I've been wondering lately... Is there any benefit to going the ARX route at this point? .NET has been easy to learn and get moving, but at the same time there are many options that aren't available though the managed API. I find myself misusing elements of the managed API because I haven't been forced to fully understand what's really happening behind it (for example, I read the my first plugin on overrules and could see right away that there was SOOO much that wasn't being explained or understood that I didn't even compile that code for use) because everything crashes nicely in .NET
Can anybody comment?
I use both. About 20% of the AutoCAD code I've written that is in use today, is written in C++ (both native C++/ObjectARX, and C++/CLI for interoperating with managed code)..
Native ObjectARX requires C++ skills, and is more tedious than managed code. Productivity is not not as great. The advantages are performance, and finer granularity in the API.
Perhaps you can cite some examples of the 'many options that aren't available though the managed API', as quite a bit of the native API is exposed to managed code, but in some cases it's not a 1:1 mapping, and the managed equivalent of native APIs may not be obvious.