By Delabassee-Oracle on Feb 19, 2015
...well almost! There are still some minor processes to go through before they are published as official RFCs but the IESG has formally approved the HTTP/2 and HPACK specifications! So clearly, it’s safe to say that HTTP/2 is done! And given how vital the HTTP protocol is in today’s world, this is really important news.
The fact that HTTP/2 is binary based (Vs. HTTP 1.x being text based) removes lot of technical barriers that (text-based) HTTP 1.x had and allows the introduction of new capabilities. Those new capabilities (e.g. stream multiplexing over a single TCP connection, stream prioritization, server-push, etc.) are more than welcome as they will reduce the overall Web latency. HTTP/2 will also help to get rid of the various “hacks” (e.g. file concatenation, domain sharding) that were put in place to work-around the HTTP 1.x limitations (e.g. Head-of-Line blocking).
HTTP/2 maintains a high-level compatibility with HTTP 1.x and preserve a lot of its predecessors concepts (e.g. URI, headers, etc.). So from a Java EE developer point of view, the HTTP/2 impact will be relatively minimal and only a few HTTP/2 aspects will be exposed to the developer through the Servlet API (Server-Push & Stream Prioritization). This work is conducted right now in the Servlet 4 (JSR 369) Experts Group (see EG discussions here).
And even if you are not a Web-tier developer, it is important to understand HTTP/2 and what it brings. For that, you are strongly encouraged to watch the session that Ed Burns and Shing wai Chan (Servlet 4 Specification Leads) gave during last JavaOne on HTTP/2 and the Servlet 4 plans (slides here).