this might be the craziest GREP yet: How about we kill SWIG as a dependency? Note: We don't want to remove functionality, and we love our Python bindings. But, could we do better? SWIG slows down our compilers, creates hundreds of kB of source code, and we don't really know what it does.
On the other hand, that's because it solves a very complex problem: We point it at headers, and it auto-creates Python libraries for us.
And of course, changing this would most likely cause some code to become no longer compatible with GNU Radio.