RISC-V: They should have known better
Dmitry Grinberg’s long critique argues that RISC-V’s optional extensions, fragmented profiles, awkward compressed encodings, and missing common addressing/bitfield operations make portable software and efficient implementations harder than the clean-slate ISA deserved. He sees RISC-V succeeding in cheap embedded controllers and accelerator sidecars, but doubts it will displace ARM in high-end general-purpose systems.
The 100+ comment HN debate supplied an important counterweight: designers value the license-free standard and can curate an embedded subset, while others agreed that incompatible extension combinations and weak feature discovery create real portability and binary-compatibility hazards.