[nanomsg] ws transport
- From: Martin Sustrik <sustrik@xxxxxxxxxx>
- To: nanomsg@xxxxxxxxxxxxx
- Date: Wed, 26 Nov 2014 13:06:25 +0100
Hi Jack,
I've spent few last days playing with the WebSocket transport. Here are
some observations:
1. As already discussed there's a lot of stuff that's not really needed.
2. I've made an equivalent of tcp tests for ws and they fail.
3. When sending a message from client, the message body is masked. Given
that message may be shared, this garbles the message daa for everyone
else.
I've tried to make a minimal implementation of ws transport (see ws2
branch in the repo). It's still missing some pieces but good news is
that all the tests (see ws.c and ws_shutdown.c) pass.
Here's the summary of how it works:
1. The connection string has the same format as tcp transport
(host+port).
2. The "resource" field in the request is always set to "/"
3. Protocol field is always set to "sp"
4. After initial WebSocket handshake, each peer sends an 8-byte SP
header as a new WebSocket message. Its format is the same as in tcp
transport (see the RFC).
Missing pieces:
1. Timeout for the intial handshake.
2. Closing handshake
3. Rigorous validation of initial request and reply
4. Detailed error handling
5. Ping/pong
Thoughts?
Martin
Other related posts:
