Re: [Spud] Fwd: New Version Notification for draft-herbert-transports-over-udp-00.txt

[Joe Touch <[e-mail]>]

On 5/19/2016 4:23 PM, Tom Herbert wrote:
> ...
>> And there is nothing in the Internet architecture that constrains any of
>> this.
>>
>> Your issue is with the OS, not the Internet.
>>
> But there is the issue issue with Internet. Middleboxes will drop
> packets they don't like for arbitrary reasons. RFC2460 clearly states
> that intermediate devices should not inspect EH and yet they do and
> often drop these packets. As I pointed out we've seen middleboxes that
> drop TCP packets that contain options unknown to it, ones that drop
> SYN packets that contain data (problem with deploying TFO),
> middleboxes that take it upon themselves to rewrite parts of the TCP
> header as they see fit. Heck, before we were using TLS we had
> middleboxes parsing TCP payloads and doing strange things to it like
> inserting their own ads into http (see accord BOF).

Now your issue is with the ISOC for not enforcing compliance, or at
least requiring validation before using terms like "supports Internet
protocols and standards".

However, like problems with OS implementation, these issues cannot be
corrected with technology. All you will accomplish is to "move the
goalposts" and escalate the problems you're already experiencing.

Joe