Sherlock, Ric wrote: >> mar, 20 Apr 2010, Sherlock Ric skribis: >>> I haven't looked at the Gtk API for J (JGtk ??) very closely all but >> would be interested in a description that relates it to the environment >> described in this article about how Gtk language bindings for Python >> (PyGtk) are being improved. >>> http://arstechnica.com/open-source/reviews/2010/04/python-support-in- >> gnome-gets-a-boost-from-hackfest.ars >> I don't know exactly how pygtk is implemented, but superficially it >> seems >> to be a library written in C linked with python so that a high-level >> interface was provided. This is very different from current J. Anyway >> I think we should worry functionality more than performance in its >> initial stage. > > I'm very much out of my depth here, but are you suggesting that the J > bindings for Gtk are written using J? If so I don't have any problem with > that - I agree that the key is functionality (especially ease of use), but > isn't that determined by the high-level interface provided? Are there things > we can learn/copy from the Gtk bindings for other programming languages to > save us re-inventing the wheel? Or would that inhibit a more J-like > interface? The article suggests that maintaining bindings to Gtk is > non-trivial, but is made easier by using "GObject introspection". I suspect > that this would only be useful where the bindings are written in C? > > I imagine that those implementing the J interface for Gtk have worked through > these sorts of issues. I suppose what I'm really asking is for some insight > into those thought processes and some idea of the "grand design".
I don't remember the exact history of this, but Bill and I worked on gtk bindings for plot and some simple demos, see http://www.jsoftware.com/trac/addons/browser/trunk/gui/gtk/source/gtk/api.ijs. Since then, a lot more bindings have been added by a combination of quick hack and manual effort. We don't have a proper generator, and I have found that adding in new bindings manually works pretty well. In most cases, what is needed is good J cover functions, and so far, we have built a reasonable set for the J session. BTW, the IDE is at the point where it can be used to bootstrap itself. I plan to clean it up and add some docs this weekend and get some preliminary feedback. We might get the first beta of it out in maybe 2-3 weeks. Chris ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
