I listened to a two part Dr. Dobbs TechNetCast presentation from John Ousterhout on "building a business around open source." This talk was recorded in 1999 back when Scriptics was called Scriptics and still a going concern.
In that talk, Ousterhout makes the case that Tcl entered into a wide variety of systems and projects through the back door. Engineers tend[ed] to know about Tcl, and some of their managers, and maybe departmental directors. But knowledge about Tcl hits a glass ceiling about one or two levels below the CxO level officers of a company. [*]
Then Ousterhout makes the case that the next wave of innovation in technology will be about integration -- scripting or gluing disparate systems together. The premise that [d]com / .net / ejb / corba / http / rest / etc. will be the one true umbrella that will cover all software development is fundementally flawed. Over time, each of these integration technologies doesn't reduce complexity, it increases heterogeneity. The old stuff never goes away, it just finds a cozy spot alongside whatever this year's "one true technology" is. So the real problem isn't about converting your old com services to use ejb, but how to get your com and ejb development efforts talking together.
From there, he makes the assertion that Tcl is the ideal platform for doing this work. Perhaps it was. But looking back with 5 years' hindsight, it's obvious that integration was important, even though the industry rejected Tcl, or at least ignored it utterly. That kind of integration happened through web services instead, which were simpler to use since they were not programming languages, but generic interfaces into any programming language you use today or tomorrow. (At least, that's the theory the CxO's buy, even if it's not the reality we see in the trenches.)
Funny how a five year old talk is just as relevant today as it would have been ten years ago. Who says this industry changes quickly? ;-)
*: Perhaps this is why Scriptics/Adjuba/Whatever was destined to die a slow, quiet death...
