Hi,
On Sat, Apr 15, 2006 at 07:54:12PM -0400, Dimi Paun wrote:
On Sat, 2006-04-15 at 16:17 +0100, Karl Lattimer wrote:
I seriously doubt that as far as users are concerned that dependencies would be an issue, the user in general just wants something that works, and they don't care that 4 extra dependancies are required (python, gtk, pygtk, pygtk-glade)
For what is worth, I personally think you've made the best choice of tools for this project. The problem with the "least common denominator" approaches is that you get a sub-par result. Don't let the rhetoric slow you down.
I'm also voting for not going with a C-only version here. It's important to stand on the shoulder of giants and use what's there already. Using C code for a frontend with complex interactions with various subsystems where you'd have to re-code several parts on your own again just doesn't sound right. One should go with a *useful* GUI toolkit and decide this based on its distribution prevalence and its features. One could even see this as a positive vote *for* a certain toolkit, and given some time and a wise choice it will turn out that more and more applications chose the same toolkit thus making it a pretty permanent feature of any Linux distribution. That's just the natural way of OSS development evolution: stuff that's good will be re-used over and over again, and stuff that's not good will die very quickly (incidentally thus punishing those projects that didn't choose a GUI toolkit wisely), and stuff that's worse than a newcomer will get deprecated eventually once the newcomer is fully established.
All this means that it's most likely a bad idea to go back to the 80s and code a complex GUI app which has to interface to all sorts of standard Linux desktop services(!) in C again, unless there's a complete and rich interface to all required services available, which I doubt.
Andreas Mohr