For user interface development, Tk is flexible and powerful. We have found that [incr Tcl] and [incr Tk] greatly improve the structure and clarity of medium-sized user-interface programs (Tycho is currently 64,000 lines of [incr Tcl]). Compared to the legacy Tcl/Tk code in Ptolemy, the Tycho code is marvelously well-organized. We would strongly recommend [incr Tcl] and [incr Tk] for Tk user-interface development for any substantial program.
[incr Tcl] and [incr Tk] are not without shortcomings. Many are largely inherited from Tcl/Tk: poor performance (which the new byte-compiler will greatly alleviate), and extremely lax parsing. In programs of this scale, run-time errors caused by an inadvertent extra parenthesis (for example), are extremely annoying and do nothing to give us confidence in the stability and robustness of our code. We have had problems with Tk's focus mechanism and the interaction of [incr Tcl] objects with Tk's update mechanism. As much as possible, we have worked around these, and other developers may wish to use the work-arounds incorporated into Tycho.
On the whole though, our experience in this project has been very positive and we are excited about future prospects. We urge Sun's Tcl/Tk group to focus on integrating Tcl/Tk (and [incr Tcl]/[incr Tk]) and Java into a seamless scripting/user-interface/code-development environment. This we see as the logical future of platform-independent computing.
Tycho is freely available under the unrestrictive UC Berkeley license
- see the Tycho home page:
https://ptolemy.eecs.berkeley.edu/tycho.