3 ms·
This whole patent is annoying because it reflects something that has been bothering me about the web for a long time. We built this enormous architecture that i
by jerf 17y ago
This whole patent is annoying because it reflects something that has been bothering me about the web for a long time. We built this enormous architecture that is built on TCP/IP, then carefully, oh-so-carefully made damn sure that it was much, much less functional than the base it was built on. Only recently are we sort of getting back to the base capabilities of TCP/IP with web sockets, and by the very phrase "web sockets" you can guess that they're a bit mucked up. Which they are.
It should hardly be patent-worthy when somebody figures out a way to undo the damage done to TCP/IP by a higher layer. Obviously TCP/IP is useful, why would it be remotely surprising that bi-directional communication is a good idea? It's not like the implementations are that surprising; channels of data are channels of data. Anyone who doesn't find that an obvious statement is not "skilled in the art".
I suppose I should rush out to file a patent on layering SSL back onto web sockets. Or proxying web sockets back to real sockets. If somebody hasn't already.
- realstevejobs 17y agoI share your outrage. Fortunately, WebSockets are specified to be able to connect over TLS, and proxying WebSockets to TCP was one of the first use cases for WebSockets. You could patent using something like SlowAES (http://code.google.com/p/slowaes/ http://code.google.com/p/slowaes/) to send encrypted data over a plaintext WebSocket, but obviously that should not be a patentable "invention." On a side note, I question why you call WebSockets "mucked up." The connection handshake differs from TCP to fit the web security model. Overall it is very close to the wire.
- jerf 17y agoJust as you say at the end there. It makes me grumpy that instead of being able to just pop open a socket to an XMPP server I have to jump through hoops. I give a specific example, but the principle holds generally. (What will happen of course is that XMPP servers will adapt to be able to run over web sockets. Speaking for the server I've spent the most time with as a developer, ejabberd, it's nearly trivial.) Security and domain stuff I'm fine with (don't need random web pages open random connections to SMTP servers and sending out spam!), but we still end up shut off from a lot of useful existing stuff for what I see as dubious security benefits since it'll all be proxied and/or re-implemented to work with websockets anyhow.