Re: [hybi] Criteria for evaluating handshake proposals

[Zhong Yu <[e-mail]>]

On Wed, Dec 8, 2010 at 3:55 PM, Zhong Yu <[e-mail]> wrote:
> It's ironic that nobody emphasizes on success rate now. At least I
> haven't heard any recently. Yet that was the reason that current
> WebSocket proposals are contained in HTTP connections on port 80.
>
> If we are happy with the 70-80% success rate with a "raw" WebSocket
> protocol over today's internet infrastructure, and if we are in a
> hopeless gridlock if we try to improve the success rate by dressing
> WebSocket in HTTP's clothes, maybe we should just abandon HTTP and go
> with a "raw" WebSocket on a new port.
>
> At least, we deserve a new trial after such a long time. The original
> proponents of building WebSocket in HTTP should examine the new
> evidences and discussions and remake the case.
>
> The success rate is only an initial problem anyway. After WebSocket is
> widely deployed, the obstacles that prevent successful WebSocket
> connections will be under great pressure to correct themselves. People

And if today's firewalls block WebSocket connections, we should not
count that as failure. The firewalls are doing what they are supposed
to do. Whether they should be modified to allow WebSocket is up to the
firewall admins. We should not make that decision for them. What gives
WebSocket the right to *cheat* these firewalls? If the cheating is
successful, aren't we designing something that *breaks* security?

> are dying to have WebSocket. It's hard to see why this little initial
> problem can have any significant impact on adoption rate. As far as
> success rate is concerned, everyday WebSocket isn't deployed is a day
> it has 0% success rate.
>
> - Zhong Yu
>