The ability to continue a connection after switching networks is really cool, but everything else seems like just a reimplementation of TCP. Whatever benefits you might get by combining encryption with transport will get cancelled out by adding the overhead of UDP+QUIC. If applications have to integrate stream prioritization themselves, they may as well implement the actual prioritization themselves from a compatibility standpoint.
Am I missing something here? I’m baffled that the IETF would consider using it, let alone making it the basis for HTTP/3.
No comments:
Post a Comment