A New Approach to Application Integration
I'm sure we've all heard that line before. Redmond
magazine Editor Ed Scannell and I got a briefing from a company called OpenScan (www.openspan.com
) last week that touted just such a claim. The announcement accompanying the briefing was under embargo until today (yes, sometimes we still get pre-briefings), so I decided to wait until today to even post a description of what the company is doing. When I worked at Compuware's NuMega Lab, one of our most successful products (BoundsChecker) injected debugging code into the memory space of a running process (and yes, that is also what a virus does). This code identified traced code execution and determined the values contained in variables, among other things. The important thing was that it could see many things that weren't being exposed by the application.
OpenSpan also injects code into the memory space of a running process, but to identify objects (a term, I believe, used loosely rather than strictly) and interfaces to those objects. OpenSpan CEO Francis Carden referred to those interfaces as APIs, but that is rather a misnomer. Because processes have different characteristics across platforms, the limitation here is that the process had to be running on a Windows box. You explicitly don't need source code.
Now here is the amazing part. Using an IDE called OpenSpan Studio, you could wire together those objects, without programming, so that they exchanged data. Carden demonstrated this by hooking the ubiquitous Windows calculator to an IE session displaying Google. By typing a number in the calculator, he sent that number as a search request to Google, which returned the results in the browser. Were it an appropriate UI for search results, he could have returned those results to the calculator display.
Way cool. I'll be writing more on this in the future for Redmond magazine (redmondmag.com) and Redmond Developer News (reddevnews.com).
Posted by Peter Varhol on 04/23/2007 at 1:15 PM