Attending the Tcl 2002 conference, I saw a number of presentations that mentioned in passing a problem that I am dealing with in my application:
Automatic (or semiautomatic) generation of Tk GUIs based on metadata descriptions.
The applications of this take several forms.
I can see where a front-end/back-end system of reading and interpreting file contents or user gestures then calling constructors to build the interface might make some sense.
Here are the questions:
Conceivably, the generated interface might be on one system connected to a remote system.
Thoughts? Pointers to wiki pages?
David S. Cargo ([email protected])
Traditionally, this sort of thing has been done via the option database I believe; I think Brent Welch's book describes how to do it. XML seems like a nicer solution to me, though. But I don't think I'd tie "user gestures" in too strongly. On the other hand, it's easy to see how a GUI Builder could store the program's structure as a DOM tree, which could then be loaded and saved as an XML document. Then, if they were ordered appropriately, the app could avoid the cost of including the DOM by parsing the XML document as a sequence of streaming events (a la TclXML) and building the app as it goes. -- WHD
It's true that the GUI builder could and should store the interfaces' current definition somewhere (so you can save and edit it later), but if your GUI builder has a button for creating a widget, some version of the widget needs to be created as soon as the button is pushed. That's the kind of "user gesture" I had in mind.
I see a process of perhaps three parts:
David S. Cargo ([email protected])