We want to have inetutils to work on as many platforms as possible with as few external dependencies as possible. While libnl could provide much ease for us, it is still not a standard library. If route can be made to work in its most basic function without a requirement for libnl (for example by doing the work of libnl it self, and if libnl is avaiable we use libnl), and depend on libnl for extra features, then we can consider depending on libnl.
If we can simply include libnl in inetutils, then that too is a possible solution. But asking users to install libnl for the sake of getting such a basic, and important tool as route to work is not something we can consider. While it is true that somethings become easier when using a library, some don't. We cannot possibly know if the author intends to support legacy GNU/Liniux systems. Nor do we know if the API will be stable, I did a quick check and it seems that libnl has only existed since late 2007 (this was when the git tree was created, maybe it dates back to before that?). If the library is that new, we definitly cannot depend on it for core functionality. Currently there are to many open questions to be able to consider libnl; including no patch that uses it for inetutils. For now it is better if we do not depend on it.
