I was the VP of Engineering of Ousterhout's spinout company, Scriptics, and worked for him. I don't know the story of how Sun made the decision but John has a longer History of Tcl here: https://web.stanford.edu/~ouster/cgi-bin/tclHistory.php
"At the same time, it became clear that Sun needed to focus its language evangelism around Java, which was released shortly after I arrived at Sun and had become hugely popular. Though Java and Tcl are very different languages, used for very different purposes, it would have been too hard for Sun to evangelize both of them simultaneously. I came to the conclusion that, overall, Java offered more benefits for Sun than Tcl did."
John was a great person to work for. He hired me despite the fact that in the interview with him I told him I'd never used Tcl and when he extolled its virtues he told me that it was "easy to learn and had powerful regular expressions" and I said "Don't those two things contradict each other?" He stopped talking and had a rather surprised expression on his face. I thought I'd blown it. Actually, I'd done the most important thing I could possibly have done: made him think something new.
I got exposed to TCL/TK while working for a National Lab a long time ago. At the time I was a hardcore C++ developer and after a week or two with TCL/Tk I was almost dying laughing about how easy it was to build really cool applications. I truly miss working with that platform.
Argonne National leaned heavily on Tcl relatively recently for a petascale super-computer project of theirs[0] (I was introduced to it by Justin Wozniak of Argonne at a Tcl conference in Manassas).
In his google talk on the philosophy of software design, he said that the hires who worked out best for him were the ones where he just really enjoyed the conversations with them during the interviews [1]
There's definitely a reality split between developers who accept early in their careers that they need to learn SQL and regular expressions and developers who cling to the hope that if they hold on just a few more years they won't be important skills any more.
"At the same time, it became clear that Sun needed to focus its language evangelism around Java, which was released shortly after I arrived at Sun and had become hugely popular. Though Java and Tcl are very different languages, used for very different purposes, it would have been too hard for Sun to evangelize both of them simultaneously. I came to the conclusion that, overall, Java offered more benefits for Sun than Tcl did."
John was a great person to work for. He hired me despite the fact that in the interview with him I told him I'd never used Tcl and when he extolled its virtues he told me that it was "easy to learn and had powerful regular expressions" and I said "Don't those two things contradict each other?" He stopped talking and had a rather surprised expression on his face. I thought I'd blown it. Actually, I'd done the most important thing I could possibly have done: made him think something new.