Hello public,
I’m putting some effort into GtkGlAreaSharp again. The supporting bits aren’t really production-ready, though, so rather than hacking on the widget itself and its associated examples, I’m beefing up the underlying platform.
The first step in the process is making Prebuild a viable meta-build system. Right now it works pretty well for generating NAnt, Visual Studio 2003, Visual Studio 2005 and MonoDevelop build systems. The Autotools target is a bit lacking, however. There is not currently support for recursive projects, ie, solutions with multiple projects. There is also no support for versioned solutions or projects. This makes it difficult to strong-named assemblies, without which, libraries can’t be added to the GAC. Production-quality code is usually added to the GAC, so getting the feature addressed is pretty high priority.
The reason I’m focusing on Prebuild is that the Tao framework, a cross-platform C# binding to the OpenGL API, uses Prebuild to generate its build systems. I use Tao in the examples distributed with GtkGLAreaSharp in order to demonstrate the widget’s usage, how one might allow one’s users to interact with objects rendered in the OpenGL widget, etc.
My work so far on the Prebuild Autotools templates is living in my own subversion repository until I get it to a better state than the stuff at sourceforge. If you’re interested in taking a peek, browse on over to https://svn.colliertech.org/mono/dnpbAutotools/
After we convince Prebuild to generate robust Autotool build systems, I’m going to see how much effort it takes to push Prebuild into the Ubuntu core. After there is an Ubuntu package, it will probably take much less effort to get it packaged up for Debian. One day I will become a DD, but that takes a lot of time, and I hear a rumor that someone has to kick the bucket before they start accepting more :)
There are also a couple of nearly-complete bugs blocking gtkglarea from working smoothly on win32. This seems like a good place to focus effort after Prebuild does what I want. Plus, then, I will move the gigantic project of adding OpenGL support to GTK+ core forward a few steps.
One response to “Prebuild Autotools target”
sounds interesting. Prebuild targets for nant, vs2005 can create strong-named assemblies easily. I’ll admit that the autotools target needs work. I think the xlst approach is very interesting and maybe the results could form the basis of a rewrote of prebuild.