Re: [hybi] With deflate-stream / websocket traceroute action

[Andy Green <[e-mail]>]

On 02/25/2011 01:34 PM, Somebody in the thread at some point said:

> However, requiring this would be complicate some simple intermediaries
> dramatically -- I don't know if the value gained is worth anything.

Well if it made trouble it could just be ignored and your intermediary 
is invisible to it without breaking anything.  But I guess most 
intermediaries which could consider it are going to be parsing the 
websocket stuff enough to understand the opcode anyway, it doesn't seem 
like any radical complication to do it from that basis.

>   Also, there would seem to be a problem exposing this to JS -- you
> wouldn't want hostile JS to be able to get a layout of the network
> behind the firewall.

That's why it was a MAY, just like some places filter ICMP ECHO in and out.

Some ninja intermediaries might not want to participate in a headcount 
either for their own reasons.

> What does this gain over using ICMP or UDP-based traceroute?

Well, traceroute is telling you about things that at least perform a 
routing action whereas this websocket-traceroute is telling you about 
websocket intermediaries on your path that might be meddling with your 
websocket stuff AS websocket stuff.

It's not critical but since it's only a small addition to existing 
opcodes, I thought I would describe it since adding opcodes to do it was 
mentioned.

-Andy