. Or maybe I should bang out a prototype so I can show people what I’m talking about.

It differs from Firefox’s Web Developer tools because it would be smarter about CSS. I think it could go above and beyond “inspect element and show its CSS for editing” by presenting you all of the classes/ids that specify font or color properties. You’d be able to tell it to “show me only the classes/ids and their rules that pertain to FONTS” or “only those that pertain to COLOR”, depending upon what you need to tweak.

It could also identify classes and ids that set the same color/font property value so you could edit them all at once (you’d only have to type in a new value ONCE). It would also keep FULL REVISION HISTORY of every edit you’ve made, because sometimes you have to backtrack to a previous color combination that worked better than what you have now.

I’ve even fancied having part of the tool assist you in simplifying your CSS by removing redundancies, making more global use of inheritance, etc. But that’d just be an intellectual challenge for me. Doubt any designers would care about having condensed CSS.

Ok, that’s enough for now.

Placement:
Placement and presentation of the video is trick too. You want the user to be able to follow along with the video. This rules out pop-ups and full screen overlays.

In-browser, CSS Editing Tool:
Firefox add-on
inspect element (jumps straight to the element in the HTML code) and you might be able to edit the CSS.

vnc’s viewer can be run in listening mode where it’s client (the viewed remote computer) actually initiates the session. there’s also a java/web based interface with all of it. i’m fairly certain the java and c is open source. that would be a great platform to start with. maybe an ajax/java collaboration.

as far as the decentralized social services and notifications, i’d have to say the k.i.s.s. method would = e-mail. the only other way to have something decentralized will involved spreading usernames, passwords, e-mails, etc. across multiple servers or using a centralized authentication database (e.g. openid) to store stuff and letting remote servers access their information to notify them.

“The intent is not to remove the need to visit facebook, but rather to serve as a funnel for some common online social interactions.”

that tells me there’s a dashboard to be built that is open and will pull from all associated sites with modules or plugs or something from a specific site (e.g. facebook, flickr, etc.)

i think that would be much better. an open social dashboard with open extensions and such.

Using the survey questions to find out more about my audience is great advice. Because really, I don’t know much about how drumlines acquire cadences these days. Not knowing who I should be targeting, plus the fact that I feel out of touch with the current state of marching percussion (and the high school drummers themselves), makes things even more difficult for me.

I’m actually using a few open source tools, my own marching percussion notation (which gets converted into formats understood by those open source tools), and a soundfont of marching percussion instruments I found online. I have considered buying professionally recorded samples/soundfonts (and I actually did, but it’s for non-commercial use … more about that mixup here).

Last night I wrote in-depth about how my toolset came about. That post is here.

On November 8 I started using another tool to create wav/mp3 files from MIDI+soundfont, and it resulted in a much better sound. I imagine you heard the updated versions… the previous demos were even worse! So cold and empty. In the new ones you can tell the samples were made in a gymnasium (at least that’s what it sounds like to me).

http://listen.grooveshark.com/#/song/Refined_Fuji_Garbage/21918785