On Wed, 29 Dec 2004, MenTaLguY wrote:
My feelings on Glade are pretty much summed up in one word: don't.
I think it's pretty telling that nearly every project I've ever seen that has started out using Glade has eventually abandoned it. I was around for the de-Glading of Sodipodi, and I really don't want to reintroduce a libglade dependency in Inkscape.
Yup, I remember that.
Not because it's hard to use or anything like that, but because it makes makes reorganization and refactoring of code that incorporates it very difficult (except in the one sense that you don't need to care about where in the widget hierarchy a widget is to find it, but if you're writing your code properly that shouldn't be an issue anyway).
Just my few cents... admittedly much of the GUI code in Inkscape we inherited is an example of how _not_ to do things manually, but we can do better with the gtkmm rewrite.
Ironically, I think a large measure of the motivation to move to Gtkmm is due to the problems we've had with the historically glade-generated code.
If you think about it, the areas where glade would pay off - being able to generate masses of highly repetitive code - are exactly the sorts of things that Gtkmm enables us to eliminate. Instead of having to generate a lot of repetitive code, we can abstract it into a base class.
However, I wanted to give Glade another try, since it's been a few years, and because Murray wrote some encouraging words on our Gtkmmification page about using it. However after spending an evening struggling with it, my previous opinions are reaffirmed. In fact, now I am even skeptical of glade's usefulness for doing mockups; it's kind of a pain to use.
Bryce