Buckybase 2.0 in 2010

A new release of Buckybase, the database that would be king, will hit the web sometime in 2010.

The most important features are:

• Combination of outline, wiki, and blog
• Shared and private spaces, in addition to public ones
• More interactive user interface

See you soon!

--Manuel

Buckybase 2.0 Authentication

Originally, I wanted to use OpenID 2.0 for authentication, but the spec is too heavy for me, and I don't want to incorporate large libraries at this point.

At the urging of Master Earl to DTSSTCPW, I decided to go with a simple username/password scheme for now.

Being a perfectionist, I have chosen the following route:

• All communications are encrypted via SSL. This gets rid of complex approaches to counter cookie theft.
• Session cookies contain a HMAC digest (paper) with a rotating random key on the server.
  
• Passwords on the server are stored salted using the bcrypt hash function, which can be made slower as computers get faster, to make brute-force attacks harder.

I think this approach is reasonably perfect.

DO-IO preview screenshot

Another day of CSS tweaks. :)

Buckybase NG preview screenshot

A screenshot of the next generation Buckybase and its GUI (codenamed DO-IO), which will be released under the GNU Affero GPL in time.

Buckybase: The Next Generation

A plan for the next Buckybase.

Links are searches

A link (e.g. Hello world) leads to a search results page.

If there is a "very exact" title match, the matching page is displayed prominently.

Multi-device synchronization

Central server + 1 mobile phone + 1 laptop (usually).

One device is primary for prolonged periods (phone while on the road, laptop at home).

Before switching primary device: sync with server! (No need for diff/merge.)

Smart clients, dumb server

Cycles to burn on the client.

Server is basically a stupid "network fileserver" (for now).

Before starting work, client needs to download all files into local store (Gears SQLite).

Data model

A file consists of a headers dictionary and a binary body.

Required headers are ID (unique, numeric) and Content-type (MIME). Title is optional and not required to be unique.

Plus

All the good stuff from Buckybase Classic: Linked data, simplicity, social views, data export, ...

And: privacy, workgroup sharing, ...

Per-user search

Buckybase now has per-user search thanks to the wonderful Linked Custom Search Engines.





All it takes is a dynamically generated CSE input file per Buckybase user account.

This file assigns the arbitrary label buckybase_cse to all pages under the glob /manuel/*, and then tells CSE to display only pages with that label (FILTER).

(There are a lot more possibilities, such as boosting or eliminating pages with certain labels, and even assigning synonyms.)

Thanks to Ramanathan V. Guha and the other Googlers behind CSE (yes, that R.V. Guha).

Visualizing Buckybase with Tabulator

Buckybase embeds RDF data in the HTML (previously).

TBL's Tabulator Firefox extension adds another visualization option to Buckybase.

With Tabulator installed, an icon in the address bar indicates that the page contains RDF data:



Clicking on the icon puts Firefox into Super Triple Pursuit Mode:



Now, we can quickly zoom deep into the data:



We can even display pages inline:



Update: For some unknown reason, Tabulator is currently unable to properly extract the triples.

The simplest semantic web UI

Buckybase probably has the simplest user interface to enter linked data.

Now, triples are embedded in the HTML using RDFa.



Point the RDFa distiller at e.g. Bob Dylan's Time out of Mind and it will emit this RDF resource:

<rdf:Description rdf:about="http://buckybase.appspot.com/manuel/time-out-of-mind">
   <bb:rating rdf:resource="http://buckybase.appspot.com/manuel/5"/>
   <bb:date rdf:resource="http://buckybase.appspot.com/manuel/1997"/>
   <bb:artist rdf:resource="http://buckybase.appspot.com/manuel/bob-dylan"/>
   <bb:category rdf:resource="http://buckybase.appspot.com/manuel/album"/>
   <bb:location rdf:resource="http://buckybase.appspot.com/manuel/new-orleans"/>
   <bb:location rdf:resource="http://buckybase.appspot.com/manuel/usa"/>
</rdf:Description>

Update: The RDFa Highlight bookmarklet can visualize the triples:

More Google gadgets fun

With Google Spreadsheets' importHtml formula, we can link Buckybase data to visualization gadgets (described yesterday).

Here are some more examples:

Album ratings gauges

I always wanted to do this :)



Grouping tables

I wouldn't know what to do with this, but Earl shows the way:



Word cloud

It's "hip" and gets you VC moneys.



