Working on the implementation of the bindings often results in overlooking how the use of the various commands and options is accomplished, ie. coding problems are not identical to those of scripting. Some aspects of the gnocl package are not the outcome of the binding code but are inherited from the Gtk library itself, whether these be strengths or weaknesses. Looking to produce a module building system for Gnocl is a good return to scripting for me as it not only show just how good Gnocl is (even though I say it myself) but also where some of the underlying issuee remain. These issues can be both code and methodologies for getting complex scripts to work. Creating an installation script with gnocl::assistant is good because is a contemporary mixture of UI graphics and 'old-fashioned' menu screens.
Working on the project has given me the chance too to explore the use of a number of pure Tcl/Gnocl megawidgets that I created some time ago. They were built, supplied with a test script and left on the shelf. Resurrecting this package for this job has been useful, not simply because it has provided me with some nice off-the shelf widgets for forms creation, but also because it has highlighted the need to tweek the gnocl core sourcode to refine the gnocl::fileChooseButton command and also the need to create a purely Tcl/Gnocl gnocl::labelPathEntry megawidget. Here's a screenshot:
Also, yesterday I uploaded the latest snapshot of Version 0.9.94g to sourceforge and announced its release on Freshmeat.