Dashboard, a compelling articulation for realtime contextual information

Miguel de Icaza and Nat Friedman were keynoting at OSCON this morning. It was a great talk about the Mono project and a cool demo of Dashboard. I managed to convey some of the presentation to the #dashboard folks who couldn’t be present. It was also really refreshing to see source code, system exceptions and actual open source on the big keynote display.

I discovered dashboard this week thanks to Edd, who has been doing some neato hacking with some dashboard front and backends already. Dashboard shows itself as a little GUI window on which information sensitive to what you’re currently doing (receiving an IM message, sending an email, looking at a webpage, for example) is shown.

The heart of dashboard is a matching and sorting engine that receives information (in the form of “cluepackets” – how evocative is that?) from frontend applications (like your IM and email clients) and asks the plugged-in backends to find stuff relevant to that information, which is then displayed in the sidebar-style window, designed to be glanced at rather than pored over. It’s a lovely open architecture in that you can (build and) plug in whatever frontend or backend lumps of code you think of.

I’ve been musing about an SAP backend – wouldn’t it be interesting if the engine could get a match from R/3 on a purchase order number, for example? Of course, there’s nothing out of the box on the R/3 side that could be used, but as our talk at OSCON (hopefully) showed, there are plenty of opportunities for the wily hacker.

And what about Jabber? While glueing Jabber stuff onto the front end is one thing, building a pubsub style Jabber backend could get really interesting; coordinated matching, CRM style features … Ooo, the world definitely could get very lobster-like.

And I know it annoys Nat, but I just had to point out that the GraphViz output for matching clues looks very arc-and-nodey … and we all know what thatleads to :-)