OK, enough for now, but we'll return to this.

Hat tip to Phil Jones for telling me to investigate Google Visualizations.

Sneaking Buckybase data into Google Spreadsheets for Fun and Gadgets

Google Spreadsheets can visualize data with an armada of gadgets.

And its importHtml formula can import Buckybase data – simply by parsing the Buckybase HTML table:


(Here's the published spreadsheet.)

As described previously, I've put a couple of albums into Buckybase.

Using the "show these fields" button I display only the location and title of each album:



That's the Google Maps gadget's input format: two columns, 1=address, 2=label.

Now I use the page's URL in the formula =importHtml("http://buckybase.appspot.com/manuel/album/category/of?view=table&fieldnames=+location%2C+title", "table", 1) in the first cell of a new spreadsheet:



And it automatically fetches and parses the data:



Then I Insert > Gadget..., choose the Maps gadget and I'm done:


(Here's the published spreadsheet.)

What's more, Google automatically reloads the data continuously (every couple of minutes or so).

Sortable tables

Buckybase can now display tables such as this one:



I have put a couple of albums into Buckybase, and tagged each with "category: Album", e.g. Grateful Dead's Live Europe 1972:



The page "Album" has automatic backlinks to all albums:



Clicking on "category of" shows a listing of these albums:



The "view as table" button displays them as a sortable table:



Note that it automatically picks the most common fields (columns).

* * *
All pages from all users titled "Buckybase" as a table:



* * *
BIG THANKS to the jQuery Plugin Tablesorter which reduced the client-side work to one line of code:

$(document).ready(function() { $("#data").tablesorter(); });

oEmbed support

Buckybase can embed content from oEmbed providers such as Flickr.

Just write the Flickr page URL on a line by itself:

oEmbed Test

tag: Paris, Sunset

http://www.flickr.com/photos/stuckincustoms/194863333/

This is how the page looks in Buckybase:



The caption and author of the image are automatically retrieved.

Notes:

• Currently only Flickr is supported, but it will be easy to add other oEmbed providers.
• Please only post your own images and/or images that are appropriately licensed.

Update: Amazon product images are now supported via the marvellous oohEmbed. Simply paste an Amazon product URL into the page (sample). Thanks!

Building a John Maeda-styled PIM with Buckybase and iGoogle

I am using my Buckybase web database (introduced here yesterday) itself to manage Buckybase development.

For example, here are my issues I have tagged with "priority: high".



I can add these and other issues to my John Maeda-themed iGoogle, via the new feed support:





I love this Unix-style approach to web apps!

Update: In theory. In practice iGoogle doesn't update the feeds. Like ever.

Introducing Buckybase, a social web database with bidirectional hyperlinks

Buckybase is in alpha at http://buckybase.appspot.com/.

It is a social web database with bidirectional hyperlinks.

The basic idea is that you write pages in a simple format, e.g.:

Buckybase

platform: Google App Engine
inspiration: del.icio.us, Wikipedia, Craigslist, WikiWikiWeb, Vanilla
author: Manuel Simoni

Buckybase is a social web database with bidirectional hyperlinks.

Buckybase parses the information in that page and makes it accessible through hyperlinks.

That same page in my Buckybase account is here: http://buckybase.appspot.com/manuel/buckybase



Note that all items have become hyperlinked, and that you can also click on a field name, e.g. inspiration, to see a listing of the field values:



Furthermore, when you look at the page for Google App Engine, you will see that it has an automatically created backlink "platform of" back to Buckybase.

The final major feature is that you can look at all pages with the same title from all users, e.g. here is the page Buckybase from all users:




Get your Buckybase account and feel free to try things out as you like.

Notes

FEEDBACK APPRECIATED. I am very grateful to any and all comments, questions, and problem reports regarding Buckybase. Please contact me at my msimoni Gmail account, via Jabber at msimoni@jabber.org, or in the comments section.

ALPHA WARNING. Buckybase seems stable but I have just finished coding today. Do not put any important data into Buckybase atm. Arbitrary data loss and corruption may occur.

APP ENGINE WARNING. Buckybase is built on the Google App Engine environment. App Engine itself is still in prerelease (and has just had some prolonged downtime) so if you experience hiccups, please don't give up, and if they persist, check back later. Thanks.

Best regards,
-- Manuel Simoni