i mean this thing https://ipfs.io https://github.com/ipfs/ipfs https://en.wikipedia.org/wiki/InterPlan ... ile_System
and despite relatively low CPU performance of modern routers, its kinda temporary issue.
how bout atleast cjdns ? https://en.wikipedia.org/wiki/Cjdns https://github.com/cjdelisle/cjdns
there was several smilar to mesh potato projects flying, but most of them use slightly tweaked hwm, hwmp, batman protocol and interoperability remain major issue of adhoc networks, sadly. cjdns kinda resolve that, despite ipv6 flaws.

A quick look over your links suggests these are end-point tech/protocols.
How do they relate to routers?

A quick look over your links suggests these are end-point tech/protocols.
How do they relate to routers?

directly. its mesh routing tech, blockchain-style, comverging other tech within from web-severs, file-severs to anything else, imaginable(including presently developed commercial applications(but afaik they had auth over ipv6 there, yet. because bogus DRM implementation by vendor))
basically title "A peer-to-peer hypermedia protocol
to make the web faster, safer, and more open" say/sum it all ~ more or less accurate.
if you already used some of adhoc/mesh -based products - just think about it "same, but on steroids", with portions of other products, both commercial and free.
its ~ computation -intensive(because nature of tech), but significantly better scale than ROS routing itself(for example), but CCR wouldn't had problems with.

So, it belongs in servers, not on routers.

So, it belongs in servers, not on routers.

thats can be applied to 3/4 of ROS features, then.
SMB ? belong to servers ! tftp ? belongs to servers !! packing ? belong to servers ! webcache ? belong to servers ! and etc and etc so on and so on.